AI model predictions. For entertainment and educational use. Past performance does not guarantee future results. Not financial or gambling advice.
Model Backtest
Only counts predictions where predictions.generated_at < events.starts_at — i.e., the model picked the game BEFORE it was played. Diagnostic confirmed
~99.998% of historical "predictions" in this database were training-time
backtests written after games settled (metadata.actual_value populated, generated_at years after the game).
Pre-filter the page would have shown 33 "+EV" buckets at 75–85% cover —
all in-sample training accuracy, none real edge. Post-filter, only forward
picks survive. Most models won't have any until the prediction pipeline runs
ahead of upcoming slates with the leak fixed upstream.
Per-(model × sport × market) cover rate vs −110 break-even (52.38%). Built on 1,494 legit forward picks. Once the prediction pipeline starts emitting genuine pre-game forecasts, this page will populate automatically — the nightly grader cron refreshes the MV.
| Model | Sport / Market | N | Record | Cover | Edge | ROI/pick | Units | Verdict |
|---|---|---|---|---|---|---|---|---|
[Auto] MLB Run Line (HistGradientBoosting) | MLB · Spread | 115 | 0-0 | 0.0% n=0 · insufficient | — | 0.0% n=0 · insufficient | +0.0u | n<5 |
[Auto] MLB Run Line (Gaussian Process) | MLB · Spread | 115 | 0-0 | 0.0% n=0 · insufficient | — | 0.0% n=0 · insufficient | +0.0u | n<5 |
[Auto] MLB Run Line Model (XGBoost) | MLB · Spread | 115 | 0-0 | 0.0% n=0 · insufficient | — | 0.0% n=0 · insufficient | +0.0u | n<5 |
[Auto] MLB Run Line (MARS) | MLB · Spread | 115 | 0-0 | 0.0% n=0 · insufficient | — | 0.0% n=0 · insufficient | +0.0u | n<5 |
[Auto] MLB Run Line (KNN) | MLB · Spread | 115 | 0-0 | 0.0% n=0 · insufficient | — | 0.0% n=0 · insufficient | +0.0u | n<5 |
[Auto] MLB Run Total (LightGBM) | MLB · Total | 115 | 0-0 | 0.0% n=0 · insufficient | — | 0.0% n=0 · insufficient | +0.0u | n<5 |
[Auto] MLB Moneyline Classifier (Random Forest) | MLB · Moneyline | 115 | 67-48 | 58.3% n=115 · strong [49.2%–67.3%] | +5.9pts | 11.2% n=115 · strong | +12.9u | +EV |
[Auto] MLB Moneyline (Ridge Classifier) | MLB · Moneyline | 115 | 66-49 | 57.4% n=115 · strong [48.4%–66.4%] | +5.0pts | 9.6% n=115 · strong | +11.0u | +EV |
[Auto] MLB Moneyline (ExtraTrees) | MLB · Moneyline | 115 | 65-50 | 56.5% n=115 · strong [47.5%–65.6%] | +4.1pts | 7.9% n=115 · strong | +9.1u | +lean |
[Auto] MLB Moneyline (TabPFN) | MLB · Moneyline | 114 | 64-50 | 56.1% n=114 · strong [47.0%–65.2%] | +3.8pts | 7.2% n=114 · strong | +8.2u | +lean |
[Auto] MLB Moneyline Predictor (XGBoost) | MLB · Moneyline | 115 | 64-51 | 55.7% n=115 · strong [46.6%–64.7%] | +3.3pts | 6.2% n=115 · strong | +7.2u | +lean |
[Auto] MLB Elo Power Ratings Moneyline (LogReg) | MLB · Moneyline | 115 | 61-54 | 53.0% n=115 · strong [43.9%–62.2%] | +0.7pts | 1.3% n=115 · strong | +1.4u | flat |
[Auto] MLB Moneyline SHAP Analysis (XGBoost) | MLB · Moneyline | 115 | 61-54 | 53.0% n=115 · strong [43.9%–62.2%] | +0.7pts | 1.3% n=115 · strong | +1.4u | flat |
Methodology: cover_rate = wins / (wins + losses) excluding pushes. edge_vs_break_even = cover_rate − 0.5238 (the 52.38% breakeven for a −110 standard payout). roi_per_pick = total_units / decisions, where each settled pick risks 1u and a win pays +0.909u (unless metadata.juice overrides). Materialized view refreshed via SELECT public.refresh_model_cover_rate_mv() — currently auto-refreshed nightly with the multi-sport grader.