Every winner call scored against what happened, event by event, with the method calls beside it. Only picks locked before their card segment started count; anything scored after the fact is marked BACKTEST and left out. The ledger only grows.
About this page+
Every winner call scored against what happened, event by event, with the method calls beside it. Only picks locked before their card segment started count; anything scored after the fact is marked BACKTEST and left out. The ledger only grows.
Winners calledOnly picks locked before the start count. A pick made later, or priced after the fact, never enters it.
Locked before the start, graded after
Our picks
69%
99–45
Betting favorite
69%
99–45
The same 144 priced games, winners only. Where a game is priced our pick is built on the market's price, so the two mostly match; we claim no edge. When we took the other side: 8–8.
Every graded pick, priced or not
Won–lost
103–48
Win rateWhere the true win rate likely sits (90% range). Under 30 picks, too early to read.
68%
90% range 62–74%
Graded
151
picks
Expected winsThe sum of its stated win probabilities. If those are right, wins land in this range 9 times in 10.
96.9
90% range 87.4–106.4
Win rate 68%90% rangeIts probabilities said 64%
Won 103. Its stated probabilities expected 96.9 (90% range 87.4–106.4): inside that range, in line with its probabilities.
By method: model alone 69–31 (100), market-anchored 34–17 (51).
Expected vs actual winnersThe dashed line is the wins its own probabilities expected, shaded for luck's 90% band. Solid is what happened.
151 picks
050100150
model alone ← | → market-anchored
Pick 1 · Jul 18picks in the order the games started →Pick 151 · Oct 24
Won 103Expected 96.990% band 87.4–106.4
Dotted line, pick 100: the method changed. Before it the calls were the model alone; after it a priced game’s number follows the market’s price, so the stretches are not the same product.
The solid line ends inside the shaded band: the record is where honest probabilities would put it.
Show as a table
After pick
Date
Won
Expected
90% band
19
Jul 25
16
12.5
9.2–15.9
38
Aug 1
31
24.8
20.1–29.5
57
Aug 29
43
36.4
30.5–42.3
76
Sep 12
56
48.4
41.6–55.1
94
Sep 26
66
60.0
52.5–67.5
113
Oct 3
81
73.1
65.0–81.3
132
Oct 17
92
85.2
76.3–94.1
151
Oct 24
103
96.9
87.4–106.4
When it said 70%, did it win 70%?Each lane is a confidence band. Calibrated means the dot lands in its gray band; the bar shows luck's range.
live
40%50%60%70%80%90%100%
Said 50–60%
65%won 40 of 6290% range 54–74%
Said 60–70%
66%won 29 of 4490% range 54–76%
Said 70–80%
71%won 24 of 3490% range 57–82%
Said 80–90%
91%won 10 of 1190% range 68–98%
The band it statedLive win rate90% range
Live picks on main UFC cards, locked before their card segment started. Lanes with fewer than 10 graded picks are faded.
Not counted above: 33 picks made after their card segment started are not counted; they appear per-event marked AFTER START. 25 graded bouts had no prediction locked before the fight — an early card we scored retroactively, plus bouts a data fault left unlocked. Rebuilt with the same model but reading profiles the card itself has since filled in, they carry hindsight the locked picks do not; they appear per-event marked BACKTEST. 43 graded picks are from development cards (Contender Series) — locked before the fights like any other, but most of those fighters have little or no tracked tape, so the picks run largely on career records and priors. Their cards are graded below, separately. Every headline number on this page is locked picks from main UFC cards only.
Against the betting favoriteSame games, winners only. The favorite is the side with the shorter no-vig moneyline when the pick was locked.
144 priced games
Our pickon the games that had a price99–4569%n 144
The betting favoritethe same games, no model99–4569%n 144
When we picked the favorite91–3771%n 128
When we took the other sidethe games we disagreed with the market8–850%n 16 · too early to read
Picks locked before Sept 28 are the model alone; since then a priced bout's win % is the market's no-vig price. 7 picks have no stored price and sit in none of these rows. Under 30 games a record is too early to read either way. Every pick, one row each →
Method callsHow each bout ends, graded against the probability the model gave its call. Not part of the winners record.
Winner + method
53–9835%n 151 · 29–42%
Distance call
98–5365%n 151 · 58–71%
Winner + method is a six-way call (either fighter by KO, submission or decision); always guessing the favorite by decision lands about 26%. The distance call is whether the bout reaches the judges.
UFC 333: Volkanovski vs. Evloev 2026-10-24Winners9/14Exact5/14Distance11/14
Dana White's Contender Series: Season 10, Week 10 2026-10-13Winners3/5Exact3/5Distance2/5
Development card — graded like any other, but excluded from the running totals above: most of these fighters have little or no tracked tape, so the model prices them mostly off career records and division priors.
Dana White's Contender Series: Season 10, Week 9 2026-10-06Winners3/5Exact2/5Distance4/5
Development card — graded like any other, but excluded from the running totals above: most of these fighters have little or no tracked tape, so the model prices them mostly off career records and division priors.
Dana White's Contender Series: Season 10, Week 8 2026-09-29Winners3/3Exact1/3Distance3/3
Development card — graded like any other, but excluded from the running totals above: most of these fighters have little or no tracked tape, so the model prices them mostly off career records and division priors.
Dana White's Contender Series: Season 10, Week 7 2026-09-22Winners4/5Exact0/5Distance2/5
Development card — graded like any other, but excluded from the running totals above: most of these fighters have little or no tracked tape, so the model prices them mostly off career records and division priors.
Dana White's Contender Series: Season 10, Week 6 2026-09-15Winners4/5Exact3/5Distance5/5
Development card — graded like any other, but excluded from the running totals above: most of these fighters have little or no tracked tape, so the model prices them mostly off career records and division priors.
Dana White's Contender Series: Season 10, Week 5 2026-09-08Winners5/5Exact4/5Distance3/5
Development card — graded like any other, but excluded from the running totals above: most of these fighters have little or no tracked tape, so the model prices them mostly off career records and division priors.
Dana White's Contender Series: Season 10, Week 4 2026-09-01Winners4/5Exact3/5Distance4/5
Development card — graded like any other, but excluded from the running totals above: most of these fighters have little or no tracked tape, so the model prices them mostly off career records and division priors.
Dana White's Contender Series: Season 10, Week 3 2026-08-25Winners4/5Exact2/5Distance5/5
Development card — graded like any other, but excluded from the running totals above: most of these fighters have little or no tracked tape, so the model prices them mostly off career records and division priors.
Dana White's Contender Series: Season 10, Week 2 2026-08-18Winners2/5Exact2/5Distance4/5
Development card — graded like any other, but excluded from the running totals above: most of these fighters have little or no tracked tape, so the model prices them mostly off career records and division priors.
Dana White's Contender Series: Season 10, Week 1 2026-08-11Winners0/0Exact0/0Distance0/0
Development card — graded like any other, but excluded from the running totals above: most of these fighters have little or no tracked tape, so the model prices them mostly off career records and division priors.
5 of these 5 bouts had no pick locked before the fight. They are marked BACKTEST below and are excluded from the totals above.
Jon Kunneman vs Joe KropschotBACKTESTNo pick was locked before this fight; rebuilt afterward, so it's left out of the record.
Conor McGregor vs Max HollowayMain eventBACKTESTNo pick was locked before this fight; rebuilt afterward, so it's left out of the record.
Model
Max Holloway60%-280
Max Holloway by KO/TKO (21%)-120
Result
Max Holloway
by KO/TKO · R1
Winner ✓Exact ✓Dist ✓
Scored automatically as each bout ends. “Exact” requires the single most likely outcome (winner + method) to hit — a 6-way call where always guessing favorite-by-decision runs ~26%. Method-of-victory prices shown beside the top outcome are sourced from published sportsbook props (our odds feed carries no MMA method markets); prices marked “R1 KO” or “UD” are the nearest documented market variant. Hold-out results, not a promise of future returns. Model is experimental. For information only, not betting advice. 21+ only. Gambling problem? Call or text 1-800-GAMBLER.
Picks grade as each bout ends. For information only, not betting advice. 21+ only. Gambling problem? Call or text 1-800-GAMBLER.