Implied probability is the conversion of odds into the win percentage the bookmaker's price represents. It's the foundation for all EV calculations and model comparisons.
American odds conversion: - Negative odds: Probability = |odds| / (|odds| + 100) × 100%. Example: -150 → 150/250 = 60% - Positive odds: Probability = 100 / (odds + 100) × 100%. Example: +130 → 100/230 = 43.5%
Decimal odds: Probability = 1 / decimal odds. Example: 2.10 → 47.6%
The critical step after calculating implied probability is removing the vig to get the fair probability (see no-vig odds). A -110 line implies 52.4% but the fair probability is 50%.
Model-building tip: Your model should output probabilities, not predictions. Then you compare your model's probability to the no-vig implied probability. When your model says 58% and the fair probability is 50%, you have a potential edge — if the model is calibrated correctly.
Calibration means your 60% predictions should win 60% of the time at scale. A well-calibrated model's probability estimates are directly actionable for EV and Kelly calculations.