UFC
Noche UFC: Silva vs. Delgadotimes in ETBouts
13 scheduled · 0 with a report · 11 with a supported read
- Sat, Sep 12 · 2:00 PM ETscheduledTim Elliott vs Édgar CháirezÉdgar Cháirez · 78%model forecast · DEC most likelycard-level report
- Sat, Sep 12 · 2:00 PM ETscheduledMuslim Salikhov vs Ignacio BahamondesIgnacio Bahamondes · 79%model forecast · DEC most likelycard-level report
- Sat, Sep 12 · 2:00 PM ETscheduledYousri Belgaroui vs Djorden SantosYousri Belgaroui · 64%model forecast · KO most likelycard-level report
- Sat, Sep 12 · 2:00 PM ETscheduledDrakkar Klose vs Tommy GanttTommy Gantt · 52%model forecast · DEC most likelycard-level report
- Sat, Sep 12 · 2:00 PM ETscheduledRafa Garcia vs RongzhuRongzhu · 52%model forecast · DEC most likelycard-level report
- Sat, Sep 12 · 2:00 PM ETscheduledJessie Rosas vs Sean KingNo supported readcard-level report
- Sat, Sep 12 · 2:00 PM ETscheduledJJ Aldrich vs Regina TarinRegina Tarin · 65%model forecast · DEC most likelycard-level report
- Sat, Sep 12 · 5:00 PM ETscheduledJean Silva vs Jose Miguel DelgadoJean Silva · 68%model forecast · KO most likelycard-level report
- Sat, Sep 12 · 5:00 PM ETscheduledBrandon Moreno vs Joseph MoralesBrandon Moreno · 59%model forecast · DEC most likelycard-level report
- Sat, Sep 12 · 5:00 PM ETscheduledTommy McMillen vs Marwan RahikiNo supported readcard-level report
- Sat, Sep 12 · 5:00 PM ETscheduledAlexa Grasso vs Manon FiorotManon Fiorot · 57%model forecast · DEC most likelycard-level report
- Sat, Sep 12 · 5:00 PM ETscheduledCurtis Blaydes vs Waldo Cortes AcostaWaldo Cortes Acosta · 59%model forecast · KO most likelycard-level report
- Sat, Sep 12 · 5:00 PM ETscheduledDan Ige vs David MartinezDavid Martinez · 65%model forecast · DEC most likelycard-level report
| Start | Bout | Status | Our read | Report |
|---|---|---|---|---|
| Sat, Sep 12 · 2:00 PM ET | Tim Elliott vs Édgar Cháirez | scheduled | Édgar Cháirez · 78%model forecast · DEC most likely | card-level report |
| Sat, Sep 12 · 2:00 PM ET | Muslim Salikhov vs Ignacio Bahamondes | scheduled | Ignacio Bahamondes · 79%model forecast · DEC most likely | card-level report |
| Sat, Sep 12 · 2:00 PM ET | Yousri Belgaroui vs Djorden Santos | scheduled | Yousri Belgaroui · 64%model forecast · KO most likely | card-level report |
| Sat, Sep 12 · 2:00 PM ET | Drakkar Klose vs Tommy Gantt | scheduled | Tommy Gantt · 52%model forecast · DEC most likely | card-level report |
| Sat, Sep 12 · 2:00 PM ET | Rafa Garcia vs Rongzhu | scheduled | Rongzhu · 52%model forecast · DEC most likely | card-level report |
| Sat, Sep 12 · 2:00 PM ET | Jessie Rosas vs Sean King | scheduled | No supported read | card-level report |
| Sat, Sep 12 · 2:00 PM ET | JJ Aldrich vs Regina Tarin | scheduled | Regina Tarin · 65%model forecast · DEC most likely | card-level report |
| Sat, Sep 12 · 5:00 PM ET | Jean Silva vs Jose Miguel Delgado | scheduled | Jean Silva · 68%model forecast · KO most likely | card-level report |
| Sat, Sep 12 · 5:00 PM ET | Brandon Moreno vs Joseph Morales | scheduled | Brandon Moreno · 59%model forecast · DEC most likely | card-level report |
| Sat, Sep 12 · 5:00 PM ET | Tommy McMillen vs Marwan Rahiki | scheduled | No supported read | card-level report |
| Sat, Sep 12 · 5:00 PM ET | Alexa Grasso vs Manon Fiorot | scheduled | Manon Fiorot · 57%model forecast · DEC most likely | card-level report |
| Sat, Sep 12 · 5:00 PM ET | Curtis Blaydes vs Waldo Cortes Acosta | scheduled | Waldo Cortes Acosta · 59%model forecast · KO most likely | card-level report |
| Sat, Sep 12 · 5:00 PM ET | Dan Ige vs David Martinez | scheduled | David Martinez · 65%model forecast · DEC most likely | card-level report |
UFC
Predictions · experimentalWinner, method and finishing round for every bout on the next card. Paper and educational.
What changed, and what is still refused
Next card
Card and model read 2026-09-06T16:03:21Z
Sooner, but not read · Dana White's Contender Series: Season 10, Week 5 is on September 8. We have UFC history for both fighters in 0 of its 5 bouts, so the model has nothing to say about it and we publish no read. The card below is the next one it can read.
Noche UFC: Silva vs. Delgado
Sat, Sep 12, 2:00 PM ET · Desert Diamond Arena · 13 bouts
Jean Silva
Jose Miguel Delgado12-2-04-1 in his last 5. 83% of his wins come by finish; 14% of his fights reach the judges.
Brandon Moreno
Joseph MoralesBrandon Moreno 18 tracked bouts against 5 — and the matchup profiles as one that reaches the judges.
2-3 in his last 5. 55% of his wins come by finish; 61% of his fights reach the judges.
3-2 in his last 5. 100% of his wins come by finish; 20% of his fights reach the judges.
Tommy McMillen
Marwan RahikiNeither fighter has enough UFC history in our corpus to build a read from.
2-0 in his last 2. 100% of his wins come by finish; 0% of his fights reach the judges.
2-0 in his last 2. 100% of his wins come by finish; 0% of his fights reach the judges.
Alexa Grasso
Manon FiorotManon Fiorot wins 89% of his bouts to Grasso's 64% — and the matchup profiles as one that reaches the judges.
3-2 in his last 5. 33% of his wins come by finish; 71% of his fights reach the judges.
4-1 in his last 5. 38% of his wins come by finish; 67% of his fights reach the judges.
Curtis Blaydes
Waldo Cortes AcostaThe model reads it 59% — nothing in either tracked record separates them cleanly, so the number rests on the model's wider inputs rather than a headline stat.
2-3 in his last 5. 57% of his wins come by finish; 35% of his fights reach the judges.
3-2 in his last 5. 50% of his wins come by finish; 62% of his fights reach the judges.
Dan Ige
David MartinezDavid Martinez wins 100% of his bouts to Ige's 52%, the shorter record belongs to him (3 vs 21), so the read leans on his opponent's history — and the matchup profiles as one that reaches the judges.
1-4 in his last 5. 55% of his wins come by finish; 67% of his fights reach the judges.
3-0 in his last 3. 33% of his wins come by finish; 67% of his fights reach the judges.
Tim Elliott
Édgar CháirezÉdgar Cháirez wins 67% of his bouts to Elliott's 45%, 75% of his wins come by finish — and the matchup profiles as one that reaches the judges.
3-2 in his last 5. 30% of his wins come by finish; 64% of his fights reach the judges.
4-1 in his last 5. 75% of his wins come by finish; 50% of his fights reach the judges.
Muslim Salikhov
Ignacio BahamondesIgnacio Bahamondes he finishes 83% of his wins and Salikhov has been finished in 80% of his losses — and the matchup profiles as one that reaches the judges.
3-2 in his last 5. 56% of his wins come by finish; 36% of his fights reach the judges.
3-2 in his last 5. 83% of his wins come by finish; 50% of his fights reach the judges.
Yousri Belgaroui
Djorden SantosYousri Belgaroui wins 100% of his bouts to Santos's 33%, he finishes 100% of his wins and Santos has been finished in 50% of his losses — and the finish profile points to a knockout.
2-0 in his last 2. 100% of his wins come by finish; 0% of his fights reach the judges.
1-2 in his last 3. 0% of his wins come by finish; 67% of his fights reach the judges.
Drakkar Klose
Tommy GanttTommy Gantt wins 100% of his bouts to Klose's 77%, he finishes 100% of his wins and Klose has been finished in 67% of his losses — and the matchup profiles as one that reaches the judges.
4-1 in his last 5. 20% of his wins come by finish; 69% of his fights reach the judges.
1-0 in his last 1. 100% of his wins come by finish; 0% of his fights reach the judges.
Rafa Garcia
RongzhuThe model separates these two by very little — 52% is close to a coin flip, and nothing in either record breaks the tie cleanly.
4-1 in his last 5. 29% of his wins come by finish; 73% of his fights reach the judges.
3-2 in his last 5. 33% of his wins come by finish; 50% of his fights reach the judges.
Jessie Rosas8-1-0Neither fighter has enough UFC history in our corpus to build a read from.
JJ Aldrich
Regina TarinRegina Tarin wins 100% of his bouts to Aldrich's 65%, the shorter record belongs to him (1 vs 17), so the read leans on his opponent's history — and the matchup profiles as one that reaches the judges.
4-1 in his last 5. 9% of his wins come by finish; 82% of his fights reach the judges.
1-0 in his last 1. 0% of his wins come by finish; 100% of his fights reach the judges.
Three markets publish — winner, method and ending round — each tested separately by walking forward through 3,557 held-out fights from a corpus of 8,642 bouts (1994-03-11 to 2026-08-08). A head that fails its bar is withheld while the others still publish.
| Market | Model | Baseline | Log loss |
|---|---|---|---|
| Winner | 60.6% | 50.0% | 0.661 vs 0.693 |
| Method | 52.7% | 46.7% | 0.969 vs 1.025 |
| Ending round | 60.6% | 55.3% | 0.915 vs 0.955 |
Accuracy is measured against a baseline that always answers the historical base rate. Paper and educational — research estimates, never advice, and never presented as an advantage over the sportsbook price.
Moneyline: This card publishes the model's own probabilities and carries no price beside them. UFC prices are captured separately and live in the odds and moneyline-grading artifacts; they are not joined into this document.
Schedule source: espn_scoreboard
Record
Current era · fight model vs de-vigged priceHow the model's picks actually turned out
Fight-winner picks: the model's chosen fighter and its probability, graded against the official result.
16 graded picks is far too few to say anything about accuracy. These are published so the predictions can be checked, not as evidence that they work.
The market's own de-vigged probability for the same bout is shown alongside. This is the only sport here where the two are recorded together.
| Date | Event | Model said | What happened | Result |
|---|---|---|---|---|
| 2026-08-29 | Umar Nurmagomedov vs Song Yadong | Umar Nurmagomedov · 52.5% | Song Yadong | MISS |
| 2026-08-29 | Denise Gomes vs Yan Xiaonan | Denise Gomes · 69.6% | Denise Gomes | HIT |
| 2026-08-29 | Andre Lima vs Namsrai Batbayar | Andre Lima · 61.1% | Andre Lima | HIT |
| 2026-08-29 | Rei Tsuruya vs Kevin Borjas | Rei Tsuruya · 73.4% | Rei Tsuruya | HIT |
| 2026-08-29 | Sean Woodson vs Jack Jenkins | Sean Woodson · 66.6% | Sean Woodson | HIT |
| 2026-08-29 | Julia Polastri vs Jingnan Xiong | Julia Polastri · 74.5% | Julia Polastri | HIT |
Showing the 6 most recent of 16 graded.
Settled archiveArchive · no live coverageretired moneyline era — one settled card
Settled from ESPN MMA scoreboard (site.api.espn.com/apis/site/v2/sports/mma/ufc/scoreboard, event 600058854, STATUS_FINAL). Winners + ending round/time + scheduled rounds are official from the ESPN MMA feed. Specific finish method (KO/TKO vs submission) is not exposed by the connected feed, so method grading is at the finish-vs-decision level only; KO-vs-submission is needs_review. Expanded model-only props graded for calibration only — no betting P&L. Paper-only educational tracking.
This is one settled event (7 graded fights). A single-card record is an outcome log, not model validation — the moneyline model was retired unvalidated.
The two eras never share a metric. The current record grades a three-head fight model against the de-vigged price recorded before each card; the archive is the retired read's outcome log from before any model was scored against anything. Adding them together would average two different questions.
UFC 329: McGregor vs. Holloway 2 (July 11, 2026) was listed with market-implied reads only before the event; no model picks were published for it and no official settlement was ingested, so no record is claimed for it.
What the model is most confident about today
What each simulation is most confident about today, ranked by the model's own probability — not by any gap against a sportsbook price. A watchlist, not a bet.
UFC · Fight winner · Welterweight · model reads DEC78.7%UFC · Fight winner · Flyweight · model reads DEC77.6%UFC · Fight winner · Featherweight · model reads KO67.5%UFC · Fight winner · Bantamweight · model reads DEC65.1%UFC · Fight winner · W Flyweight · model reads DEC64.9%ufc — The one model here that cleared its preregistered bar, on a 3,557-fight held-out sample. Since 2026-08-22 each settled bout is graded beside the market's own de-vigged probability — 16 so far, far too few to support any claim either way.
NFL — not listed: the model does not currently tell these teams apart — its team-strength term was measured and found not to be statistically significant, so it produces near-identical reads for every game.
Paper-only and educational. Nothing here is a pick or a recommendation to wager, and no stake is filled in anywhere on this site.