RECOMMENDED 🏈 NFL ATS spread Predict NFL spreads using Vegas lines, team Elo, and weather. Trained on 12 seasons. 7 blocks RECOMMENDED 🎯 NBA player props Predict NBA player points using rolling stats. Great for over/under picks on PrizePicks. 5 blocks 🔬 Hypothesis test Test one idea quickly with two signals and a simple model. Minimal, fast to train. 4 blocks ⚙️ Autotune-ready Full NFL spread pipeline pre-wired for automated hyperparameter search. Let the optimizer do the work. 8 blocks ▢ Blank canvas Start empty — drag your own bricks into the slots. Advanced — start from scratch
Start from the closest working blueprint Ranked for the current sport and modeling objective, with off-context templates moved out of
the way instead of mixed into the first choices.
Context fit NFL / spread
Template search 49 matched exactly
NFL Recommended Blowouts Beat the Line, Grinds Don't The belief is that big‑score games are under‑priced by the market while tight games are over‑priced. Our Engine spots the game vibe, splits the slate into blowout or grind blocks, and tests the spread against the real result. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Cold‑Weather Home Cover Edge We think cold‑weather home teams beat the spread more often than Vegas gives them credit. Shark Snip’s Engine tracks the line, team power ratings, and cover odds to test the claim. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Cold‑Weather Home Favorite Edge We think cold‑weather home teams beat the spread more than Vegas prices. The engine watches temps, home advantage and spread lines to test that edge. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Cold‑Weather Home Teams Beat the Spread We think cold‑weather home teams cover the spread more often than the line suggests. The Shark Snip Engine tracks Elo, EPA differentials and the Vegas spread to test that belief. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Cold‑Weather Home Teams Cover the First Half We think cold‑weather home teams are more likely to cover the first‑half spread than the Vegas line suggests. The engine watches early‑down success and pass‑rate advantage to spot those edges. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Cover Probability Ensemble Stacking a trained cover-probability feature with ELO power ratings and market efficiency creates an ensemble that beats simple point-spread models; this calibrated blend finds consistent ATS value over a large sample. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Cross-Country Travel Fade Teams traveling across multiple time zones show measurable early-game performance degradation; this model quantifies the timezone-adjusted travel disadvantage to find ATS fade value in West-to-East matchups. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Division Dogs Bite Back Rivalry games tighten up—know thy enemy means closer scores and live dogs. The Engine watches divisional rematch history, line movement, and rest edges to spot where the public's sleeping on the dog. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Divisional dog — Spread A model built on the thesis that divisional underdogs are sharper than the market — familiarity compresses margins toward the dog; trained with bayesian ridge and fixed 1u@1.0.0 staking, sliced to divisional games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Divisional dog — Spread A model built on the thesis that divisional underdogs are sharper than the market — familiarity compresses margins toward the dog; trained with calibrated ensemble and variance capped@1.0.0 staking, sliced to divisional games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Early down — Spread A model built on the thesis that early-down success rate is a stickier cover signal than box-score yards; trained with calibrated ensemble and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Early down — Spread A model built on the thesis that early-down success rate is a stickier cover signal than box-score yards; trained with quantile xgboost and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Early down — Spread A model built on the thesis that early-down success rate is a stickier cover signal than box-score yards; trained with catboost regressor and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Explosive — Spread A model built on the thesis that explosive-play differential predicts cover better than yardage the market anchors on; trained with catboost regressor and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Explosive — Spread A model built on the thesis that explosive-play differential predicts cover better than yardage the market anchors on; trained with elastic net and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Home dog — Spread A model built on the thesis that home underdogs win outright more than the moneyline implies — venue is under-priced for weak home sides; trained with calibrated ensemble and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Injury-Adjusted Spread Model Offensive injury reports create systematic market inefficiencies that persist even after lines move; this model uses offensive injury impact differential and sharp lean to find ATS value when injury adjustments are lagged. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Kicking EPA Cover Model Teams with superior net kicking EPA (field goals + kickoffs) win margin differential is underpriced by sharp lines; this model uses rolling kicking EPA differential and field-position advantage to uncover ATS edges. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Late Money Knows Best The sharpest cash hits closer to kickoff, and the closing line is where the smart money lands. This model tracks how lines move from open to close and hunts spots where the late action points to a cover. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Line move steam — Spread A model built on the thesis that steam and sharp line movement carry signal the closing number only partly absorbs; trained with calibrated ensemble and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Line move steam — Spread A model built on the thesis that steam and sharp line movement carry signal the closing number only partly absorbs; trained with catboost regressor and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Minimal Signal ATS Baseline A parsimonious ATS model using only minimal signals — Elo, EPA differential, and spread line — to establish a baseline beat rate; useful as a benchmark against more complex models and for identifying overfit risk. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Momentum — Spread A model built on the thesis that recent-form momentum is a trap the market over-extrapolates — fade the streaky favorite; trained with calibrated ensemble and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Momentum — Spread A model built on the thesis that recent-form momentum is a trap the market over-extrapolates — fade the streaky favorite; trained with quantile xgboost and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Momentum — Spread A model built on the thesis that recent-form momentum is a trap the market over-extrapolates — fade the streaky favorite; trained with catboost regressor and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended NFL Spread — Bayesian Uncertainty MC-dropout Bayesian net outputs a margin AND its uncertainty — bet bigger when confident, smaller when not. Variance-capped staking. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Pass Rush Edge This theory claims that a team's ability to pressure the quarterback gives them an edge over opponents who struggle to protect theirs. Our model watches how teams with strong pass rushes perform against those with weaker offensive lines to test this idea. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Playoff Calibrated Spread Model Playoff games have different ATS dynamics than regular season — tighter margins, superior opponent quality, and heightened preparation; this model is explicitly calibrated on playoff data using Elo, EPA, and divisional factors. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Podcast Buzz Beats the Line If the buzz on the latest NFL podcasts pushes a team higher, the market line usually lags behind. Our Engine watches podcast sentiment, line data, and prop consensus to spot the mismatch. Matches NFL + Matches spread 4/5 slots Partial registry match NFL Recommended Qb injury — Spread A model built on the thesis that the market over-reacts to QB news on the open and under-reacts to backup quality; trained with bayesian ridge and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Qb injury — Spread A model built on the thesis that the market over-reacts to QB news on the open and under-reacts to backup quality; trained with calibrated ensemble and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Rest & Cold This theory claims that teams playing at home with more rest and in colder conditions have an edge over the spread. Our model tracks how Vegas lines, team power ratings, weather, and rest days impact the spread to test this idea. Matches NFL + Matches spread 4/5 slots Partial registry match NFL Recommended Rest diff — Spread A model built on the thesis that rest and short-week differentials are under-priced — the fatigued side covers less than the market implies; trained with catboost regressor and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Rest diff — Spread A model built on the thesis that rest and short-week differentials are under-priced — the fatigued side covers less than the market implies; trained with elastic net and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Second-Half Momentum Spread Teams with strong second-half momentum — measured by rolling MOV and late-game EPA splits — cover second-half spreads at above-market rates; this model identifies momentum-driven second-half ATS edges. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Sharp Lean CLV Spread Model Sharp bettors consistently generate positive closing line value (CLV) on spreads; this model uses both sharp lean signals to pre-empt line moves and capture ATS edge before markets fully correct. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Short-Week Fatigue Spread Edge Teams on a short week coming off a physical game show degraded ATS performance against rested opponents; this model quantifies the fatigue disadvantage by combining the short-week flag with injury news count and rest differential. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Situational — Spread A model built on the thesis that situational spots (letdown, lookahead, sandwich) are under-modeled and exploitable on the spread; trained with quantile xgboost and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Situational — Spread A model built on the thesis that situational spots (letdown, lookahead, sandwich) are under-modeled and exploitable on the spread; trained with catboost regressor and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Situational — Spread A model built on the thesis that situational spots (letdown, lookahead, sandwich) are under-modeled and exploitable on the spread; trained with bayesian ridge and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Transformer Cold Weather Ats A transformer encoder trained on cold‑weather games with enriched form/EPA features and Kelly‑fraction staking will achieve a higher ATS hit rate than the current GBM baseline. Matches NFL + Matches spread 7/7 slots Full registry match NFL Recommended Travel fade — Spread A model built on the thesis that long-travel and timezone-shifted road teams under-cover — the body-clock edge is real and unpriced; trained with catboost regressor and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Travel fade — Spread A model built on the thesis that long-travel and timezone-shifted road teams under-cover — the body-clock edge is real and unpriced; trained with bayesian ridge and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Travel fade — Spread A model built on the thesis that long-travel and timezone-shifted road teams under-cover — the body-clock edge is real and unpriced; trained with elastic net and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Travel‑Tired Teams Lose the Spread Edge When a team crams a game into a short week and flies across the country, they tend to be slower and give up points. Our Spot watches rest days, travel miles, and the Vegas spread line to spot the mispriced games. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Turf vs Grass Surface Spread Teams built around explosive skill players perform better on turf while run-heavy teams favor grass; this model uses surface type, team YPC, and explosive play rate to find ATS value when surface mismatch is large. Matches NFL + Matches spread 4/4 slots Full registry match NFL Recommended Weather — Spread A model built on the thesis that cold-weather games are systematically over-priced on the total — wind and temperature suppress scoring the market under-adjusts for; trained with catboost regressor and variance capped@1.0.0 staking, sliced to cold weather@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended Weather — Spread A model built on the thesis that cold-weather games are systematically over-priced on the total — wind and temperature suppress scoring the market under-adjusts for; trained with elastic net and quarter kelly@1.0.0 staking, sliced to cold weather@1.0.0, will beat the closing spread on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL + Matches spread 5/5 slots Full registry match NFL Recommended YPC Advantage Cover Model Teams with a sustained net yards-per-carry advantage control game tempo and cover spreads more consistently than EPA-based models imply; this model uses net YPC differential and field position to find hidden ATS value. Matches NFL + Matches spread 4/4 slots Full registry match NBA Compatible Add Referee Features Lab candidate: add-referee-features. Matches spread 2/4 slots Partial registry match NFL Compatible Away Team Total Offense Model High-EPA away offenses visiting warm indoor stadiums score more points than implied away team totals suggest; this model uses rolling offensive EPA, travel context, and implied away total to find team-total over value. Matches NFL 4/4 slots Full registry match NBA Compatible Bayesian Spread Fader Bayesian Ridge regression provides uncertainty-aware spread forecasts, particularly useful when sample sizes are small and recent form conflicts with season-long ratings. Matches spread 4/4 slots Full registry match NFL Compatible Box Stackers Get Buried Defenses selling out to stop the run don't just lower a back's average—they crush the ceiling. This Engine watches box count, offensive line grades, and game script to spot when the median line is pricing a normal day for a back who's about to hit a wall. Matches NFL 4/4 slots Full registry match NFL Compatible Cold Weather Unders This theory claims that NFL games played in cold weather tend to have lower scores than expected. Our model tracks cold-weather games, wind speeds, and temperature to test if the under bet is a smarter choice in these conditions. Matches NFL 5/5 slots Full registry match NFL Compatible Cold-Weather Home Edge We think cold-weather home teams beat the odds more often than the market expects. Our Engine tracks weather, home field advantage, and moneyline odds to see if that edge holds. Matches NFL 5/5 slots Full registry match MLB Compatible Cold‑Weather Pitcher Edge We think cold‑weather starters are more likely to beat the run line than the odds suggest. Our engine spots the edge by tracking pitcher temperature, venue stats, and recent performance. Matches spread 4/4 slots Full registry match NFL Compatible Deep Threats Pay This theory claims that NFL wide receivers who run deeper routes and get better separation from defenders will have more receiving yards than the market thinks. Our model tracks route depth and separation to test if these factors can give us an edge when betting on player props. Matches NFL 4/4 slots Full registry match NFL Compatible Divisional dog — Moneyline A model built on the thesis that divisional underdogs are sharper than the market — familiarity compresses margins toward the dog; trained with quantile xgboost and fixed 1u@1.0.0 staking, sliced to divisional games@1.0.0, will beat the closing moneyline on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Divisional dog — Moneyline A model built on the thesis that divisional underdogs are sharper than the market — familiarity compresses margins toward the dog; trained with elastic net and quarter kelly@1.0.0 staking, sliced to divisional games@1.0.0, will beat the closing moneyline on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Divisional Rivalry ML Upset Divisional rivalry games compress expected margins and create moneyline upsets at above-market rates for home underdogs; this model uses divisional flag, Elo differential, and home field to identify high-value ML plays. Matches NFL 4/4 slots Full registry match NFL Compatible Dome Game Pace Totals Stack Dome games with fast-pace offenses on both sides produce totals that consistently exceed the market line; this model stacks dome indicator, team EPA pace, and rolling offensive efficiency to find totals over value. Matches NFL 5/5 slots Full registry match NFL Compatible Dome Games Go Over the Line When the wind is out of the picture, scores jump higher than the market expects. Our Engine watches dome‑only matchups, tracks the total line, and checks if the over hits more often than the line suggests. Matches NFL 5/5 slots Full registry match NFL Compatible Dome overs — Total A model built on the thesis that dome games favor the over — controlled conditions raise efficiency above the posted total; trained with bayesian ridge and variance capped@1.0.0 staking, sliced to dome games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Dome overs — Total A model built on the thesis that dome games favor the over — controlled conditions raise efficiency above the posted total; trained with elastic net and fixed 1u@1.0.0 staking, sliced to dome games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Dome overs — Total A model built on the thesis that dome games favor the over — controlled conditions raise efficiency above the posted total; trained with calibrated ensemble and quarter kelly@1.0.0 staking, sliced to dome games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Elo Edge This theory claims that teams with an edge in power ratings and performance metrics will outperform the market. Our model tracks Elo power ratings, rolling EPA differentials, and other key stats to test this idea and make informed picks. Matches NFL 4/4 slots Full registry match NBA Compatible Elo Pace Ensemble A stacking ridge ensemble of GBM models using ELO and pace features on divisional games will outperform the incumbent ATS model. Matches spread 5/6 slots Partial registry match MLB Compatible Elo Pace Mlb Ensemble A quantile-xgb model with team form, elo, and pace features, using a calibrated blend ensemble and kelly-fraction staking, will beat the incumbent ats on the same eval window. Matches spread 6/6 slots Full registry match NBA Compatible Elo Pace Weather Ensemble An ensemble model combining elo, pace, and weather features with a quantile xgb architecture and kelly fraction staking will outperform the incumbent ats model on the same eval window. Matches spread 6/6 slots Full registry match MLB Compatible Elo Pace Weather Mlp A tfjs-mlp model using elo, pace, and weather features will outperform the incumbent ATS model on the same evaluation window. Matches spread 4/4 slots Full registry match NBA Compatible Elo Power Spread Elo power ratings provide stable long-run team quality assessments. LightGBM uses Elo gap between teams to project spread outcomes when short-term form diverges from true strength. Matches spread 4/4 slots Full registry match MLB Compatible Elo Weather Ensemble A quantile XGB model with team form, elo, and weather features, trained on dome games and using a calibrated blend ensemble, will outperform the incumbent ATS model on the same evaluation window. Matches spread 7/7 slots Full registry match NBA Compatible Expert Committee Spread A mixture-of-experts model routes NBA spread predictions through specialist sub-models for pace, rest, and defensive matchups, then blends outputs for a consensus spread projection. Matches spread 4/4 slots Full registry match NFL Compatible Explosive — Total A model built on the thesis that explosive-play differential predicts cover better than yardage the market anchors on; trained with bayesian ridge and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NBA Compatible Fast Pace, Fat Spreads High-pace teams create larger score variance, making them unreliable covers as favorites. This XGBoost model exploits pace asymmetry to find mispriced spreads. Matches spread 4/4 slots Full registry match MLB Compatible First Half Run Line Starter Model First-half run lines are driven almost entirely by starting pitcher quality before bullpen effects dilute the edge. This model targets the first 5 innings window using starter xwOBA and recent form exclusively. Matches spread 4/4 slots Full registry match NFL Compatible First-Half Total Pace Edge High-pace offenses and favorable weather stack to push first-half totals over lines that lag behind full-game adjustments; this model uses team pace, game total, and dome/weather indicators to find first-half over value. Matches NFL 4/4 slots Full registry match NBA Compatible Full-Feature Spread Model XGBoost spread model ingests all NBA team features including net rating, pace, rest, home/away, win streak, and SOS for the most complete spread projection available. Matches spread 4/4 slots Full registry match NBA Compatible Gaussian Process Spread Gaussian Process regression provides probabilistic spread predictions with built-in uncertainty estimates, allowing Kelly-staking to scale with prediction confidence rather than fixed units. Matches spread 4/4 slots Full registry match NBA Compatible Graph Network Spread GNN-based matchup model represents NBA teams as nodes with roster-quality edges, capturing complex inter-team dynamics that tabular models miss when predicting spreads. Matches spread 4/4 slots Full registry match NFL Compatible High-Volume QBs Beat the TD Over/Under We think quarterbacks who crank out a lot of passes tend to push more TDs than the bookies expect. Our Engine tracks pass attempts and TD counts to see if the over/under line is consistently off. Matches NFL 4/4 slots Full registry match NFL Compatible High-Wind Under Model Games with pre-kick wind speeds above 15mph show systematic under performance on totals regardless of teams involved; this model uses NWS wind speed, precipitation probability, and total line to bet unders in windy outdoor games. Matches NFL 4/4 slots Full registry match NBA Compatible Histogram Boost Spread Histogram gradient boosting handles large NBA datasets efficiently and natively handles missing injury-report data, making it robust to daily lineup uncertainty in spread prediction. Matches spread 4/4 slots Full registry match NBA Compatible Home Court Premium Home court advantage varies dramatically by arena and roster type. LightGBM quantifies true home lift per team, finding games where the spread under- or over-prices venue effects. Matches spread 4/4 slots Full registry match NFL Compatible Home dog — Moneyline A model built on the thesis that home underdogs win outright more than the moneyline implies — venue is under-priced for weak home sides; trained with elastic net and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing moneyline on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Home dog — Moneyline A model built on the thesis that home underdogs win outright more than the moneyline implies — venue is under-priced for weak home sides; trained with quantile xgboost and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing moneyline on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Home Team Total Offense Model Home teams with high-EPA offenses in dome environments score more points than market team totals imply; this model uses rolling offense EPA and home field signals to bet team-total overs for elite home units. Matches NFL 4/4 slots Full registry match NFL Compatible Line move steam — Moneyline A model built on the thesis that steam and sharp line movement carry signal the closing number only partly absorbs; trained with quantile xgboost and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing moneyline on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Line move steam — Moneyline A model built on the thesis that steam and sharp line movement carry signal the closing number only partly absorbs; trained with bayesian ridge and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing moneyline on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NBA Compatible Lock-Down Defense Covers Elite defensive teams consistently limit opponent scoring below expectations, making them reliable ATS bets even when underdogs. XGBoost models the spread using defensive rating gaps. Matches spread 4/4 slots Full registry match NFL Compatible Long Drives Push Totals Over We think that big, sustained drives in the later half of a game tip the total toward the over. Our Engine watches each play‑by‑play drive, tallies drive length, and checks if the final total beats the Vegas line. Matches NFL 4/4 slots Full registry match NBA Compatible MARS Spread Finder Multivariate Adaptive Regression Splines find non-linear ATS breakpoints in net-rating and pace features that simple linear models miss, uncovering spread edges in close matchups. Matches spread 4/4 slots Full registry match MLB Compatible Mlb Elo Weather Quantile Xgb Quantile XGB with team form, ELO, and weather features on dome games beats the incumbent ATS Matches spread 6/6 slots Full registry match NBA Compatible Momentum Spread Model Teams with sharp recent form outperform season-long ratings in spread coverage. XGBoost weights recent game outcomes to capture momentum swings the market prices slowly. Matches spread 4/4 slots Full registry match NBA Compatible Multi-Task Spread Learner Multi-task learning jointly predicts spreads and totals, sharing representation across tasks to improve both predictions when labeled game data is limited early in the season. Matches spread 4/4 slots Full registry match NBA Compatible Nba Elo Weather Ensemble A mixture-of-experts model trained on NBA primetime games with team form, elo, and weather features, using kelly-fraction staking and calibrated-blend ensemble will achieve a higher ATS hit rate than the current GBM baseline. Matches spread 5/6 slots Partial registry match NBA Compatible Nba Primetime Gbm Ats A gradient boosted model trained on team form, Elo, pace, rest, and travel features for primetime NBA games will achieve a higher ATS hit rate than the current baseline model. Matches spread 4/6 slots Partial registry match NBA Compatible Nba Rest Advantage Gbm Kelly A GBM trained on NBA games with rest-advantage slice (>=1 day extra rest) and Kelly-fraction staking will achieve higher ATS than the flat-staked baseline by exploiting schedule-induced fatigue asymmetries. Matches spread 3/6 slots Partial registry match NBA Compatible Nba Rest Advantage Moe A mixture-of-experts model trained on NBA games with rest-advantage slices and confidence-scaled staking will achieve higher ATS hit rate than the current GBM baseline by exploiting schedule-induced fatigue asymmetries. Matches spread 3/6 slots Partial registry match NBA Compatible Nba Rest Advantage Transformer A transformer encoder using rest-advantage slice and Kelly-fraction staking will achieve higher ATS hit rate than the current GBM baseline. Matches spread 3/6 slots Partial registry match NFL Compatible NFL Team Totals — Pace Vegas implied team totals + pace/EPA into a team-total model. Team totals are softer than the game total. Matches NFL 4/4 slots Full registry match NBA Compatible Offense-Defense Mismatch When an elite offense meets a weak defense, the spread does not always reflect the full scoring gap. XGBoost exploits asymmetric offensive vs. defensive rating mismatches in spread prediction. Matches spread 4/4 slots Full registry match NFL Compatible Pace overs — Total A model built on the thesis that high-pace matchups beat the total — possession volume is under-weighted vs efficiency; trained with quantile xgboost and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Pace overs — Total A model built on the thesis that high-pace matchups beat the total — possession volume is under-weighted vs efficiency; trained with catboost regressor and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Pace overs — Total A model built on the thesis that high-pace matchups beat the total — possession volume is under-weighted vs efficiency; trained with bayesian ridge and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Playoff unders — Total A model built on the thesis that playoff intensity tightens defenses — totals run under the regular-season-anchored number; trained with elastic net and variance capped@1.0.0 staking, sliced to playoff games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Playoff unders — Total A model built on the thesis that playoff intensity tightens defenses — totals run under the regular-season-anchored number; trained with quantile xgboost and quarter kelly@1.0.0 staking, sliced to playoff games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Playoff unders — Total A model built on the thesis that playoff intensity tightens defenses — totals run under the regular-season-anchored number; trained with calibrated ensemble and fixed 1u@1.0.0 staking, sliced to playoff games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Primetime unders — Total A model built on the thesis that primetime totals run high on public over-bias — the under is the disciplined side; trained with elastic net and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Primetime unders — Total A model built on the thesis that primetime totals run high on public over-bias — the under is the disciplined side; trained with quantile xgboost and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Primetime unders — Total A model built on the thesis that primetime totals run high on public over-bias — the under is the disciplined side; trained with catboost regressor and quarter kelly@1.0.0 staking, sliced to all games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Primetime unders — Total A model built on the thesis that primetime totals run high on public over-bias — the under is the disciplined side; trained with calibrated ensemble and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible QB Completions CPOE Edge Quarterbacks with high completion percentage over expectation (CPOE) consistently exceed completion prop lines; this model blends CPOE signal with game-pace and market context to find value on QB completion props. Matches NFL 4/4 slots Full registry match NFL Compatible QB DFS Score Optimizer Daily fantasy QB scoring is driven by passing volume, game script, and opposing secondary quality; this model predicts DraftKings QB points using rolling EPA, CPOE, and market total as inputs to identify high-ceiling tournament plays. Matches NFL 4/4 slots Full registry match NFL Compatible Qb injury — Moneyline A model built on the thesis that the market over-reacts to QB news on the open and under-reacts to backup quality; trained with elastic net and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing moneyline on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible QB Interceptions Under Pressure Quarterbacks facing high defensive pressure have elevated interception rates; this model uses net pressure rate and rolling turnover margin to predict interception prop overs when defenses create consistent chaos. Matches NFL 4/4 slots Full registry match NFL Compatible QB Passing Yards Pass-OE Model Quarterbacks with high pass-attempts over expectation (pass OE) generate more opportunities and exceed yardage props; this model uses pass OE differential, NGS time-to-throw, and total implied points to predict yardage. Matches NFL 4/4 slots Full registry match NFL Compatible QB TD Throws Red Zone Model Quarterbacks with high red-zone touchdown rates and passing volume from high-pace offenses outperform their TD prop lines; this model blends red-zone advantage with EPA and CPOE to capture multi-TD game probability. Matches NFL 4/4 slots Full registry match NBA Compatible Quantile Spread Uncertainty Quantile XGBoost spread model outputs prediction intervals around the spread, allowing confidence-adjusted staking that avoids bets where the interval straddles the posted line. Matches spread 4/4 slots Full registry match NFL Compatible Quick Release This theory claims that quarterbacks who release the ball quickly tend to throw for more yards. Our Shark Snip model tracks quarterback time-to-throw and other key player stats to test this idea and make informed bets on passing yards props. Matches NFL 4/4 slots Full registry match NFL Compatible RB DFS Score LightGBM Running back DFS scores hinge on snap share, game script, and opponent run-defense EPA allowed; this LightGBM model identifies underpriced RBs in favorable game environments for DraftKings GPP lineups. Matches NFL 4/4 slots Full registry match NFL Compatible RB Workload Carries Model Running back carry counts are more predictable than yardage and reflect game-script; this model uses EPA team offense, game total, and player-store features to predict carries props when one RB holds a clear lead-back role. Matches NFL 4/4 slots Full registry match NFL Compatible Red‑Zone Touchdown Edge The theory says that a player’s red-zone snaps and targets predict extra touchdown chances beyond what the betting line shows. The approach tracks each player’s red-zone usage and compares the predicted TD rate to the actual odds to spot mismatches. Matches NFL 4/4 slots Full registry match NBA Compatible Referee Bias Feature Set Adding a referee bias feature to the GBM model will increase the ATS hit rate for NBA games above the current baseline. Matches spread 3/5 slots Partial registry match NFL Compatible Referee Penalty Rate Totals Referee crews with high historical penalty rates and total-points-allowed profiles inflate game totals above market lines; this model uses ref total-points and penalty-rate features to bet totals overs in high-flag games. Matches NFL 4/4 slots Full registry match NBA Compatible Rest Advantage Gbm Kelly A GBM trained on NBA games with rest-advantage >2 days and Kelly-fraction staking will achieve higher ATS than the flat-staked baseline by exploiting schedule-induced fatigue asymmetries. Matches spread 3/6 slots Partial registry match NBA Compatible Rest Advantage Mlp Teams with 2+ days rest advantage show systematic market inefficiency exploitable by MLP with travel and pace features Matches spread 3/6 slots Partial registry match NBA Compatible Rest Advantage Quantile Xgb Training a quantile XGBoost model on NBA games where a team has at least two days of rest, using team form, Elo, pace, and injuries-count features, will yield a higher ATS hit rate than the incumbent baseline on the same walk-forward evaluation window. Matches spread 5/6 slots Partial registry match NFL Compatible Rest diff — Moneyline A model built on the thesis that rest and short-week differentials are under-priced — the fatigued side covers less than the market implies; trained with bayesian ridge and fixed 1u@1.0.0 staking, sliced to all games@1.0.0, will beat the closing moneyline on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Rest diff — Moneyline A model built on the thesis that rest and short-week differentials are under-priced — the fatigued side covers less than the market implies; trained with calibrated ensemble and variance capped@1.0.0 staking, sliced to all games@1.0.0, will beat the closing moneyline on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match MLB Compatible Run Line Full Scoring Env MoE Run line edges exist in multiple scoring environment regimes — high-scoring, pitcher-duel, and balanced. This mixture-of-experts model routes each matchup to the appropriate sub-regime model for accurate run-line prediction. Matches spread 4/4 slots Full registry match MLB Compatible Run Line Home Form Hist-GBM Home teams in strong recent form cover the run line at a significantly higher rate than road teams. Hist-GBM detects the threshold where home form advantage exceeds the run-line price efficiently. Matches spread 4/4 slots Full registry match MLB Compatible Run Line Pitcher xwOBA Calibrated Run line coverage correlates strongly with starter xwOBA differential — elite pitchers suppress runs enough to cover consistently. This calibrated ensemble fuses pitcher quality, ELO gap, and home advantage. Matches spread 4/4 slots Full registry match MLB Compatible Run Line Recent Form ELO Model ELO ratings updated by recent form provide a dynamic measure of team quality that responds faster than season stats. This model blends ELO rating differential with rolling 10-game form for run line prediction. Matches spread 4/4 slots Full registry match MLB Compatible Run Line Runs Allowed Defense Teams that suppress runs cover the run line at a higher clip because covering by 1.5 runs requires only modest offensive production when defense is elite. This model targets defensively dominant run-line covers. Matches spread 4/4 slots Full registry match MLB Compatible Run Line Scoring Variance Edge High scoring variance teams cover run lines in volatile game environments while low-variance teams cover in stable ones. This model features scoring variance as the primary run-line edge signal. Matches spread 4/4 slots Full registry match MLB Compatible Run Line Win Streak Model Teams on hot win streaks often cover the run line at a higher rate than their point spread implies. This model tracks win streaks, run differential momentum, and rest advantage to time the run-line edge. Matches spread 4/4 slots Full registry match NBA Compatible Schedule-Adjusted Spread Teams rated favorably by their record often face an inflated spread when transitioning to tougher opponents. Calibrated ensemble adjusts spread projections using true strength-of-schedule gaps. Matches spread 4/4 slots Full registry match NBA Compatible Soft Schedule Fade Teams coming off a soft schedule look better than they are, and the market overpays for recent results. CatBoost with strength-of-schedule features fades inflated favorites. Matches spread 4/4 slots Full registry match NFL Compatible Targeted Receivers This theory claims that NFL players who get a lot of attention from their quarterbacks and can create space from defenders will get more receptions than expected. Our model watches how these players' target share and separation metrics impact their reception totals to test this i Matches NFL 4/4 slots Full registry match NFL Compatible Team Momentum Moneyline Model Teams on a multi-game winning streak with improving EPA cover moneylines at above-expected rates; this model uses advanced momentum, rolling EPA, and market spread to find ML value on momentum-driven teams. Matches NFL 4/4 slots Full registry match NFL Compatible Turnover Margin ML Edge Rolling turnover margin is one of the strongest predictors of close-game outcomes; this model pairs turnover differential with EPA and Elo to find moneyline value when one team has a large and sustained turnover advantage. Matches NFL 4/4 slots Full registry match NFL Compatible Weather — Total A model built on the thesis that cold-weather games are systematically over-priced on the total — wind and temperature suppress scoring the market under-adjusts for; trained with quantile xgboost and quarter kelly@1.0.0 staking, sliced to cold weather@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Weather — Total A model built on the thesis that cold-weather games are systematically over-priced on the total — wind and temperature suppress scoring the market under-adjusts for; trained with bayesian ridge and fixed 1u@1.0.0 staking, sliced to cold weather@1.0.0, will beat the closing total on a held-out eval window. Kill criterion: ROI below 0 vs the close after 200 graded bets. Matches NFL 5/5 slots Full registry match NFL Compatible Wind & Pace Beat the Total We think wind‑blown, high‑pace games tend to be lower‑scoring than the line suggests. The model watches wind speed, EPA pace and the Vegas total to see if the market is over‑estimating points. Matches NFL 4/4 slots Full registry match NFL Compatible Windy Games & Under‑Friendly Refs Beat the Line We think wind and under‑friendly refs keep scores down, so totals land under the line more than the bookies expect. The model watches wind speed, temperature and referee under‑call rates to test that belief. Matches NFL 5/5 slots Full registry match NFL Compatible WR DFS CatBoost Edge Wide receiver DFS ceilings depend on target share, separation, and pace of game; this CatBoost model identifies WRs primed for high-upside games by weighting NGS route depth and pace features against DraftKings salary. Matches NFL 4/4 slots Full registry match NFL Compatible WR Receiving Yards Pace Stack Wide receivers in high-pace games with large implied totals see inflated target share and yardage totals; this model uses game total, pace-implied volume, and NGS separation to find over value on WR yardage props. Matches NFL 4/4 slots Full registry match NFL Compatible WR Separation Reception Edge Wide receivers with superior NGS separation scores attract more targets and convert them at higher rates; this model pairs separation metrics with route depth to predict reception prop overs for elite route runners. Matches NFL 4/4 slots Full registry match NHL Outside context AdaBoost Focus on Hard-to-Predict Goal Scoring Cases AdaBoost iteratively focuses on the hardest-to-predict goal-scoring cases — slumping snipers and breakthrough role players — building a model that captures tail-end market inefficiencies. Review before applying to this context 4/4 slots Full registry match NBA Outside context AdaBoost Moneyline AdaBoost focuses on misclassified games in training, building a strong ensemble that corrects systematic market errors in moneyline pricing for teams on momentum turns. Review before applying to this context 4/4 slots Full registry match NHL Outside context AdaBoost Points Model Focuses on Breakout Producers AdaBoost iteratively corrects misclassified cases — undervalued breakout producers and overvalued declined veterans — building a model that captures edge cases the market systematically misprices. Review before applying to this context 4/4 slots Full registry match NBA Outside context All-Around Feast Weak defensive teams allow high PRA totals as stars score, rebound, and set up teammates freely. LightGBM maps defensive weakness to all-around stat surplus. Review before applying to this context 4/4 slots Full registry match NBA Outside context B2B Boards Fade Big men lose boxing-out intensity on back-to-back nights, suppressing rebound totals below their season averages. Random forest model targets prop lines that have not adjusted. Review before applying to this context 4/4 slots Full registry match NBA Outside context B2B Rim Protector Fade Shot-blockers lose vertical explosiveness on back-to-back nights, reducing block counts. LightGBM targets block props that remain too high for fatigued rim protectors on B2Bs. Review before applying to this context 4/4 slots Full registry match NBA Outside context Back-to-Back Scoring Fade Tired legs cost buckets. This model watches rest days, minutes logged, and travel miles to spot when a player's points line hasn't dropped enough. Review before applying to this context 4/4 slots Full registry match MLB Outside context Batter DFS Full Feature XGBoost Batter DFS value is determined by the intersection of six independent signal streams. This full-feature XGBoost model integrates all available batter and game-context signals into a single high-coverage DFS score. Review before applying to this context 4/4 slots Full registry match MLB Outside context Batter DFS Platoon Stacking Model DFS lineup construction benefits from stacking batters with favorable platoon matchups. This stacking meta-learner fuses opposite-handed advantage, recent xwOBA, and lineup position into a DFS roster score. Review before applying to this context 4/4 slots Full registry match MLB Outside context Batter DFS Rest Days LightGBM Batters with 2+ days of rest show elevated DFS production versus fatigue-compressed schedules. LightGBM efficiently captures the rest day interaction with lineup spot and recent xwOBA form. Review before applying to this context 4/4 slots Full registry match MLB Outside context Batter ELO-Glicko Rating Model Batter performance against specific pitcher types can be modeled as a head-to-head rating system. This ELO-Glicko model tracks batter vs. pitcher-archetype matchup history to produce calibrated prop predictions. Review before applying to this context 4/4 slots Full registry match MLB Outside context Batter Runs xwOBA Mix LightGBM High-OBP batters lead off innings more and score runs at above-average rates. LightGBM blends xwOBA quality with pitch-mix contact-friendliness to project runs scored above or below the market line. Review before applying to this context 4/4 slots Full registry match MLB Outside context Batter Strikeout AdaBoost Model Batters who chase spin in the zone get called out frequently against high-whiff pitchers. AdaBoost iteratively re-weights the misclassified examples to nail the volatile K count distribution. Review before applying to this context 4/4 slots Full registry match MLB Outside context Batter Strikeout ELO Context Model Batters face more strikeouts when their team is trailing against high-ELO opponents because managers force aggressive swings. This model conditions batter K rate on team ELO deficit and lineup slot position. Review before applying to this context 4/4 slots Full registry match NBA Outside context Bayesian Assist Projector Bayesian Ridge captures uncertainty in assist predictions for playmakers when role changes or lineup shuffles create unstable baselines. Uses priors from season-long patterns. Review before applying to this context 4/4 slots Full registry match NHL Outside context Bayesian Hit Model for Stable Physical Role Players Physical role players maintain highly stable hit rates that Bayesian ridge models with tight posterior distributions. Markets often misprice these stable props, creating reliable over/under edges. Review before applying to this context 4/4 slots Full registry match NHL Outside context Bayesian Save Count Model with Opponent Shot Rate Prior Bayesian ridge save prediction incorporates prior beliefs about opponent shot rates as informative priors, updating efficiently when confirmed starters face known shot-heavy offenses. Review before applying to this context 4/4 slots Full registry match NHL Outside context CatBoost Assist Forecaster for Elite Playmakers Primary playmakers on contending teams facing defensive-minded opponents generate assist totals CatBoost can predict with higher accuracy than the market line. Captures categorical team/role interactions natively. Review before applying to this context 4/4 slots Full registry match NHL Outside context CatBoost Captures Ice Time Role Signals Line assignment, penalty-kill usage, and power-play role interact in complex categorical ways that set ice time distributions. CatBoost captures these categorical interactions without manual encoding. Review before applying to this context 4/4 slots Full registry match NHL Outside context CatBoost Goalie Save Model for Pace-Extreme Matchups Pace-extreme matchups produce non-linear save count distributions that CatBoost handles better than linear models. Identifies goalies who systematically outperform market save props in specific opponent matchup categories. Review before applying to this context 4/4 slots Full registry match NBA Outside context Chaos Ball Turnover Spike High-pace, transition-heavy games lead to more turnovers as players rush decisions. Poisson regression models player turnover rates in chaos-ball environments. Review before applying to this context 4/4 slots Full registry match NHL Outside context Confirmed Starters vs Weak Offenses Dominate DFS Confirmed starting goalies facing offenses in the bottom quartile of shots-per-game score the most DraftKings DFS points in NHL daily lineups. Bayesian ridge models the uncertainty of starts and matchup difficulty. Review before applying to this context 4/4 slots Full registry match MLB Outside context Contact Hitter Singles Model High-contact batters exploit spray-angle tendencies and weak infields. CatBoost captures the categorical interactions between park, pitcher handedness, and batter zone rates that define singles volume. Review before applying to this context 4/4 slots Full registry match NBA Outside context Defense Wins Moneylines Elite defensive units win more games than offense alone predicts. LightGBM uses defensive-rating asymmetry to find moneyline value in defense-decided games. Review before applying to this context 4/4 slots Full registry match NHL Outside context Defensive Blueliners Beat Block Lines in High-Shot Matchups Top defensive pairs against shoot-heavy opponents log blocked shot totals in the upper quantile of their distribution. Quantile regression precisely captures the tail-heavy distribution of blocks in such matchups. Review before applying to this context 4/4 slots Full registry match NBA Outside context Defensive Matchup Scorer Scorers facing weak defensive teams on the road see inflated point props. XGBoost models individual scoring output using defensive rating and matchup quality features. Review before applying to this context 4/4 slots Full registry match NBA Outside context DFS B2B Value Play Stars are rested on B2B nights, opening salary room for backup players who see elevated minutes. CatBoost identifies DFS value plays that emerge when depth is thrust into starting roles. Review before applying to this context 4/4 slots Full registry match NBA Outside context DFS Elo Matchup Finder Players facing teams with lower Elo ratings see inflated DFS value as the matchup advantage creates extra scoring and stat accumulation opportunities underpriced by salary. Review before applying to this context 4/4 slots Full registry match NBA Outside context DFS High-Ceiling GPP Play Tournament DFS requires ceiling plays with high scoring variance. Quantile XGBoost targets the 90th percentile DK outcome for high-variance players in favorable matchups. Review before applying to this context 4/4 slots Full registry match NBA Outside context DFS Home Court Cash Players at home outperform their salary-implied DK ceiling due to consistent home scoring lifts. LightGBM identifies home DFS underpricing before lineup lock. Review before applying to this context 4/4 slots Full registry match NBA Outside context DFS Net Rating Stacker Stacking meta-learner combines net-rating, pace, and home-split features to produce DK salary-adjusted value scores identifying underpriced players on high-efficiency offenses. Review before applying to this context 4/4 slots Full registry match NBA Outside context DFS Pace Optimizer DFS points concentrate in high-pace games where more possessions yield more counting stats. XGBoost selects value players with elevated pace-adjusted DK upside. Review before applying to this context 4/4 slots Full registry match NHL Outside context DFS Stacking: Line-Mates on Pace-Up Slates Full line stacks on DraftKings NHL slates where the matchup creates a high total allow individual skater DFS points to reach GPP-winning ceilings. Calibrated ensemble anchors the floor and ceiling of each player's output. Review before applying to this context 4/4 slots Full registry match NBA Outside context DFS Weak Defense Feast Players facing weak defenses dramatically exceed their DK salary expectations as points, assists, and rebounds pile up. XGBoost maps defensive weakness to DFS upside. Review before applying to this context 4/4 slots Full registry match NBA Outside context Dominant Team Rebounder Big men on dominant net-rating teams spend more time in comfortable positions to secure offensive and defensive boards. CatBoost quantifies the team quality effect on rebound props. Review before applying to this context 4/4 slots Full registry match NBA Outside context Dominant Team Three-Pointer Shooters on high net-rating teams get more open catch-and-shoot opportunities as star players draw attention. XGBoost maps team net-rating to individual three-point prop upside. Review before applying to this context 4/4 slots Full registry match NBA Outside context Dominant Team Turnover Fade Dominant teams force opponents into turnovers but also protect the ball well. XGBoost projects player turnover props using team quality differential. Review before applying to this context 4/4 slots Full registry match MLB Outside context Doubles Pitcher Soft Contact Model Batters hit more doubles when facing fly-ball pitchers who give up soft contact. This model fuses pitcher xwOBA allowed, batter pull-rate, and park gap depth to project doubles production. Review before applying to this context 4/4 slots Full registry match MLB Outside context Earned Runs xwOBA LightGBM Model Pitcher earned runs allowed track closely with xwOBA allowed to contact, making it the best predictor of blowup risk. LightGBM efficiently trees through the interaction of contact quality and run-scoring context. Review before applying to this context 4/4 slots Full registry match NHL Outside context Elastic Net Point Model — Parsimony for Stable Props Elastic net identifies the smallest set of features that best predict point totals for consistent NHL producers, avoiding overfitting on season-to-season noise and finding durable market edges. Review before applying to this context 4/4 slots Full registry match NHL Outside context Elastic Net Regularized Goal Scorer Model Elastic net regularization selects the most predictive features for goal scoring from a large feature set, reducing overfitting and identifying which signals have persistent predictive value across seasons. Review before applying to this context 4/4 slots Full registry match NBA Outside context Elastic Net Total Regressor Elastic Net combines L1 and L2 penalties to select the most informative pace and rating features for total prediction, preventing overfitting during high-collinearity early season periods. Review before applying to this context 4/4 slots Full registry match NBA Outside context Elite Defense Block Count Rim-protecting bigs on elite defensive teams accumulate more blocks as opponents attempt riskier layups under pressure. Random forest identifies block prop undervaluation on top defensive teams. Review before applying to this context 4/4 slots Full registry match NHL Outside context Elite Defensemen Log More Ice Time in Close Games Top-pairing defensemen see dramatically elevated ice time in tight-margin contests, especially in third periods. Quantile regression captures the skewed distribution of TOI under pressure. Review before applying to this context 4/4 slots Full registry match NHL Outside context Elite Goalies Rack Up Saves Against Shoot-First Teams Shot-heavy teams with high peripheral shot rates generate save counts for elite goalies that line setters systematically undervalue. Random forest captures the non-linear interaction between shot quality, volume, and goalie skill. Review before applying to this context 4/4 slots Full registry match NHL Outside context Elite Lines vs Tired Goalies Score DFS Points Top-six forwards stacked on the same line against fatigued or backup goalies produce the highest DraftKings DFS upside. LightGBM models line synergy, goalie fatigue, and pace interactions. Review before applying to this context 4/4 slots Full registry match NBA Outside context Elo Calibrated ML Elo ratings provide a stable long-run power ranking that the market occasionally underweights in favor of recent streaks. Calibrated ensemble converts Elo gap to moneyline probability. Review before applying to this context 4/4 slots Full registry match NBA Outside context Expert Committee Total Mixture-of-experts combines pace-specialist, scoring-variance, and rest-effect sub-models to produce a calibrated game total projection that outperforms single-model baselines. Review before applying to this context 4/4 slots Full registry match MLB Outside context Extra-Base Gap Shot Model Doubles production is a function of gap-hitting power and deep outfield dimensions. Random forest captures the non-linear interaction between batter pull-rate, park doubles factor, and pitch velocity. Review before applying to this context 4/4 slots Full registry match NBA Outside context Fast Game Shot Swatter High-pace teams drive more frequently at the rim, creating extra block opportunities for elite shot-blockers. LightGBM leverages pace features to project block totals accurately. Review before applying to this context 4/4 slots Full registry match NBA Outside context Fast-Game Rebound Spike High-pace games generate more misses and therefore more rebound opportunities. XGBoost finds when big-men rebound props are systematically low in fast-paced matchups. Review before applying to this context 4/4 slots Full registry match NBA Outside context Fast-Paced Dish Machine High-pace games generate more offensive possessions, inflating assist opportunities for primary ball-handlers. XGBoost identifies when assist props are set too low for playmakers in fast games. Review before applying to this context 4/4 slots Full registry match MLB Outside context First-Half Total Starter Form Model First-half totals are a pure starter quality bet before bullpen erosion kicks in. Recent form and pitch-mix evolution determine whether starters will go under their first-half total with high confidence. Review before applying to this context 4/4 slots Full registry match NBA Outside context Full-Feature Assist Projector XGBoost assist model combining player history, pace, defensive rating, home/away, and recent form for complete assist prop coverage across all scheduling and matchup contexts. Review before applying to this context 4/4 slots Full registry match NBA Outside context Full-Feature DFS Optimizer CatBoost DFS model combining player history, pace, defensive rating, home/away, rest, win streak, and SOS for the broadest DK points projection and salary value identification. Review before applying to this context 4/4 slots Full registry match NBA Outside context Full-Feature ML Ensemble Calibrated ensemble moneyline model ingests all NBA team features for the highest-accuracy win probability estimate, converting to fractional Kelly staking for edge extraction. Review before applying to this context 4/4 slots Full registry match NBA Outside context Full-Feature Points Projector CatBoost points model combines player history, defensive rating, pace, home/away split, and rest to produce calibrated points prop projections across all matchup contexts. Review before applying to this context 4/4 slots Full registry match NBA Outside context Full-Feature PRA Projector LightGBM PRA model ingests player store, defensive rating, pace, home/away, and rest features to produce the most comprehensive all-around player prop projection available. Review before applying to this context 4/4 slots Full registry match NBA Outside context Full-Feature Rebound Projector Random forest rebound model combining player history, pace, defensive rating, home split, and SOS for comprehensive rebound prop coverage across all game contexts. Review before applying to this context 4/4 slots Full registry match NBA Outside context Full-Feature Total Ensemble Calibrated ensemble total model ingests all NBA team features including pace, offense, defense, rest, and schedule to produce the most comprehensive game total projection available. Review before applying to this context 4/4 slots Full registry match MLB Outside context Game Over Runs Rate XGBoost Teams trending upward in runs-scored rate often push games over in the short term before books adjust. XGBoost detects this momentum surge using rolling runs-scored windows and team form streaks. Review before applying to this context 4/4 slots Full registry match MLB Outside context Game Total Bayesian Ridge Baseline Bayesian ridge regression provides a principled baseline for game totals by constraining coefficient magnitudes to prevent overfitting on small recent windows. Useful for early-season calibration and reference. Review before applying to this context 4/4 slots Full registry match MLB Outside context Game Total Full Signal LightGBM Game totals require all six scoring environment signals to price correctly. This full-signal LightGBM model integrates pitcher quality, team scoring rates, variance, ELO, and home-away park context. Review before applying to this context 4/4 slots Full registry match MLB Outside context Game Total Home-Away Gaussian Process Game totals exhibit smooth spatial dependence on ballpark run factors that a Gaussian process captures with principled uncertainty. This model provides calibrated uncertainty bounds on total scoring. Review before applying to this context 4/4 slots Full registry match MLB Outside context Game Total Pitcher Rest Gaussian Rested pitchers suppress runs at a higher rate, pushing game totals under. A Gaussian process models the smooth dependence of total scoring on pitcher rest days with principled uncertainty quantification. Review before applying to this context 4/4 slots Full registry match MLB Outside context Game Total Rest Days GBM Model Teams with extra rest tend to score more runs early in games due to fresh lineup legs and reduced fatigue. Gradient boosting identifies the rest-day scoring uplift interaction with ballpark and opponent quality. Review before applying to this context 4/4 slots Full registry match NHL Outside context Gaussian Process Uncertainty for Volatile TOI Markets Ice time props have high game-to-game variance driven by penalties, line changes, and game state. Gaussian processes model the smooth underlying trend and uncertainty intervals for informed over/under positioning. Review before applying to this context 4/4 slots Full registry match NHL Outside context GBM Goals-Against Model for Exposed Netminders Goalies with high recent goals-against rates facing offensive powerhouses are systematically underpriced in goals-against props. GBM captures the compounding interaction of goalie fatigue and opponent scoring depth. Review before applying to this context 4/4 slots Full registry match NHL Outside context GBM Power-Play Point Predictor for PP1 Shooters First-unit power-play shooters on teams with elite PP efficiency generate point totals gradient boosting can predict with strong accuracy. Captures PP time allocation, shot quality, and opponent PK weakness. Review before applying to this context 4/4 slots Full registry match NHL Outside context Goalie Win Probability for Strong Team/Weak Opponent Spots Starting goalies on teams with top-five goal differentials facing teams on losing streaks win at rates far exceeding implied market probability. Bayesian ridge quantifies uncertainty in goalie win props. Review before applying to this context 4/4 slots Full registry match NHL Outside context Grinders Exceed Hit Props in Rivalry Games Physical forwards and defensemen consistently exceed their hit lines in divisional rivalry matchups. Our model identifies underleveraged physicality signals the market ignores. Review before applying to this context 4/4 slots Full registry match NHL Outside context High-Volume Shooters in Pace Matchups — Stacked Meta-Learner When fast-paced teams face each other, total shot attempts spike and individual shot-on-goal counts exceed projections. A stacking meta-learner combines multiple base models for robust shot volume predictions. Review before applying to this context 4/4 slots Full registry match NHL Outside context High-Volume Shot Teams Inflate Goalie Save Counts Goalies facing high-shot opponents in defensive-structure matchups consistently clear the save line. XGBoost captures shot-rate, pace, and team defensive identity interactions. Review before applying to this context 4/4 slots Full registry match NBA Outside context Histogram Boost Points HistGBM projects individual scoring output by processing full player-history feature sets with efficient binning, outperforming standard GBMs in speed while matching accuracy. Review before applying to this context 4/4 slots Full registry match NBA Outside context Histogram Boost Total HistGBM efficiently processes high-dimensional NBA feature sets for game total prediction, with native NaN handling for missing pace and rating data in early-season games. Review before applying to this context 4/4 slots Full registry match NHL Outside context Histogram GBM Finds Goal-Scoring Regime Changes Hist-gradient boosting efficiently handles large feature spaces and detects regime changes in goal-scoring patterns — useful for identifying when a player transitions from slump to hot streak. Review before applying to this context 4/4 slots Full registry match NHL Outside context Histogram GBM Spots Underpriced Point Totals Hist-gradient boosting identifies non-linear feature interactions between ice time, power-play usage, and opponent quality that standard models miss, finding persistent edges on player point total lines. Review before applying to this context 4/4 slots Full registry match MLB Outside context Hits All-Signal Stacking Model Batter hits are predicted best when all signal streams are combined. This stacking meta-learner layer-2 model fuses batter form, pitcher quality, handedness, rest days, and park signals for maximum accuracy. Review before applying to this context 4/4 slots Full registry match MLB Outside context Hits Flow Like a Counting Game Every trip to the plate is a fresh roll, not a season-long average. This Engine watches how pitcher stuff, park dimensions, and lineup protection actually shape hit probability—then sizes the Stack where the market still prices old news. Review before applying to this context 4/4 slots Full registry match MLB Outside context Hits Form-Mix Calibrated Ensemble Batter hits are influenced by form cycles and pitcher pitch-mix adaptation. A calibrated ensemble combines form-based and pitch-mix sub-models to produce probability-calibrated hit over/under predictions. Review before applying to this context 4/4 slots Full registry match MLB Outside context Hits Run-Differential Context Model Batters on teams with strong run differentials tend to receive better counts and more hittable pitches due to low-leverage scoring contexts. This model conditions hit probability on run differential advantage. Review before applying to this context 4/4 slots Full registry match MLB Outside context Hits xwOBA Poisson Dual-Signal Combining both batter and pitcher recent xwOBA into a Poisson model gives a dual-matchup signal that single-sided models miss. Expected hit count is estimated as a product of individual at-bat probability and plate appearances. Review before applying to this context 4/4 slots Full registry match NBA Outside context Home Boards Advantage Rebounders perform better at home due to crowd energy and familiarity with backboard angles. LightGBM models home rebound lift and finds when the market underprices it. Review before applying to this context 4/4 slots Full registry match NBA Outside context Home Court All-Around Star players accumulate more PRA at home due to comfortable rhythm, supportive crowds, and strategic advantages. Random forest models home PRA lift and finds games where it is underpriced. Review before applying to this context 4/4 slots Full registry match NBA Outside context Home Court Heroes This theory claims that NBA players are more productive at home, leading to better than expected fantasy performances. Our model tracks home and away splits in points, rebounds, and assists to test if this edge holds up against the market. Review before applying to this context 4/4 slots Full registry match NBA Outside context Home Court Pickpocket Perimeter defenders generate more steals at home as crowd energy lifts lateral quickness and anticipation. XGBoost models home steal rates versus away for lock-down wing players. Review before applying to this context 4/4 slots Full registry match NBA Outside context Home Floor General Point guards dish more assists at home when the crowd elevates energy and teammates attack more confidently. LightGBM pinpoints home/away assist splits by playmaker type. Review before applying to this context 4/4 slots Full registry match NBA Outside context Home Heavy Rotation Star players log more minutes at home when coaches protect leads and matchup advantages. LightGBM predicts when home-game minutes will exceed book props. Review before applying to this context 4/4 slots Full registry match NHL Outside context Home Hot Streaks This theory claims that NHL players on a hot streak at home will keep scoring more goals than expected. Our model tracks player performance Data and Stats to test if betting on the over for these players is a winning Spot. Review before applying to this context 4/4 slots Full registry match NBA Outside context Home Rim Protector Shot-blockers are more aggressive at home and see more rim-attack attempts from emboldened visitors. XGBoost finds when block props are set below expected home output. Review before applying to this context 4/4 slots Full registry match MLB Outside context Home Run Form CatBoost Kelly Batters in recent power form against fastball-heavy pitchers show elevated HR probability that books are slow to price. CatBoost with Kelly sizing targets the value window before line adjustment. Review before applying to this context 4/4 slots Full registry match MLB Outside context Home Run Mixture-of-Experts Model Home run hitters fall into distinct archetypes — pull-power, all-fields power, and contact-first. This mixture-of-experts model routes each batter to the appropriate sub-model based on batted-ball profile. Review before applying to this context 4/4 slots Full registry match MLB Outside context Home Run Power Model Power hitters hit more home runs against soft contact pitchers at hitter-friendly parks. This model tracks xwOBA matchups, park factors, and batter recent form to find the sweet spot. Review before applying to this context 4/4 slots Full registry match MLB Outside context Home Run Win-Streak Confidence Power hitters on winning teams enter each at-bat with heightened confidence and face more fastballs in hittable counts. This model detects the win-streak confidence uplift on HR probability for top-of-order sluggers. Review before applying to this context 4/4 slots Full registry match NBA Outside context Home Streak Scorer Players on winning streaks at home score more aggressively as confidence compounds. CatBoost identifies when the market under-prices home win-streak scorers. Review before applying to this context 4/4 slots Full registry match NBA Outside context Home Three-Point Specialist Shooters attempt and make more threes at home, yet books often set props from overall averages. LightGBM identifies when three-point props understate the home advantage for shooters. Review before applying to this context 4/4 slots Full registry match NBA Outside context Home Turnover Suppression Players commit fewer turnovers at home due to comfortable surroundings reducing mental errors. LightGBM quantifies home turnover suppression for prop bets. Review before applying to this context 4/4 slots Full registry match MLB Outside context Home-Away Total Runs Split Model Teams score significantly more runs at home than on the road due to lineup construction and park familiarity. This model isolates the home-away scoring split to find game total edges when teams play outside their comfort zone. Review before applying to this context 4/4 slots Full registry match NHL Outside context Home-Ice Advantage Inflates Goal Scorer Props Snipers playing in front of their home crowd benefit from crowd energy and familiar ice, scoring at higher rates than away games. LightGBM quantifies the home-ice amplification effect on individual goal totals. Review before applying to this context 4/4 slots Full registry match NBA Outside context Hot Rebounder Streak Rebounders in strong recent form lock into consistent board-grabbing routines that extend beyond their historical season averages. CatBoost targets rebound prop undervaluation during hot streaks. Review before applying to this context 4/4 slots Full registry match NBA Outside context Hot Scorer Win Streak Star scorers on winning streaks maintain elevated usage and shot volume as coaches let them operate freely in favorable game scripts. XGBoost identifies when points props lag behind streak-driven output. Review before applying to this context 4/4 slots Full registry match NHL Outside context Hot Shooters vs Backup Goalies — Calibrated Ensemble Forwards on multi-game scoring streaks facing confirmed backup goalies historically over-perform their projected goal totals. A calibrated blend of regressors captures non-linear interactions in this high-value spot. Review before applying to this context 4/4 slots Full registry match NBA Outside context Hot Streaks Pay This theory claims that players on a hot streak will keep scoring big when they face weaker defenses at home. Our model watches how these players perform in these situations to see if their hot streaks really do translate to more points than the market thinks. Review before applying to this context 4/4 slots Full registry match NBA Outside context Hot Team DFS Ownership Players on hot-streak teams are under-owned in DFS as casual players follow season averages. LightGBM targets streak-fueled scoring runs for DFS tournament plays. Review before applying to this context 4/4 slots Full registry match NBA Outside context Hot Team Pickpocket Teams on winning streaks play with elevated defensive intensity, boosting steal rates across the roster. XGBoost identifies when individual steal props are mispriced during team hot streaks. Review before applying to this context 4/4 slots Full registry match NHL Outside context K-Nearest Neighbor Goal Scorer Similarity Model KNN finds historically similar player-game situations — same player type, same matchup quality, same team momentum — and uses those comparable games to predict goal output beyond what parametric models capture. Review before applying to this context 4/4 slots Full registry match NBA Outside context Leak Hunt Playmaker Weak defensive teams allow penetration and kick-outs, creating extra assist chances for creators. CatBoost targets assist props for playmakers facing poor defensive units. Review before applying to this context 4/4 slots Full registry match NHL Outside context Linear Model for Stable Playmaker Assist Markets Established playmakers have highly stable assist rates that a simple linear regression can model effectively, identifying systematic over-lines in conservative market pricing. Review before applying to this context 4/4 slots Full registry match NBA Outside context Load Management DNP Fade Star players are rested on the second game of back-to-backs, but books set minute lines too high. XGBoost predicts load-management minute restrictions before official injury reports. Review before applying to this context 4/4 slots Full registry match NHL Outside context MARS Regression Finds Non-Linear Thresholds in Hit Props MARS (Multivariate Adaptive Regression Splines) detects non-linear threshold effects in physical play — certain role players dramatically increase hits above specific usage thresholds. Review before applying to this context 4/4 slots Full registry match NBA Outside context Minute Load Scheduler Coaches manage minute loads based on upcoming schedule difficulty. LightGBM projects minutes props combining SOS with back-to-back and win-streak context. Review before applying to this context 4/4 slots Full registry match NHL Outside context Mixture-of-Experts DFS Skater Model for Slate Variance NHL DFS slates have multiple game-environment regimes (back-to-back, high total, late-season) that require different models. MoE routes predictions through specialized sub-models for each slate type. Review before applying to this context 4/4 slots Full registry match NHL Outside context Mixture-of-Experts Goalie Win Prediction Model Goalie wins depend on team quality, matchup, and individual save performance in distinct regimes. MoE assigns predictions to specialized sub-models for different opponent strength categories and game contexts. Review before applying to this context 4/4 slots Full registry match MLB Outside context MLB Total Bases Total bases from the MLB player store into XGBoost. Fixed-1u. The power-hitter prop with the cleanest signal. Review before applying to this context 4/4 slots Full registry match MLB Outside context Moneyline ELO Graph Network Team strength relationships form a network where ELO ratings and recent form create graph edges. This GNN model captures the directional win-probability signal that flat stat models miss. Review before applying to this context 4/4 slots Full registry match MLB Outside context Moneyline ELO Linear Probability Simple ELO-based win probability provides a clean baseline that outperforms naive moneyline pricing in low-information environments. Linear regression converts ELO difference to calibrated win probability. Review before applying to this context 4/4 slots Full registry match MLB Outside context Moneyline Pitching Depth Model Teams with deeper bullpen quality win close games at higher rates. This model uses team runs allowed rate, pitcher xwOBA, and rest days to score pitching depth and identify moneyline value. Review before applying to this context 4/4 slots Full registry match MLB Outside context Moneyline Run Differential MARS Run differential is one of the best predictors of future win probability, but the relationship is non-linear. MARS regression discovers the hinge points where run differential edges collapse or amplify. Review before applying to this context 4/4 slots Full registry match MLB Outside context Moneyline Scoring Variance XGBoost High scoring variance teams upset favorites more often than their ELO implies due to big-inning explosiveness. XGBoost captures when variance creates an upset edge that the moneyline underprices. Review before applying to this context 4/4 slots Full registry match NHL Outside context More Ice = More Shots When a skater's minutes and matchup usage tick up, shot volume follows. This Engine tracks ice-time trends, line assignments, and opponent goalie rotations to Spot shooters the books are sleeping on. Review before applying to this context 4/4 slots Full registry match NHL Outside context More Ice Time Means More Points We think heavy‑minute players tend to beat their point lines. Our Engine tracks ice time and recent point stats to spot the over‑performers. Review before applying to this context 4/4 slots Full registry match MLB Outside context Multi-Stat Batter Multi-Task Model Hits, runs, and RBIs are correlated outputs from the same at-bat events. Multi-task linear regression shares information across correlated outputs to reduce variance and improve all three props simultaneously. Review before applying to this context 4/4 slots Full registry match NBA Outside context Nearest Neighbor Scorer KNN regression finds historically similar pace-and-matchup scenarios and averages the scoring outcomes, providing a non-parametric baseline for player points props. Review before applying to this context 4/4 slots Full registry match NBA Outside context Net Rating Edge Teams with superior net ratings covering both offense and defense consistently beat closing moneylines. LightGBM captures non-linear net-rating thresholds the market underweights. Review before applying to this context 4/4 slots Full registry match NBA Outside context Net Rating Total Projector Combined net ratings of both teams strongly predict game totals. Bayesian Ridge quantifies uncertainty and avoids overfitting to short recent-game samples. Review before applying to this context 4/4 slots Full registry match MLB Outside context Over Scoring Environment Model High-scoring environments emerge when two weak rotations meet in a hitter-friendly park with wind blowing out. This model scores the game environment using run differential, scoring variance, and probable starter quality. Review before applying to this context 4/4 slots Full registry match NBA Outside context Pace Dog Moneyline Slow-pace underdog teams that control tempo surprise fast-pace favorites who cannot find their rhythm. XGBoost finds moneyline value for pace-controlling underdogs in speed-versus-control matchups. Review before applying to this context 4/4 slots Full registry match NBA Outside context Pace Mismatch Scorer Slow-team players scoring props drop when they face fast-paced defenses, creating fade opportunities. GBM models the pace delta and its effect on individual point output. Review before applying to this context 4/4 slots Full registry match NBA Outside context Pace-Adjusted Totals When two fast-paced teams meet, the total line rarely adjusts enough for extra possessions. XGBoost trained on pace matchups finds over-under edges before the market catches up. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher DFS Home Advantage XGBoost Home pitchers face more favorable mound conditions and umpire familiarity, yielding higher DFS ceilings. XGBoost captures the home advantage on pitcher DFS output beyond what xwOBA alone explains. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher DFS Meta-Stack Model Pitcher DFS value peaks when ownership is low and matchup quality is high. This stacking meta-learner fuses strikeout rate, xwOBA allowed, and rest context into a single DFS-optimized score. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher DFS Variance-Capped XGBoost Pitcher DFS outcomes have fat tails from blowups and dominant gems. XGBoost with variance-capped staking identifies the right ceiling plays to target without over-allocating to volatile outcomes. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher DFS xwOBA Quantile Model Pitchers facing lineups with low xwOBA against right-handers produce higher DFS floors. This quantile model outputs the 25th-75th percentile range to identify safe DFS targets and ceiling plays. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher K Full-Season Prophet Trend Pitcher strikeout rates follow seasonal stamina curves — high early, fatigue dips mid-season, then playoff pushes. Prophet captures these trends plus the weekly rest cycle for projection stability. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher K Home-Away Split Forest Many pitchers show significant home-road K differentials due to crowd noise effect on batter timing and familiar mound conditions. Random forest identifies the home-away K split most likely to be mispriced. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher K Run Support Forest Pitchers on teams with strong run differentials receive more strikeout attempts per 9 innings because they work deep into counts rather than pitching to contact. Random forest identifies this run support interaction. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Outs ELO Quality Model Pitchers facing weaker teams by ELO rating record more outs because managers let them work through weaker lineups. This model uses team ELO differential to adjust outs-recorded projections upward for dominant matchups. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Outs Pitch-Mix Hist-GBM Pitchers who recently shifted to a slider-heavy mix retire batters faster and record more outs per start. Hist-GBM identifies the pitch-mix shift signal that predicts extended outings and deep-count efficiency. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Outs Prophet Trend Model Pitcher workload follows seasonal and weekly cycles that trend models capture better than point-in-time regressors. Prophet decomposes pitcher outs into trend, weekly rest cycle, and recent form components. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Outs Recorded Quantile Starting pitcher outs recorded are driven by opposing lineup wOBA and manager hook tendencies. This quantile model gives calibrated intervals for deep-dish vs. early-exit pitchers. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Outs Variance-Adjusted Model High-variance offensive teams shorten pitcher outings through early big innings. This model adjusts outs-recorded projections downward when the opposing team shows high scoring variance and run momentum. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Walk Rate Bayesian Edge Wild pitchers issue walks at a rate the market consistently underweights. This Bayesian model tracks pitcher control metrics and batter patience to find the edge. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Win ELO Matchup Model Pitcher wins correlate strongly with team ELO gap favoring the pitcher team. Larger ELO advantages translate to more run support AND lower opposing run output, jointly driving win probability above market price. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Win Full Context MoE Pitcher wins emerge from three overlapping conditions: quality start, run support, and bullpen quality. This mixture-of-experts model routes to separate sub-models based on which condition dominates for each game. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Win Pitch-Mix Forest Model Pitchers who have sharpened their two-pitch approach in recent starts show better command and earn wins at a higher rate. Random forest identifies which pitch-mix evolutions correlate with win probability uplift. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Win Probability Elastic Net Pitcher wins depend on run support AND quality of start in a joint-probability structure that naive win-total lines miss. Elastic net regularization handles the collinear offensive support features cleanly. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitcher Win Random Forest Model Pitcher wins are a noisy outcome jointly determined by quality of start, run support, and bullpen performance. Random forest ensembles across hundreds of decision boundaries to denoise this joint probability. Review before applying to this context 4/4 slots Full registry match MLB Outside context Pitching Duel Moneyline Underdog When two elite starters face off in a pitching duel, underdogs win at a higher rate because the game outcome is effectively a coin flip after neutralizing pitching advantage. This model identifies those neutralization zones. Review before applying to this context 4/4 slots Full registry match NHL Outside context Player ELO Ratings Predict Scoring Point Upside A player-level ELO/Glicko rating system that tracks consistent scorers vs opponents captures market inefficiency in point props. Players with rising ELO ratings facing opponents with declining ratings are systematically underpriced. Review before applying to this context 4/4 slots Full registry match NHL Outside context Player ELO Supplemented Shot Volume XGBoost Incorporating player-level ELO ratings into XGBoost shot models captures individual skill-level trends that raw statistics miss. Rising ELO players consistently outperform static shot props. Review before applying to this context 4/4 slots Full registry match NBA Outside context Playmaker Elo Rating Playmaker Elo tracks assist rates adjusted for the strength of defenses faced. When a point guard has a large Elo advantage over the opposing defense, assist props are mispriced. Review before applying to this context 4/4 slots Full registry match NHL Outside context Playmakers on High-Scoring Teams Beat Assist Lines Primary playmakers on top-ten scoring teams generate assists at rates the market consistently underestimates, especially in high-paced home games. Poisson regression models the count distribution of assists accurately. Review before applying to this context 4/4 slots Full registry match NHL Outside context Poisson Count Model for Elite Sniper Goal Props Goal scoring follows Poisson-like count distributions, and elite snipers' game-level rates diverge from market props in predictable ways. Poisson regression captures the count structure of goal events better than standard regression. Review before applying to this context 4/4 slots Full registry match NHL Outside context Power Play Facilitators Beat Assist Lines Power-play specialists who see heavy PP ice time generate more assists than sportsbooks price in. This model uses XGBoost to exploit market inefficiency in assist lines for top PP units. Review before applying to this context 4/4 slots Full registry match NHL Outside context Power-Play Quarterbacks Exploit Weak Penalty Kills Offensive defensemen who quarterback the first power-play unit against teams with bottom-third penalty kill rates accumulate PP points at rates well above their season averages. XGBoost captures unit-quality and matchup interactions. Review before applying to this context 4/4 slots Full registry match NHL Outside context PP Point Regime-Switching via Mixture-of-Experts Power-play production varies dramatically by opponent PK quality, PP unit personnel, and game flow. MoE allocates predictions across sub-models calibrated for different power-play production regimes. Review before applying to this context 4/4 slots Full registry match NHL Outside context PP1 Producers vs Weak Penalty Killers First-unit power-play regulars facing teams with bottom-third penalty-kill rates generate more PP points than sportsbooks price. Poisson regression captures the count-based nature of PP point accumulation. Review before applying to this context 4/4 slots Full registry match NBA Outside context Quantile Points Projector Quantile XGBoost models the full distribution of player scoring, flagging when a prop line sits above the 75th percentile of predicted outcomes — a systematic fade signal. Review before applying to this context 4/4 slots Full registry match NBA Outside context Quantile PRA Projector Quantile XGBoost predicts the 25th and 75th percentile PRA outputs, identifying when player prop lines sit above the upper quartile or below the lower quartile of the distribution. Review before applying to this context 4/4 slots Full registry match NHL Outside context Quantile XGBoost Goalie Save Distribution Model Save count distributions have heavy right tails when strong goalies face shoot-heavy teams. Quantile XGBoost models the 75th-percentile save ceiling for over prop betting in high-shot-volume matchups. Review before applying to this context 4/4 slots Full registry match NHL Outside context Random Forest Block Model for Defensive Blueliners Defensive defensemen on shot-suppressing teams generate consistent blocked shot counts that random forest models with low variance. Identifies reliable over bets on block props for high-usage defensive pairs. Review before applying to this context 4/4 slots Full registry match NHL Outside context Random Forest Goal Prediction with Team Context Random forest aggregates diverse signal trees capturing shooting percentage regression, goalie strength, ice time, and team scoring pace to generate accurate goal prop predictions. Review before applying to this context 4/4 slots Full registry match NHL Outside context Random Forest Goalie DFS Projections with Confirmation Gate Random forest-based goalie DFS projections gated by starting lineup confirmation outperform unconstrained models. Leverages team defensive identity and pace features alongside confirmation timing signals. Review before applying to this context 4/4 slots Full registry match NHL Outside context Random Forest Ice Time Predictor for Defensive Pairs Top defensive pair ice time is more predictable than forward ice time due to consistent role assignments. Random forest captures the interaction of defensive pairing rank, opponent forward usage, and game script. Review before applying to this context 4/4 slots Full registry match NHL Outside context Random Forest Point Total Predictor for Role Players Role players' point totals are highly sensitive to line assignment, PP usage, and matchup — complex interactions that random forest models effectively without manual feature engineering. Review before applying to this context 4/4 slots Full registry match MLB Outside context RBI Opponent Runs Allowed Edge Playing against a team that allows many runs creates more RBI opportunities via base traffic. This model pairs individual batter quality with opposing team runs-allowed rate to identify inflated RBI props. Review before applying to this context 4/4 slots Full registry match MLB Outside context RBI Pitcher xwOBA SVR Model RBI opportunities multiply when facing pitchers who allow high xwOBA because more runners reach base. SVR with an RBF kernel smooths the non-linear relationship between pitcher contact quality and RBI output. Review before applying to this context 4/4 slots Full registry match MLB Outside context RBI Run Production Engine RBIs cluster around lineup spots that see high-OBP runners ahead of them. This engine models run-scoring opportunity via batting order context, recent form, and team run differential. Review before applying to this context 4/4 slots Full registry match MLB Outside context RBI Win-Streak Momentum XGBoost Offenses on win streaks cluster RBIs across multiple hitters in the lineup due to sustained pressure. XGBoost identifies batters who benefit most from their team streak context based on lineup spot and recent form. Review before applying to this context 4/4 slots Full registry match NBA Outside context Rebound Surge This theory claims that players tend to grab more rebounds in games with a faster pace, giving them an edge over their props. Our model tracks pace, minutes played, and opponent rebounding stats to test whether this edge is real and exploitable. Review before applying to this context 4/4 slots Full registry match NBA Outside context Rebounder Elo Rating Rebounder Elo tracks board-grabbing ability normalized by opposing rebounding strength. Matchups with a large Elo gap for rebounders signal mispriced rebound props. Review before applying to this context 4/4 slots Full registry match NBA Outside context Recent Form Points Lift Players in strong recent scoring form exceed their season-average points lines. LightGBM tracks short-window scoring trends that precede sustained prop line undervaluation. Review before applying to this context 4/4 slots Full registry match NHL Outside context Regime-Switching Points Model for Top-Six Forwards Top-six forwards operate in distinct scoring regimes — hot streaks vs slumps — that standard regression misses. Mixture-of-experts routes predictions through the right sub-model for each regime. Review before applying to this context 4/4 slots Full registry match NBA Outside context Rest and Schedule Total Totals are suppressed when both teams are on B2Bs coming off a tough schedule. LightGBM jointly models rest and schedule difficulty to project reduced scoring. Review before applying to this context 4/4 slots Full registry match NHL Outside context Role Players Exceed Hit Lines on Penalty-Kill Units Penalty-kill specialists and defensive role players log more hits when matched against high-usage power-play units. XGBoost identifies role players the market prices as scorers but who generate outsized physical value. Review before applying to this context 4/4 slots Full registry match NBA Outside context Run-And-Gun Three-Pointer Fast-paced games produce more three-point attempts as teams push transition. Poisson regression models expected three-pointers made in high-pace matchups. Review before applying to this context 4/4 slots Full registry match MLB Outside context Runs Scored Poisson Counter Runs scored are a counting event shaped by lineup position and team run-scoring environment. This Poisson model targets the over on fast-scoring offenses with strong home splits. Review before applying to this context 4/4 slots Full registry match MLB Outside context Runs+RBIs Joint Hist-GBM Model Runs and RBIs are jointly determined by the same offensive environment signals. Hist-GBM with the combined offensive context features surfaces lineup slot and run-environment interactions efficiently. Review before applying to this context 4/4 slots Full registry match NBA Outside context Schedule Adjusted Boards Rebound prop lines set from raw averages do not adjust for schedule quality. XGBoost projects rebound output corrected for opponent rebounding weakness. Review before applying to this context 4/4 slots Full registry match NBA Outside context Schedule Boost Pickpocket Perimeter defenders steal more against weak offensive teams that rely on high turnover rates. CatBoost adjusts steal props for schedule quality and opponent offensive efficiency. Review before applying to this context 4/4 slots Full registry match NBA Outside context Schedule Boost Playmaker Playmakers accumulate more assists against weak defensive teams. Random forest combines SOS and defensive rating to find assist prop undervaluation. Review before applying to this context 4/4 slots Full registry match NBA Outside context Schedule-Corrected Scorer Player scoring props are set from raw averages that do not adjust for schedule quality. XGBoost corrects for SOS and projects true scoring expectation against the current opponent. Review before applying to this context 4/4 slots Full registry match NBA Outside context Scorer Elo Rating Player-level Elo/Glicko ratings track scoring ability adjusting for defensive quality faced. When a scorer faces a weaker defensive Elo opponent, points props are systematically underpriced. Review before applying to this context 4/4 slots Full registry match NHL Outside context Seasonal Goal-Scoring Trend Predicts End-of-Season Surges NHL scorers exhibit predictable seasonal trends — early-season adjustment, mid-season peak, playoff push — that Prophet time-series models capture well, finding systematic over/under edges on point totals late in the season. Review before applying to this context 4/4 slots Full registry match NBA Outside context Shoot Over Leaky Defense Defenses that surrender high three-point percentages inflate shooter volume and accuracy. CatBoost finds matchups where books underestimate shooter upside against porous defenses. Review before applying to this context 4/4 slots Full registry match NHL Outside context Shot-Blocking Defenders Beat the Line on Defense-First Teams Defensive stalwarts on low-scoring franchises generate shot-block counts the market consistently underestimates. LightGBM captures complex interactions between role, minutes, and opponent shot volume. Review before applying to this context 4/4 slots Full registry match MLB Outside context Singles Home Run-Differential Model Batters playing for dominant teams face more aggressive pitching that paradoxically inflates singles. When runs are already secured, pitchers attack the zone more, boosting contact and singles rates. Review before applying to this context 4/4 slots Full registry match MLB Outside context Singles xwOBA Elastic Net Model Singles rates are the most stable component of batting average, shaped by contact quality and batted-ball profile. Elastic net regularization cuts through the multicollinearity in xwOBA-derived features. Review before applying to this context 4/4 slots Full registry match NHL Outside context Stacked Ensemble for Consistent Scorers Point Props Consistently productive NHL scorers have stable underlying distributions that stacking ensembles model better than single learners. Combines XGBoost, LightGBM, and Bayesian ridge via a meta-learner for refined point prop accuracy. Review before applying to this context 4/4 slots Full registry match NBA Outside context Stacked ML Predictor A stacking meta-learner blends predictions from pace, net-rating, and form base models into a final moneyline probability superior to any single model alone. Review before applying to this context 4/4 slots Full registry match NBA Outside context Stacked PRA Machine A stacking meta-learner combines defensive rating, pace, and home/away base-model outputs into a calibrated PRA projection for all-around NBA players. Review before applying to this context 4/4 slots Full registry match NHL Outside context Standard GBM Shot Forecaster for Top-Line Shooters Gradient boosting captures the cumulative effect of ice time, line chemistry, and opponent defensive strength on individual shot volume. Reliable baseline model for first-line NHL shot props. Review before applying to this context 4/4 slots Full registry match MLB Outside context Stolen Base Speed Model Elite base stealers run more against slow catchers and right-handed pitchers with long deliveries. This gradient boosting model integrates speed metrics, matchup context, and rest days. Review before applying to this context 4/4 slots Full registry match MLB Outside context Stolen Base Win-Streak Poisson Teams on win streaks run the bases more aggressively, leading to above-average stolen base attempts by lead-off speedsters. This Poisson model conditions SB probability on team streak context. Review before applying to this context 4/4 slots Full registry match MLB Outside context Stolen Bases Full Context XGBoost Stolen base success combines speed, handedness matchup, rest, and team strategy. This full-context XGBoost model integrates all four factors for the most comprehensive stolen base probability estimate available. Review before applying to this context 4/4 slots Full registry match MLB Outside context Stolen Bases Home-Away CatBoost Speedsters steal more on the road because visiting catchers have unfamiliar pop time to that base. CatBoost captures the home-road, handedness, and rest interactions that shape stolen base opportunity. Review before applying to this context 4/4 slots Full registry match NBA Outside context Streak Breakers Long win streaks attract public money that distorts moneylines. Random forest model fades teams on hot streaks past their true win-probability ceiling. Review before applying to this context 4/4 slots Full registry match NBA Outside context Streak Shooter Premium Shooters in recent hot form shoot more confidently from deep and generate more three-point attempts as teammates feed them deliberately. XGBoost identifies streak-inflated three-point upside. Review before applying to this context 4/4 slots Full registry match NBA Outside context Streaker Moneyline Edge Win-streak teams attract sharp money but also massive public backing that inflates true moneyline price. Random forest isolates when a streak team is correctly priced versus overvalued. Review before applying to this context 4/4 slots Full registry match NBA Outside context Streaking Triple-Threat Players on personal scoring streaks in fast-paced games generate elite PRA outputs. CatBoost combines pace and recent-form features to identify when PRA lines are too conservative. Review before applying to this context 4/4 slots Full registry match MLB Outside context Strikeout Spot This theory claims that pitchers who dominate at home will keep racking up strikeouts. Our model tracks pitcher performance Data and Stats to test if this Spot is a reliable Bet sizing opportunity. Review before applying to this context 4/4 slots Full registry match MLB Outside context Strikeout Win-Streak Confidence Model Pitchers on extended win streaks show heightened confidence and command, producing above-expected strikeout totals. This LightGBM model surfaces the momentum signal layered on top of raw matchup quality. Review before applying to this context 4/4 slots Full registry match NHL Outside context SVR Goal Model for High-Variance Sniper Props Support vector regression with RBF kernel finds robust decision boundaries in goal-scoring feature space, particularly effective for snipers with volatile short-term shooting percentages that mask true talent. Review before applying to this context 4/4 slots Full registry match NBA Outside context SVR Total Model Support Vector Regression is robust to outlier scoring games that distort GBM models during blowouts or overtime. SVR finds stable total predictions on normalized pace and rating features. Review before applying to this context 4/4 slots Full registry match NHL Outside context Team Goal-Differential Context Sharpens Individual Goal Props Individual goal props are mispriced when team context — goal differential, recent form, home ice — is not properly integrated. XGBoost with team-level features corrects this systematic market inefficiency. Review before applying to this context 4/4 slots Full registry match MLB Outside context Team Total ELO Gap Hist-GBM When a team with a large ELO advantage plays at home, their team total is systematically underpriced. Hist-GBM captures this ELO gap threshold effect with park factor and home scoring rate. Review before applying to this context 4/4 slots Full registry match MLB Outside context Team Total Runs Scored Model Team totals are priced with less efficiency than game totals. This model isolates one team run-scoring output using recent form, opposing pitcher quality, and ballpark run factor. Review before applying to this context 4/4 slots Full registry match NBA Outside context Tired All-Around Fade Stars accumulate fewer combined points, rebounds, and assists on back-to-back nights as fatigue spreads across all stat categories. XGBoost quantifies the B2B PRA discount. Review before applying to this context 4/4 slots Full registry match NBA Outside context Tired Legs Under Back-to-back games suppress scoring as fatigue slows offensive execution. LightGBM quantifies how much a back-to-back depresses scoring relative to the set total line. Review before applying to this context 4/4 slots Full registry match NBA Outside context Tired Shooter Fade Shooters on back-to-back nights see compressed shot selection and lower three-point attempt volume. LightGBM identifies when three-point props are set too high for fatigued players. Review before applying to this context 4/4 slots Full registry match NHL Outside context Top-Line Shooters Hit High Shot Quantiles Against Weary Goalies First-line forwards against goalies playing a back-to-back game hit the upper tail of their shot distribution more reliably than the market prices. Quantile XGBoost captures the asymmetric upside in these situations. Review before applying to this context 4/4 slots Full registry match MLB Outside context Total Bases Poisson Matchup Model Total bases are a counting process driven by matchup quality at the at-bat level. Poisson regression models each base-earning event as an independent rate process shaped by batter power and pitcher flyball rate. Review before applying to this context 4/4 slots Full registry match MLB Outside context Total Bases Rest Days AdaBoost Batters with extra rest produce more total bases by swinging harder in favorable counts. AdaBoost iteratively focuses on the hard-to-classify power outbursts that rest-rested sluggers produce against tiring pitchers. Review before applying to this context 4/4 slots Full registry match MLB Outside context Total Bases Team Streak GBM Batters on win-streak teams swing with more conviction and see more fastballs in favorable counts, boosting extra-base hit rates. GBM detects the streak-powered extra-base surge before books adjust. Review before applying to this context 4/4 slots Full registry match MLB Outside context Total Bases xwOBA CatBoost Model xwOBA is the best single predictor of future total base production. CatBoost handles the categorical park-pitcher-batter interaction that inflates or deflates expected extra-base hit rates. Review before applying to this context 4/4 slots Full registry match NBA Outside context Transition Pickpocket Fast-paced games create more loose-ball situations and transition defense moments, elevating steal opportunities. Poisson models steal counts in high-pace environments. Review before applying to this context 4/4 slots Full registry match NHL Outside context Two-Way Forwards Beat Assist Lines in Transition Games Skilled two-way forwards generate rush opportunities in transition-heavy matchups, creating assist volume that books underestimate. Random forest captures non-linear interactions across usage, matchup, and team pace. Review before applying to this context 4/4 slots Full registry match NHL Outside context Uncertain Shot Markets Exploited with Bayesian Modeling Shot-on-goal lines have high uncertainty in backup-goalie situations and pace-extreme matchups. Bayesian ridge identifies where the market's uncertainty is priced incorrectly, finding persistent edge in volatile shot props. Review before applying to this context 4/4 slots Full registry match NBA Outside context Usage Fade on Tired Legs High-usage players see scoring drops on back-to-back nights that books fail to fully price in. LightGBM quantifies the fatigue discount on player points props. Review before applying to this context 4/4 slots Full registry match NBA Outside context Variance Exploiter High-variance teams produce extreme scoring outcomes that standard total lines cannot price efficiently. Quantile XGBoost finds when over or under is structurally mispriced. Review before applying to this context 4/4 slots Full registry match NBA Outside context Volume Shooters Bounce Back This belief says high-volume NBA shooters are easier to bet Over when their role is steady, even if the last box score looks ugly. Shark Snip watches minutes, three-point attempts, shot quality, and the posted prop to spot made-three overs the market may be pricing too low. Review before applying to this context 4/4 slots Full registry match MLB Outside context Walk Rate Pitcher Control XGBoost Batter walk rates spike when facing pitchers with poor zone rate and high walk tendency. XGBoost captures the interaction between pitcher control zone rate, batter plate discipline, and game situation context. Review before applying to this context 4/4 slots Full registry match MLB Outside context Walks Win-Streak Aggression Model Pitchers in losing streaks tend to challenge batters more aggressively to avoid extended innings, reducing walk rates. Gradient boosting detects this pitcher aggression shift conditioned on team win-streak context. Review before applying to this context 4/4 slots Full registry match NHL Outside context Weak Goalies Let In More Goals vs Explosive Offenses Goalies with poor recent form facing offenses that rank in the top quartile for shot quality allowed more goals than the market prices in their goals-against props. CatBoost captures goalie fatigue and offensive power interactions. Review before applying to this context 4/4 slots Full registry match NHL Outside context Win-Streak Teams Score More on Momentum Teams on extended win streaks generate more scoring chances and convert at higher rates due to positive momentum and system confidence. CatBoost identifies which individual players benefit most from team momentum. Review before applying to this context 4/4 slots Full registry match NBA Outside context Winning Team Playmaker Playmakers on winning-streak teams dish more assists as confidence and ball movement improve. Random forest identifies when assist props are systematically low for players on hot-streak teams. Review before applying to this context 4/4 slots Full registry match NHL Outside context XGBoost Goalie Win Probability in Favorable Matchups XGBoost goalie win prediction combines team strength, matchup quality, recent goalie form, and confirmation timing into accurate binary win probability, finding edges when implied market probability diverges from model probability. Review before applying to this context 4/4 slots Full registry match NHL Outside context XGBoost Goals-Against Forecaster for Struggling Goalies XGBoost captures multi-way interactions between goalie form, team defensive structure, and opponent offensive threat level to identify when goals-against props are set too low for struggling netminders. Review before applying to this context 4/4 slots Full registry match NHL Outside context XGBoost Skater DFS Optimizer for High-Total Slates XGBoost DFS point projections for NHL skaters excel on high-total slates where offensive production is elevated. Feature interactions between implied game totals, team pace, and individual usage drive accurate DFS floor/ceiling predictions. Review before applying to this context 4/4 slots Full registry match Build from scratch by adding components into the required workflow slots.
Blocks ensemble-champion-blend ensemble
feat-elo-power-ratings feature
feat-nba-player-feature-store feature
game-spread.py target
linear-regression.py architecture
nba-player-points.py target
nfl-team-game-log data
nws-wind feature
rest_advantage feature
slice-primetime-games slice
staking-fixed-1u staking
staking-quarter-kelly staking
vegas-spread-line feature
xgboost-regressor.py architecture
Data Source tables and feeds. multi 1 NFL Team Game Log nfl-team-game-log compatible v1.0.0 Features Signals the model learns from. req multi 4 Vegas Spread Line vegas-spread-line recommended v1.0.0 Elo Power Ratings feat-elo-power-ratings recommended v1.0.0 NWS Wind nws-wind recommended v1.0.0 Rest Advantage rest_advantage compatible v1.0.0 Target The market or outcome to predict. req 1 Game Spread Target game-spread.py recommended v1.0.0 Slice Optional situational filter. 1 Primetime Games slice-primetime-games compatible v1.0.0 Model Training architecture. req 2 Linear Regression linear-regression.py compatible v1.0.0 XGBoost Regressor xgboost-regressor.py compatible v1.0.0 Staking Bet sizing rules. req 2 Fixed 1U staking-fixed-1u compatible v1.0.0 Quarter Kelly staking-quarter-kelly compatible v1.0.0 Ensemble Optional model blend. 1 Champion Blend ensemble-champion-blend compatible v1.0.0 Relevant 12 Selected 0 Completeness: 0%
Features 0 bricks missing Drop feature brick Target 0 bricks missing Drop target brick Model 0 bricks missing Drop architecture brick Staking 0 bricks missing Drop staking brick
