Ankalaev Finishes It in Five, and the Card Did the Rest

Magomed Ankalaev needed five rounds to put Bogdan Guskov away, but he got there — punches, Round 5, 2:41 on the clock. That is the main event in full: the light heavyweight headliner went the distance of a championship structure and then ended emphatically anyway, Ankalaev doing exactly what a 61% model favorite and an 81% market favorite is supposed to do. Guskov had his moments — 107.2 DraftKings points for Ankalaev tells you the fight had real content — but in the end it was over by punches in the fifth.

The rest of the card moved faster and louder. Valter Walker submitted Thomas Petersen via calf slicer at 1:32 of Round 1 in the heavyweight co-main — a finish unusual enough that it is worth saying out loud: a calf slicer ended a UFC heavyweight bout in under two minutes. The model had Walker at 58% and the market at 62%; both sides liked him, and both sides were right inside 90 seconds.

Muhammad Said knocked out Dustin Jacoby with a kick in Round 2 (4:49) — the card's first real upset by the numbers. Both the model (59%) and the market (61%) had Jacoby. Said was a low-data debutant the board was blind on by design, and he made the board pay for it.

Ramazan Temirov finished Steve Erceg by punches in Round 1 at 4:21, a coin-flip bout — model and market both at 50% — that resolved as violently as a coin-flip can. Sam Patterson then stopped Santiago Ponzinibbio in Round 2 by punches at 3:06, a result the market saw coming (78% Patterson) but the model did not (53% Ponzinibbio). More on that in the receipts.

Magomed Zaynukov took a unanimous 30-27 decision over Damian Rzepecki in three. Axel Sola choked out Ismael Bonfim with a D'Arce at 4:44 of Round 1. Magomed Tuchalov edged Brendson Ribeiro by 29-28 on all three cards. Rizvan Kuniev finished Tyrell Fortune by knee in Round 3 at 1:12. Abubakar Vagaev outpointed Saygid Izagakhmaev 29-28 across the board. Nurullo Aliev swept Mike Davis 30-27 on all three scorecards. And Abdul Hussein submitted Cody Gibson with an arm triangle in Round 3 at 3:07.

Two bouts — Wellington Turman vs. Islam Dulatov and David Martinez vs. Umar Nurmagomedov — are not in the results store; they are listed as cancelled or unreported and are not graded here.

Of the 12 bouts with results, 12 produced a winner. Eight of those ended by finish — five by KO/TKO, three by submission. This card moved.

A reminder on framing before the numbers: this is the UFC model at version zero. The picks gate is closed — no historical odds validation has been done. What follows is a straight accuracy receipt on the journaled published probabilities. Nothing on this card was a pick. This is the board graded honestly, nothing more.

Across 12 gradable bouts, the model went 66.7% accurate with a log-loss of 0.6645 and a Brier score of 0.2358. The market, over the same 12 bouts, went 83.3% — log-loss 0.4024, Brier 0.1170. The market was sharper. That is on the record.

Two divergence flags came in at eight percentage points or more. Both were model-side losses, both worth naming:

• Ankalaev vs. Guskov: The model had Guskov at 39%, the market had Ankalaev at 81% — a 20-point divergence. The market's side won. The model was leaning toward Guskov; the market was leaning hard the other way. The market was right.
• Patterson vs. Ponzinibbio: The model had Ponzinibbio at 53%, the market had Patterson at 78% — a 31.6-point divergence, the larger of the two flags. The model's side lost. The market read this fight as a Patterson fight from the jump; the model called it a mild Ponzinibbio lean. Patterson won by punches in Round 2.

The model went 0-for-2 on its divergence flags. That is not a betting record — it is a calibration note. When the board was meaningfully out of step with the market on this card, the market was right both times. Nine of the 12 graded bouts were low-data bouts where the model deferred to the market by design; grade those accordingly.

Of the 12 bouts with results: five ended by KO/TKO, three by submission, four by decision. Eight finishes out of 12 decisive results. The card did not drag.

The five KO/TKO stoppages: Ankalaev over Guskov by punches in Round 5, Said over Jacoby by kick in Round 2, Temirov over Erceg by punches in Round 1, Patterson over Ponzinibbio by punches in Round 2, and Kuniev over Fortune by knee in Round 3. The three submissions: Walker over Petersen via calf slicer in Round 1, Sola over Bonfim via D'Arce choke in Round 1, and Hussein over Gibson via arm triangle in Round 3. Four decisions rounded it out.

On the sim layer: the board journaled DraftKings point bands (floor at the 10th percentile, ceiling at the 90th) for fighters pre-event. Of the 24 banded fighters, 15 landed inside their band — 62.5% coverage. The mean projection across banded fighters was 47 DK points; the mean actual was 59.9, a bias of -12.9 points. The sim was systematically low. A card that produced eight finishes and several fast, dominant performances will do that.

One important note, stated plainly: per-bout finish-method probabilities were never published on the board. The sim journals DK-point bands only. The realized methods above are results — they are not graded against a number that does not exist.

The top of the DraftKings leaderboard was dominated by finishers who went out fast and piled up activity points. In order:

1. Axel Sola — 125.7 DK points. Pre-event projection: 55.7, band: 13.3–118.1. He blew past his ceiling by 7.6 points. A first-round D'Arce choke finish will do that.
2. Ramazan Temirov — 122.0 DK points. Pre-event projection: 46.1, band: 13.5–118.5. Cleared the ceiling by 3.5. First-round TKO in a coin-flip bout.
3. Nurullo Aliev — 116.2 DK points. Pre-event projection: 50.5, band: 12.0–115.5. Cleared the ceiling by 0.7. A 30-27 sweep is thorough work.
4. Sam Patterson — 116.0 DK points. Pre-event projection: 41.8, band: 5.7–102.9. Cleared his ceiling by 13.1 points — the biggest ceiling bust on the card.
5. Abdul Hussein — 115.2 DK points. Pre-event projection: 66.1, band: 13.8–119.2. The one top scorer who landed inside his band — covered. The arm triangle finish in Round 3 came in exactly where the sim expected a winner to land.
6. Muhammad Said — 114.1 DK points. Pre-event projection: 43.4, band: 12.5–116.6. Inside the band. A low-data debutant who kicked Jacoby out in Round 2 and scored accordingly.

Four of the six top scorers broke their ceilings. The sim projection was low on this card, and the top of the scoring sheet reflects it. If you rostered the right finishers on Friday, you were well above the mean.

No soft landings here. The board had two divergence flags and went 0-for-2 on them. It had a systematic projection bias of -12.9 DK points. Nine band busts across 24 banded fighters. Let's name them all.

Divergence flag losses:

• Ponzinibbio-Patterson: The model liked Ponzinibbio at 53%. The market liked Patterson at 78%. Gap of 31.6 points — the wider flag on the card. Patterson won by punches in Round 2 and scored 116 DK points. The model was on the wrong side by a lot.
• Ankalaev-Guskov: The model had Ankalaev at 61% — a lean, not a conviction. The market had Ankalaev at 81%. Model had Guskov at 39%; market had Guskov at 19%. Ankalaev won. The result went the model's way directionally, but the board was underconfident relative to where the market sat. That 20-point divergence still counts as a flag, and the model's side — Guskov — lost.

Biggest ceiling bust — Sam Patterson: Ceiling was 102.9. Actual: 116.0. Over by 13.1 points. The sim simply did not project a second-round TKO against Ponzinibbio; when that's how it ends, the sim misses high on the winner.

Biggest floor bust — Steve Erceg: Floor was 11.0. Actual: 4.6. Under by 6.4 points. A first-round TKO loss leaves almost nothing on the board, and Erceg's 4.6 points came in well below where the sim's pessimistic tail landed.

Additional band busts: Thomas Petersen (actual 2.3, floor 7.2, under by 4.9); Ramazan Temirov (actual 122.0, ceiling 118.5, over by 3.5); Axel Sola (actual 125.7, ceiling 118.1, over by 7.6); Santiago Ponzinibbio (actual 5.6, floor 8.5, under by 2.9); Ismael Bonfim (actual 9.1, floor 11.2, under by 2.1); Tyrell Fortune (actual 9.8, floor 12.9, under by 3.1); Nurullo Aliev (actual 116.2, ceiling 115.5, over by 0.7).

The pattern is clear: fast finishes shattered the floors for losers and blew the ceilings for winners. A card this active was always going to stress the bands.

Three things this card put on record:

One. Ankalaev is the real thing at light heavyweight. He closed out a five-round main event by punches against a credible opponent. The market read him at 81% and the market was right. The model's 61% was a lean; 81% was a conviction. When the gap is that wide and the market is that sure, that gap is information.

Two. The v0 model went 66.7% accurate against the market's 83.3% on a card loaded with low-data, debutant bouts where the board deferred to the market by design. Nine of 12 graded bouts were flagged low-data. That structural limitation explains some of the gap — but not all of it. The two divergence flags, both losses, are a genuine calibration note: when this board and the market disagree by double digits, the market has earned the benefit of the doubt until the model builds a deeper track record.

Three. Eight finishes on a 12-bout card will break your sim projections. The mean actual (59.9 DK points) came in 12.9 points above the mean projection. That is not random error — that is a finishing-rate that ran hotter than the seeded sim expected. The DFS lesson is the same one MMA always teaches: roster the right finisher in the right spot and everything else is detail.