Signal Logic

A signal is where the model's estimated probability of an outcome diverges above the probability implied by the bookmaker's odds. Signals are not predictions, and a divergence is not a guaranteed advantage — it is a quantitative model's view, surfaced transparently so you can compare it against the market price yourself.

Over a large sample, positive-EV signals should yield a positive return on investment. Individual outcomes are subject to variance. A signal that does not land is not a model failure — it is the expected cost of operating under uncertainty.

Decimal odds represent the bookmaker's price for an outcome. They encode an implied probability:

Implied Probability = 1 ÷ Decimal Odds

A line of 2.50 implies a 40% chance of winning. Bookmakers add a margin (overround) so the combined implied probabilities across all outcomes in a market sum to more than 100% — this is the structural house margin built into every price.

The model computes the statistical likelihood of each outcome independently of market pricing. When the model probability diverges sufficiently from the implied probability in the odds, a signal is surfaced.

• Football — Dixon-Coles Poisson score matrix calibrated on recent form, H2H record, and rest-day fatigue, refined by a stacked XGBoost meta-model that adjusts expected goals using 17 contextual features.
• Tennis — Surface-adjusted Elo rating derived from match-level results on each court type.
• NBA — Gaussian efficiency model built on offensive and defensive ratings.

EV measures the average return per unit staked, assuming the model probability is correct. The formula is:

EV = (P_win × Profit) − (P_loss × Stake)

Which simplifies to the form used by the model:

EV = (Model Probability × Decimal Odds) − 1

A positive EV means the model estimates the expected return exceeds the cost of staking over many independent trials. EV is displayed as a percentage of the stake. Only signals above the configured threshold are surfaced.

EV quantifies the size of the model–market divergence, not certainty. High EV does not mean the outcome is likely, and it does not mean the market is wrong — it means the model's estimate diverges from the market price. Divergence is a model view, not a proven advantage.