Sportsbook prices next to our simulations
Market probabilities are de-vigged from the posted price — not the book's own numbers.
MLB prices shown here are from the Oct 28 capture. Premier League, UFC and NFL forecasts are on their own hubs.
model 58.6% − market 66.6% = −8.0 pp
Our simulation says 58.6%; the sportsbook’s margin-free price implies 66.6% — pp is the gap between two percentages, not scoring points.
A large gap is not automatically a good bet — our settled record shows the model does not beat the market — treat it as a question, not an answer.
Full reading key · every term on this page
- Model probability
- What our simulation estimates the chance of this outcome to be. It is an estimate, not a prediction of what will happen.
- Market-implied probability
- The chance the sportsbook's price implies, after removing their built-in margin (the 'vig'). This is what the market thinks.
- American odds
- The sportsbook's price. −150 means you would risk 150 to win 100; +150 means you would risk 100 to win 150. A negative number is the more likely side.
- Moneyline
- A bet on which team wins outright, with no handicap.
- Run line / spread
- A handicap applied to the favourite, so both sides price closer to even.
- Total
- The combined score of both teams, priced over or under a line.
- pp (percentage point)
- The unit of the difference column. It is the gap between two percentages — NOT scoring points, and not a percentage change. 58.6% and 66.6% are 8.0 pp apart.
- Difference
- Model probability minus market-implied probability, in pp. Positive means our model is higher than the market on the listed side; negative means lower. It is a measurement, not a recommendation.
- Data freshness
- When the price and inputs behind a row were captured. Older captures are less reliable.
- Incomplete input
- A required input (a lineup, a price, a stat) was missing when the row was built, so the model view is withheld or partial rather than guessed.
- Qualified
- A market that met the published policy for appearing on a paper card. Most rows are never qualified — that is normal.
- No-play
- The slate was checked in full and nothing met policy. It is a real answer, not a missing update.
Model-ranked picks · full list (0)
Every market the model ranked for 2026-10-30, ordered by its own ranking score. Ranking is model confidence and market reliability — not a prediction of profit, and not ordered by the size of the difference. Terms are defined in the reading key above.
No ranked picks for 2026-10-30. That is the model’s answer for this slate, not a missing update.
Game markets
Milwaukee Brewers@
Tampa Bay RaysMarket disagreement explorer
For each row: what the sportsbook price converts to, what the simulation produced, how far apart they are, what happened afterwards, and how rows that disagreed by a similar amount have performed historically. A difference is a disagreement between two estimates. On settled history the sportsbook price has been the better estimate, so a large difference is something to look at and never a recommendation. Paper and educational only.
Only rows with a pregame capture record are listed individually — rows whose observation time relative to first pitch cannot be proven are excluded from the list. They still appear inside the historical ranges below, where the number of rows behind every figure is shown.
Every settled row from 2026-05-16 to 2026-10-28, grouped by how far the simulation sat from the sportsbook price. 49,894 rows are counted. 35 were refused by an integrity check and appear in no figure here. Ranges with no observations show no rate rather than a zero.
| Range | Rows | Came in | 95% interval | Brier |
|---|---|---|---|---|
| 2 pp or more below the market | 0 | — | — | — |
| within 2 pp of the market | 6,038 | 47.3% | 46.1% – 48.6% | 0.239 |
| 2–5 pp above the market | 9,349 | 52.0% | 50.9% – 53.0% | 0.242 |
| 5–10 pp above the market | 14,123 | 51.8% | 51.0% – 52.7% | 0.245 |
| 10–20 pp above the market | 16,723 | 49.7% | 48.9% – 50.5% | 0.265 |
| 20 pp or more above the market | 3,661 | 48.0% | 46.3% – 49.6% | 0.303 |
What this page does and does not show
A difference is reported in percentage points and nothing more. Our simulations have not been shown to out-predict the sportsbook — on settled history the market price is the better estimate — so a gap is a disagreement worth looking at, never a recommendation.
Freshness describes the captured file, not individual rows. The feed carries no per-row update time, so there is no “updated N minutes ago” anywhere on this page.
There is no opening price, no line movement and no trend history here, because no retained snapshot series exists yet. When one does, it will be built from prices captured going forward.