How the Ranking Is Calculated
For Over 2.5 and BTTS
Each match receives two separate ranks: one for Over 2.5 Goals and one for Both Teams to Score (BTTS). The two ranks are built from the same idea: estimate how open the game should be by combining expected goals, scoring form and defensive vulnerability for both teams.
- Attack strength is based on season xG plus short-term scoring form (recent goals and chances created) for both teams.
- Defensive vulnerability is based on recent goals conceded, shots on target conceded and the opponent's ability to create chances.
- Openness multiplier increases the score when both teams have high BTTS rate or regularly score in both halves.
- Expected total goals combines home attack with away defense and away attack with home defense. Higher expected total boosts Over 2.5 rank.
- BTTS likelihood focuses on the probability both teams score at least once, boosted when both have consistent scoring trends.
Final output: two ranks per fixture (Over 2.5 and BTTS) plus corresponding odds. Lower data quality fixtures are placed lower by default.
For 1X2
We estimate three probabilities for each fixture: Home Win, Draw and Away Win. The approach starts by building expected goals for both teams, then converting that expectation into win draw win probabilities.
- Home expected goals (λH) combines the home team attack indicators (season xG and recent scoring form) with the away team defensive indicators (recent goals conceded, shots conceded and xGA where available).
- Away expected goals (λA) combines the away team attack indicators with the home team defensive indicators in the same way.
- Game-state adjustment adds a small multiplier when both teams have high BTTS rate or high tempo signals, reflecting a more open match.
- Scoreline probabilities are computed from λH and λA using a goal distribution model (Poisson-based). From the scoreline matrix we derive: Home Win = P(H > A) Draw = P(H = A) Away Win = P(A > H).
- Ranking is done separately for each outcome by sorting fixtures by the corresponding probability, then comparing against bookmaker odds for value checks.
Output: three probabilities per fixture (Home/Draw/Away) with odds next to them. Fixtures with missing core inputs are flagged and pushed lower to reduce false confidence.