Premium MMA Intelligence

We publish the numbers.
Including the bad ones.

Calibrated UFC fight predictions — every engine measured against the closing line, every miss shown next to every hit. No accuracy theatre. Just whether a stated 70% actually happens 70% of the time.

Enter the Prediction Lab →See the method

01 / Events

The Calendar

Every card graded before the walkouts, settled the morning after. The next one drops here the moment it's announced.

Next Event

UFC Fight Night: Ankalaev vs. Guskov

Every engine's read locked, confidence stamped, and the calibration table updated — with the results scored honestly the morning after.

2026-07-25 · Etihad Arena · Abu Dhabi, UAECalibrated · not hyped

See the predictions

2026-06-14View card

UFC Freedom 250: Topuria vs Gaethje

The White House (South Lawn) · Washington

2026-05-09View card

UFC 328: Chimaev vs Strickland

Prudential Center · Newark

2026-05-02View card

UFC Fight Night: Della Maddalena vs. Prates

RAC Arena · Perth

All events →

02 / Prediction Lab

The Honest Scoreboard

UFC betting markets are close to efficient — the closing line wins about two-thirds of the time. So we measure calibration, not bravado, and beating the market is the only result that would mean anything.

Market benchmark

67.3%

Brier 0.2075 · n=3,257

Genuine locked picks

537

314 correct · 58.5%

Best in-house · by calibration

STYLE

Brier 0.2418 · 61.5% acc

Engines beating the market

measured on Brier score

Soon AI Fight Read

Read any matchup.

Name two fighters and the reader returns a calibrated lean, a confidence band, and the one stylistic factor that tips it — in UNFOLD's honest voice. Qualitative only: no invented records, no odds, no bets.

▸ Enter two fighters and hit “Read the matchup.” The live reader leans — then tells you why it might be wrong. Launching soon.

Calibration curve

MARKET (benchmark) n=3,257STYLE (best in-house) n=30perfect calibration

Does a stated confidence mean anything?

Each point is a confidence bucket. The dashed diagonal is perfect calibration — above the line means we were too cautious, below means too sure of ourselves. The market sits almost exactly on it. That is the standard our engines are measured against.

Distance from the diagonal is the error. An engine can pick winners often and still be badly calibrated — which is why accuracy alone is a poor way to choose between models.

Engine Scoreboard

Ranked by Brier · lower is better

UFC prediction engines ranked by Brier score, with win–loss, accuracy, log loss, ECE and sample size.
Engine	W–L	Acc	Brier	Log loss	ECE	n
MARKETbenchmark	2192–1065	67.3%	0.2075	0.6019	0.0195	3,257
BOOKWORM	33–23	58.9%	0.2413	0.7329	0.1965	13
STYLE	56–35	61.5%	0.2418	0.6766	0.0926	30
ORACLE	57–34	62.6%	0.2540	0.7141	0.2298	30
MOMENTUM	45–46	49.5%	0.2550	0.7001	0.0453	30
ML	51–40	56.0%	0.2579	0.7459	0.1675	30
MASTER	53–38	58.2%	0.2620	1.0332	0.1443	43
ELO	49–42	53.8%	0.2705	0.8204	0.1204	30
TEST	51–40	56.0%	0.2717	0.7998	0.2086	30
DATA	49–42	53.8%	0.2781	0.7535	0.1283	30
BASELINE	48–43	52.8%	0.2910	0.8060	0.1457	30

MARKET is the de-vigged closing line, not a model of ours — beating it, not beating a coin flip, is the only result that would mean anything. BOOKWORM has fewer than 25 scored picks, so its Brier is not yet meaningful. Accuracy figures exclude 260 picks from two 2026 events reconstructed after the fact; they remain in the record but are not counted as predictions. Analysis and entertainment only. Not betting advice.

03 / The Fighters

The Roster

Style, division and country. Drop each fighter's current photo into their card.

Lightweight

Ilia Topuria

Pressure Striker · Georgia / Spain

Welterweight

Islam Makhachev

Sambo Grappler · Dagestan

Light Heavyweight

Alex Pereira

Kickboxer · Brazil

Bantamweight

Merab Dvalishvili

Cardio Wrestler · Georgia

Strawweight

Zhang Weili

Complete Fighter · China

Middleweight

Sean Strickland

Volume Pressure · USA

04 / Analysis

The Method

How the scoreboard is built and graded — analytical and honest, never hot-take culture.

Accuracy is a vanity metric

An engine can pick winners and still be badly wrong about how sure it should be. That's why every engine is ranked by Brier score — being right and being appropriately unsure — not by win rate.

The Method · see the scoreboard ↓

The Benchmark

What the closing line knows

The de-vigged market hits about two-thirds. Beating it — not a coin flip — is the only result that counts.

The Record

How we grade every card

Locked before bell time, scored the morning after — misses shown next to hits, reconstructions flagged.

05 / The Gallery

Fight Week, Rendered

Cinematic poster art, faceoffs and aura pieces. Drop finished renders into each tile.

Main Event · Key Art

Faceoff

Aura

Weigh-in

The Arena

06 / Follow

Never miss a lock.

Every engine's read, the calibration table, and the misses — no spam, no picks-for-profit hype.

Newsletter launching soon. Until it's live, follow along on social for every lock and every miss.

Engine

W–L

Acc

Brier

Log loss

ECE

MARKETbenchmark

2192–1065

67.3%

0.2075

0.6019

0.0195

3,257

BOOKWORM

33–23

58.9%

0.2413

0.7329

0.1965

STYLE

56–35

61.5%

0.2418

0.6766

0.0926

ORACLE

57–34

62.6%

0.2540

0.7141

0.2298

MOMENTUM

45–46

49.5%

0.2550

0.7001

0.0453

51–40

56.0%

0.2579

0.7459

0.1675

MASTER

53–38

58.2%

0.2620

1.0332

0.1443

ELO

49–42

53.8%

0.2705

0.8204

0.1204

TEST

51–40

56.0%

0.2717

0.7998

0.2086

DATA

49–42

53.8%

0.2781

0.7535

0.1283

BASELINE

48–43

52.8%

0.2910

0.8060

0.1457

We publish the numbers.Including the bad ones.