How the Beezy Score works
A 0-100 strength score for one flagged price. It is not a win probability. Hover a score to see its parts.
- Edge, up to 60: 4 points per 1% of edge at the flagged price.
- Consensus, up to 20: how many other books back the fair price (3 books is 0, 12 or more is the max). Model bets have no books count and get a neutral 10.
- Hit probability, up to 20: longshots are penalised (15% to win is 0, 50% or better is the max).
Consensus and hit probability only count when there is an edge, so a price with no edge scores 0. STRONG is 70+, LEAN is 45-69, WATCH is under 45.
On 7,259 graded +EV alerts the tiers realised +25.8% (STRONG), +6.4% (LEAN) and +3.5% (WATCH) flat-stake ROI, paper, each book counted separately. The cut points were set on that same sample, so expect them to shrink going forward, and they are not yet validated for model bets.
KEY FEATURES (RESEARCH MODEL)
- Rolling SOG, TOI, shot attempts and power-play attempts per player
- Own-team shots for and opponent shots allowed
- Home / away, rest days, position, league shot environment
- Missing-teammate shot redistribution (experimental, off by default)
Against prices. NO_EDGE on 63 days of real lines: the model's log loss was 0.6872 vs 0.6816 for the market consensus, and bets on model disagreement lost -3.6% at the best price.
What the daily board runs. The daily board runs an empirical estimator: exponentially weighted game-log average, blended with a usage-adjusted rate, scaled by an opponent-allowed factor, with a negative-binomial or normal spread fitted from the league's own game-to-game variance. NFL means are calibrated on 2021-2025 outcomes (accuracy only: the pool of quoted players is over-projected by 3-7%). It does not model injuries, lineups, minutes, weather or teammate absences, and it has not been validated against prices.
Source: NHL shots-on-goal validation, 2026-10-14. Baselines are the best trailing-average forecaster in each study.