Experimental Two public models, graded only on fights after their training cutoffs. Betting results are simulated point-in-time — the live record started July 2026 and is 0-1 so far. Not picks, not advice.
The Model Lab

How CFL's two public models work

The Model Lab — how CFL's two public models work, and the honest windows they're graded on. Value prices bets against the opening line. Fight IQ picks winners from tape alone and never sees a line. Every graded number on this page counts only fights after each model's training cutoff — fights the model never saw — and ROI is computed at the opening line, the realistic price a bettor could actually get. Earlier models (v1, v2, v4, v5) are retired; v5 is retired because it was trained on the closing line, which made its accuracy the market's own number.

Admin view Indicator chips flag historical patterns on dog picks. Hidden from non-admin users.
PREMIUM edge≥0.20 · younger · coming off W
CLV AGREE closer moved ≥+2% prob toward us
HIGH CONVICTION edge ≥ 0.20
YOUNG DOG edge≥0.20 · younger
BALANCED edge≥0.15 · ≤+250 · younger
STANCE FADE orthodox dog vs southpaw favorite
AVOID — SUB LOSS dog coming off a submission loss
CLV FADE closer moved ≥−2% prob (line drifting against us)
Upcoming picks · with bet sizing
How to use this page

What this is

Each line below is a fight on an upcoming UFC card. For every fight, each of CFL's two public models gives its own predicted win probability for each fighter, and we compare that to what the betting market is offering. When the model thinks a fighter is meaningfully more likely to win than the market price implies, the row gets flagged ★ — that's a candidate bet.

How to read a row

Model
Which model made the pick. Value (v6) prices bets against the opener. Fight IQ (v3) picks winners from tape alone and never sees a betting line.
Pick
The fighter the model favors more than the market does.
Opener
The American odds that fighter opened at (BFO consensus across multiple sportsbooks).
Op p
What the opener implies about that fighter's win probability.
Model p
What our model predicts the fighter's win probability actually is.
Edge
(Model p − Opener p). The gap. Bigger gap = better bet, in theory.
Stake
How much we recommend you bet, given your bankroll. See "Bet sizing" below.
The model flags this row as a bet (edge ≥ 5%).

3-step quick start

  1. Set your bankroll. Type your total betting capital into the Bankroll box. Stake amounts scale automatically.
  2. Check "Consensus only." A consensus pick is when both models independently flag the same fighter. The page hides everything else — these are the highest-confidence bets of the card.
  3. Bet the recommended Stake on the recommended Pick. We recommend Polymarket for execution (no sportsbook vig, no banning winners).

Bet sizing — the "Kelly" thing

Kelly is a formula that tells you the bet size that mathematically maximizes long-term growth, given your model's win probability and the odds offered. Bigger edges = bigger stakes. We default to Half Kelly because Full Kelly has brutal short-term variance — half-Kelly captures ~75% of the long-run growth with ~25% of the swings.

Max bet cap (default 5%) is a safety net against any single bet wrecking the bankroll if the model is wrong about one fighter. Even at 0% Kelly value, you'll never be exposed beyond this cap on one bet.

Edge thresholds — when to bet

We use edge ≥ 5% as the default "would_bet" threshold. Lower thresholds catch more bets but include more marginal ones; higher thresholds (7–10%) are more selective with better per-bet ROI but fewer opportunities. The model card stats above show how each threshold performed historically.

Honest caveats. This is an experimental research tool, not a tout service. Value's betting record is simulated point-in-time, not live: on fights after its December 2024 training cutoff, the shipped rule (favorites with 3+ points of edge, $100 flat bets) has gone 66–21 as of July 15, 2026 — correct on 66 of 87 bets — for +$1,478 at the opening price (+17.0%). Re-priced at closing odds, that drops to +$1,103 (+12.7%). The live record starts July 2026 and is 0-1 so far — the first settled live bet lost. Real-world results also shrink with liquidity friction and book limits, and a losing month or two is normal — not a sign the strategy is broken. Never bet money you can't afford to see swing 30–40% down before recovering. Always gamble responsibly. 1-800-GAMBLER.
How the models compare

Each model is graded only on fights after its own training cutoff — fights it never saw. Click a model for the full breakdown and recent graded picks.