Probetrics — Sports Analytics

How to use Probetrics

Five minutes, every tool. The short version: the fight pages show you what the model thinks and why, the board and edges tell you where those numbers disagree with the sportsbooks, and Results shows our graded track record so you can decide how much to trust us. Hover any ? on the site for a plain-English definition of that stat.

Every upcoming card, grouped by event. Each fight links to a full breakdown: the model's pick (winner, method, round window), tale of the tape, stat comparison, pace and finishing data, and a written read on the matchup.

When to use it: Start here in fight week. It's the whole card in one place, and each fight page is the deep dive.

  • The Model Pick banner is the headline: winner, most likely method, and a round window or goes-the-distance call.
  • Probabilities are calibrated — squeezed toward the betting market where a line exists and discounted in chaotic matchups.
  • Once a fight is scored, the page shows the locked pre-fight pick next to the actual result — no hindsight edits.
  • Fighter names link to full stat profiles with complete fight histories.

Every bout priced against the model: our win probability for each fighter next to the best available moneyline across sportsbooks, with the edge highlighted.

When to use it: When you want the whole card as numbers — one row per fight, sortable, with the value side flagged.

  • Best price carries the book's name — clicking odds opens that sportsbook.
  • Edge = our probability minus the market's with the vig removed. Bigger gaps deserve more skepticism, not more money.

Simulate any two of 600+ tracked fighters — dream fights, rematches, cross-division what-ifs — with full win/method/round probabilities and a stat-by-stat comparison.

When to use it: Any time. It's the same model that grades itself on real cards, pointed at whatever matchup you can imagine.

  • Pick two fighters and a 3- or 5-round format; championship rounds change cardio and finish math.
  • The rounds distribution shows when the fight ends if it doesn't go to the cards.

Fighters & Rankings

Open Fighters & Rankings

Official UFC rankings for every division, plus a full stat profile for every tracked fighter: output and defense rates, pace, finishing splits, complete fight-by-fight history, and how our model has scored their fights.

When to use it: Scouting a fighter — or wandering from a ranked name to their tape to a Fight Lab simulation.

  • "C" marks the champion; +/− shows rank movement since the last update.
  • Every ranked fighter we track links straight to their profile; profiles link every past opponent.
  • Each profile shows the model's receipts against that fighter's actual fights.

The model's strongest disagreements with the market, grouped by event and ranked. The shortlist, not the whole board.

When to use it: When you want the fights where our numbers and the books' numbers differ the most.

  • Edge % = model probability minus the market's implied probability with the vig removed.
  • Bigger edge does not mean safer — it means bigger disagreement. Treat 10%+ edges with more skepticism, not more money.

Auto-written, data-backed nuggets from the fight model: finishing streaks, style patterns, and matchup facts tied to bettable markets.

When to use it: Skim it for angles you wouldn't have looked for — every insight links to the underlying fight page.

  • Each insight is generated from our own ingested data — nothing is copied from touts.
  • Pattern insights ("won his last five by KO") map directly onto method-of-victory markets.

Results — our track record

Open Results — our track record

Every model prediction, locked before the fight and graded after, with running totals: winner and method hit rates, $100 flat-bet ledgers, per-event breakdowns, and calibration checks. No hindsight, no deleting misses.

When to use it: Before you trust anything else on this site, read this page. It's also where you check how the model did after each card.

  • Picks are snapshotted while fights are upcoming — what gets graded is exactly what the site displayed, misses included.
  • The $100 ledgers answer the only question that matters: would following the model have made money?
  • The calibration panel shows whether our percentages mean what they say (a 60% claim should win about 60% of the time).

Fight week with Probetrics

  1. 1 · Early week: Lines post — scan Fights for the card and Edges for the model's biggest disagreements with the books.
  2. 2 · Research: Click into each fight page — does the tape, pace data, and style read support the edge? Check both fighters' profiles for the full history.
  3. 3 · Price: The Fight Board highlights the best available price on every side — line shop before you bet.
  4. 4 · Fight night: Our picks are already locked — what you saw is what gets graded.
  5. 5 · Sunday morning: Results grades the whole card in public: winners, methods, rounds, and the money.

Probetrics is a research tool, not a tout service — every number is explained, every prediction is graded in public, and nothing is ever "guaranteed." Bet only what you can afford to lose. 21+, 1-800-GAMBLER.