What influences the prediction.
Logistic regression contributionsThese factors explain the logistic regression model, one part of the ensemble. They describe model influence, not a fighter’s strengths or weaknesses.
Matchup research
Compare UFC form, statistics and fight history.
Checking data coverage…
No matchup selected
Your matchup report
Start with two fighters.
Choose names above or load the example. The report brings their recent fights, statistics and shared opponents into one view.
Matchup report
Edit matchupAt a glance
Striking and grappling
Recent fights
Last five recorded UFC bouts, most recent first.
Earlier fights & other promotions
Explore additional records
Partial histories, with bouts already shown in the UFC record excluded.
Previous UFC meetings
Shared UFC opponents
Data coverage & limitations
Set up an estimate
Choose the date, division and bout conditions.
Model estimate
Set the fight conditions
Choose the date, division and rounds for a hypothetical matchup.
How this model performed on later fights
Original model diagnostics Earlier, uncalibrated model · kept separate from the estimate above
Model data through the saved dataset date. These probabilities have not passed an independent calibration check. A large number here does not establish high confidence.
Red corner · model probability
Fighter One
—
— model-implied odds
vs.
Blue corner · model probability
Fighter Two
—
— model-implied odds
◂
Model favors pick a fight
▸
Price comparison
Compare the market with your estimate
Moneyline comparison
Enter both moneylines from the same bookmaker. Quotes you enter are not a live feed.
| Fighter | Offered price | Break-even | Market, no margin | Your estimate | Gap | Expected return / unit |
|---|
How to read this
Break-even is the win rate needed at the offered price. The market baseline removes the bookmaker’s margin by normalizing the two implied probabilities. Gap is your selected estimate minus that baseline, in percentage points.
Expected return is probability × decimal odds − 1 per unit staked. It is conditional on the probability estimate, before commissions and other fees. It is not a verified betting edge. The audited candidate has separate calibration and later-fight evaluation, but has not demonstrated a betting edge.
Your record
Research journal
Saved in this browser, with outcomes recorded by you.
Entry removed.