A Significant Strikes Prop Model With Stylistic Adjustments

Significant-strikes props are the easiest UFC market to model and the hardest to bet. The market mostly prices each fighter's career SLpM × 15 minutes, with a small adjustment for opponent. That works for typical matchups and breaks badly for stylistic mismatches. This post walks through a closed-form prop model that beats the market by 4–7 percent on stylistic-mismatch fights, what features to feed it, and where to slot it on Tinker. The full Monte Carlo version lives in the UFC round-simulation brick; this is the single-prop, faster, closed-form companion.

Why significant strikes matter as a prop

UFC sportsbooks price significant-strikes-landed props per fighter, usually with thresholds in 5- or 10-strike increments. Typical lines:

• Over/under 65.5 for a 5-round main event between strikers.
• Over/under 45.5 for a 3-round fight with a takedown threat.
• Over/under 30.5 for a heavyweight 3-rounder where the finish probability is high.

Vig is typically -115 on each side, occasionally -120 on the favored side. Edges in this market come from the books mis-pricing the time distribution (standing vs ground) rather than the per-minute output rates. If you can project how many minutes of standing time a fight has, the strike count almost falls out.

The closed-form model

The model has three layers: a per-fighter output rate, a time-distribution model, and a finish-truncation adjustment. None of them require simulation; all three can run in under 5ms in the browser, which makes this brick the right one to use for fast in-play retunes when an injury or stance change is announced just before bell time.

Layer 1: opponent-adjusted strike rate

For each fighter, compute:

• SLpM: significant strikes landed per minute, career.
• Opponent-adjusted SLpM: SLpM × (1 + α × (current opponent SApM − league average SApM)), where α is a fit coefficient around 0.4.
• Rolling SLpM: last-5-fights weighted at 0.5, career at 0.5.

A fighter who averaged 5.5 SLpM against a typical opponent (SApM 4.0) will project to 4.8 SLpM against an elite defender (SApM 2.5). The market often misses this — it cites SLpM as if all opponents were average, especially for fighters with résumés padded against weak defense.

Layer 2: time distribution

The fight time gets split into:

• Standing distance: prior 60% in striker-vs-striker, 20% in wrestler-vs-striker (favoring the wrestler).
• Clinch: prior 10–25% depending on both fighters' clinch tendencies.
• Ground: prior depends on TDpM × (1 − opponent TDD) for the takedown-favored fighter.

The model computes an expected standing time per fighter, then applies the SLpM rate only to standing time and a much smaller residual rate (about 0.4 SLpM equivalent) to ground time. For a wrestler vs striker matchup, this is where the projection diverges from the market's naive SLpM × 15.

Layer 3: finish truncation

If the fight is expected to finish in round 1 with probability 25%, the expected strikes count has to discount the rounds that won't happen. We use a competing-risks survival model: per minute of standing time, hazard of KO is roughly KO/TKO-win-rate × 0.04 per minute. The expected strikes is integrated over the survival curve, not over the full scheduled time.

For most fights this adjusts the projection by 4–10 strikes. For high-finish-rate heavyweight bouts the adjustment can hit 20+ strikes — which is why every heavyweight prop is an under-favoring market.

Stylistic adjustments that actually move the line

The adjustments worth coding into the brick, in order of impact:

1. Wrestler with high TDpM vs low-TDD opponent: reduces opponent's standing time by 30–55%. Single biggest factor. The market often prices this at 60% of fair value, leaving room for under bets on the striker's strikes prop.
2. Pressure fighter vs counter-striker: counter-striker absorbs more clean shots, total strikes-thrown rises 8–15% above naive average.
3. Opposite stance (orthodox vs southpaw): small +3 to +5% total strikes thrown across both fighters; doesn't usually flip a line.
4. Cage size: small cage (UFC Apex) clusters fighters and increases strikes thrown by 5–8% vs a full Octagon. Mostly relevant for Fight Nights at Apex.
5. 5-round fight: rounds 4 and 5 have lower per-minute output (~85% of the round 1–3 rate) due to fatigue. Account for this in the time integral.
6. Late notice / weight miss: 10–15% output reduction for the affected fighter, and slight reduction (~5%) for the opponent due to fight-flow disruption.
7. Reach differential: small effect on accuracy but not on strikes thrown. Don't double-count.
8. Female bantamweight/featherweight: historically higher per-minute output than equivalent men's weight classes. Use sex-specific priors, not unified priors.

Worked example: striker vs wrestler

3-round bantamweight bout. Fighter A is a striker (SLpM 5.4, SApM 3.1, TDD 60%). Fighter B is a wrestler (SLpM 3.2, SApM 2.6, TDpM 3.8, KO/TKO 18% of wins).

• Expected standing time per round: 3.1 minutes (out of 5).
• Expected total standing time: 9.3 minutes if fight goes the distance.
• Distance probability: 71% (low finish rate from B, average from A).
• Fighter A expected significant strikes: 5.4 × 9.3 × opponent adjustment (0.96 for B's slightly above-average defense) × distance prob (0.71) + truncated path strikes (0.29 × ~28) = 42.4 strikes.
• Fighter B expected significant strikes: 3.2 × 9.3 × A's defense (1.04, A absorbs more than average) × distance prob (0.71) + truncated path = 25.7 strikes.

Sportsbook line on A: over 51.5 (-115). Model projection: 42.4. Edge: under 51.5, fair price about -180, market is -115 — strong under bet.

Sportsbook line on B: over 22.5 (-110). Model projection: 25.7. Edge: over 22.5, fair price about -155, market is -110 — over bet.

Both sides edges came from the time-distribution model recognizing that B will get 6+ minutes of ground control over the fight, which the public sportsbook line did not fully price in.

Where this model breaks

1. Late stylistic shift. A fighter who has changed coaches and now strikes more is hard for any rolling average to catch. Watch for camp changes and apply a manual override.
2. Reach mismatches at heavyweight. Reach matters more at heavyweight because every strike does damage. The closed-form model under-weights this; switch to the simulation brick for HW main events.
3. Fights with ringside doctor stoppages. The KO hazard model assumes natural finishes; doctor stoppages from cuts are noisier. Apply a small downward adjustment for fighters with a recent cut history.
4. Newcomers with under 3 UFC fights. Regional MMA data is noisy; project conservatively and prefer joint Kelly fractions under 0.25 on first-three-fight UFC newcomers.

Slotting the brick into your workflow

The closed-form brick lives next to the simulation brick on Tinker. Recommended workflow:

1. Run the closed-form brick first — it's fast and gives you a single-prop projection for every fighter on the card.
2. Flag fights where the model edge is more than 6 strikes vs the market line.
3. Run the simulation brick on flagged fights to confirm the joint distribution agrees with the closed-form projection.
4. Bet the agreed edges at appropriate Kelly fractions.
5. Log every bet, with stake, opening line, closing line, and outcome — feeds CLV tracking from the CLV explainer.

The closed-form model is intentionally less ambitious than the sim. It does one thing well: project significant strikes for a single fighter against a specific opponent, fast. For everything else, escalate to the sim.

Validating the model

Backtest on the last two seasons of UFC cards. Walk-forward by event. Expected results from a well-calibrated version:

• Mean absolute error on significant strikes landed: ~9 strikes (vs ~13 for a naive SLpM × 15 baseline).
• Under bets in striker-vs-wrestler bouts: ~58% hit rate at -110 (break-even 52.4%).
• Over bets in striker-vs-striker Apex events: ~56% hit rate.
• Method-implied props (no direct strikes): break-even or worse — don't bet these from this model.

Pipe results into picks for tracking and tune the style matrix every quarter.

Weight class differences worth knowing

The brick uses unified priors across weight classes by default, but several adjustments are worth wiring in if you bet across the card:

• Strawweight and flyweight (women's and men's): high per-minute strike output (SLpM averages 4.8 across the division). Low knockout rate. Decisions dominate. Total strikes props skew over.
• Bantamweight and featherweight: middle ground; balanced finish rates. Most bettable closed-form predictions.
• Lightweight: highest grappling rate in modern UFC; wrestler-vs-striker matchups dominate. Time distribution model matters more than per-minute output.
• Welterweight and middleweight: physically larger, more finish-prone. Truncation adjustment matters more.
• Light heavyweight and heavyweight: high finish rate, low strike volume even in standing time, short fights. Closed-form model under-predicts variance; prefer the simulation for these.
• Women's featherweight and women's bantamweight: small sample of fights at the elite level; priors are unstable. Use career averages with heavy shrinkage toward division means.

Cage size and venue effects

UFC has two cage sizes that get used:

• Full Octagon (Vegas T-Mobile, most international PPVs): 30-foot cage. More room to circle; striking output slightly lower.
• Apex small cage: 25-foot. Fighters get cornered faster. Strike output runs 5–8% higher on average; takedown completion goes up too (less escape room).

Tag the venue in your model and apply the +5–8% multiplier on Apex Fight Nights. The market sometimes prices it, sometimes doesn't. When it doesn't, overs on standing-strikes props at Apex are a soft edge.

Altitude and travel effects

UFC events at altitude (Denver, Mexico City) measurably reduce per-round output, especially in round 3 and beyond. The drop is small (~3% lower SLpM) but consistent. Same logic for fighters with long travel and short acclimation windows. Most public models ignore these; the brick exposes them as togglable features.

Tracking model performance per fighter

One pattern that emerges from running this brick across hundreds of fights: certain fighter styles consistently beat or miss model expectations:

• Pressure southpaws (e.g., Conor-style entries): model under-predicts standing time; they keep the fight where they want.
• Slow-starting wrestlers: model overestimates round 1 takedown attempts; they often feel out the striker before shooting.
• Faded veterans: model overweights rolling-5 features when career features are more predictive of late-career output.
• Hot prospects 2–4 fights into UFC: model underweights ceiling; small UFC sample plus regional dominance suggests they're better than priors imply.

Keep a notebook of fighter-specific overrides as you run the model. Don't add these to the main brick — they're not generalizable — but use them when you bet that specific fighter.

Putting it together: a card workflow

The full pre-card workflow:

1. Load the card on Tinker. Brick auto-fetches UFC Stats for every listed fighter.
2. Run the closed-form sig-strikes projection for every fighter on the card.
3. Pull sportsbook lines (DraftKings, FanDuel, BetMGM) for sig-strikes props on each fighter.
4. Compute edge per prop. Flag edges >5 strikes.
5. For flagged fights, run the full simulation brick to confirm joint pricing.
6. Apply weight-class, venue, and stance overrides where appropriate.
7. Size bets using Kelly fractions on the calibrated probabilities.
8. Log to picks and track CLV per fighter, per weight class, per venue.

Bottom line

UFC significant-strikes props sit at the intersection of slow market reaction and obvious stylistic factors. A closed-form model with opponent-adjusted SLpM, time-distribution decomposition, and finish-truncation is enough to find edges of 4–7 percent on the median card. Build it on Tinker, validate on two seasons of UFC Stats data, and bet the unders on strikers facing real wrestlers.

Bet responsibly — set limits, never chase losses.

UFC example board

A fight-night model should describe paths, not just winners. Islam Makhachev is the clean grappling-control example: takedown success, top time, and submission threat can all point in the same direction. Alex Pereira is the striking-power example where knockout equity can outrun minute-by-minute control metrics. Ilia Topuria and Tom Aspinall are useful pressure and early-finish examples because the model must respect both ceiling and fragility.

• Grappler path: takedown entries, control time, opponent get-up rate, and submission exposure.
• Striker path: knockdown power, defensive responsibility, five-round cardio, and judging volatility.
• Market trigger: weigh-in misses, short-notice replacements, and camp-change rumors deserve their own line-move tag.
• Prop filter: finish props need a different threshold from moneyline bets because time and method matter.

UFC update rules

Do not freeze a UFC take before weigh-ins. A bad cut, opponent change, missed limit, or short-notice travel spot can rewrite both the win probability and the finish probability. Keep the modeling workflow tied to fight-night modeling and review the price movement through CLV after the close.

Research note board

Use this board before clicking a prop, DFS build, or same-game entry. The table is intentionally about thresholds, not fake certainty.

Prop OVER hit rate vs line distance from median

Empirical hit rate of OVER bets as the prop line moves away from the player projection median, measured in standard deviations. A line set 1sd below the median hits ~84% of the time — but books price the juice to match.

Breakeven win % at common American odds

The win rate you need to break even at each price. Pick odds shorter than -150 and you must win >60% just to stay flat — a hurdle most casual handicappers never sustain.