Over 2.5 goals · Thursday, October 15
The day's fixtures that the engine rates more likely than not to produce three or more goals, most likely first. Each price is a bookmaker's where one was quoted, and the engine's own fair price where not.
Last 30 days, every settled call
59.8%183 of 306 landed, against 50.4% of all matches on record
Live, stored before kick-off
61.4%156 of 254
Reconstructed after the match
51.9%27 of 52
Over 2.5 is only counted as a pick where the engine rates it above 50%. A fixture it rates under is not a pick that failed to land. Read the rate against how often over 2.5 happens anyway, quoted above: beating that is the claim, not the raw percentage.
15 of 15 fixtures on this card had a view for this market. Prices marked ≈ are the engine’s own fair price, not a bookmaker’s.
Only the calls that landed are listed here. The rate above counts the ones that did not as well.
Over 2.5 goals means three or more goals in the match, by either side, in normal time. The probability on each row comes from a model trained on completed matches for that line, blended with the engine's Poisson goal model, which prices every scoreline from each team's attacking and defensive record.
A high number here says the match is expected to be open, not that the bet is good value. A price already reflects how likely a market thinks the outcome is, so the question worth asking of any row is whether the engine's probability is meaningfully above what the price implies. The track record above is the evidence for whether it usually is.
Three or more goals scored in the match by both teams combined. Extra time and penalty shoot-outs do not count in cup matches; the 90-minute score settles the call.
It is the engine's probability that the match finishes with three or more goals. It is a probability, not a guarantee. A 60% pick is expected to miss four times in ten.
No bookmaker price was stored for that line, so the figure shown is the pick's own fair price: 100 divided by its probability. It carries no bookmaker margin, so it is not a price anyone will pay and should not be read as value against the market.
A live call was stored before kick-off, so nobody could have known the result. A reconstructed call was generated after the match from history cut off at kick-off. Both are counted, separately, and the live figure is the stronger evidence.
It is a recent-form figure for this one market. The full record across every market is on the track-record page.