Pillar Evaluation & Probability Calibration

Pillar Scoring Framework

The system evaluates matchups by combining sport-specific parameters into an objective confidence score (C). The scoring model evaluates three core pillars, each scaling from 0 to 10:

1. Technical Sport (P_1): Evaluates pitching metrics (composite ERA calculated via a 70/30 blend of last year's regular season and spring training, or active season ERA after a 15-inning sample size), lineup quality (Elite/Hot bats), bullpen fatigue logs, and platoon splits. 2. Seasonal/Environmental (P_2): Integrates venue metrics (e.g. Coors Field hitter boost of +2.5, Petco Park pitcher boost of +1.5), weather parameters (wind speeds and directions), and ramping factors. 3. Technical Bookies (P_3): Tracks implied pricing from traditional bookmakers compared to Polymarket crowd prices to resolve Expected Value (EV):

$EV = (C / 100) - Polymarket Share Price$

Sizing and Risk Mitigation

export function getSizing(confidence: number, bankroll: number): { label: string, amount: number } {
    if (confidence >= 80) return { label: "Aggressive (7.5%)", amount: bankroll * 0.075 };
    if (confidence >= 70) return { label: "Standard (4.0%)", amount: bankroll * 0.04 };
    if (confidence >= 60) return { label: "Caution (2.0%)", amount: bankroll * 0.02 };
    return { label: "Zero (0%)", amount: 0 };
}

If the system detects a performance drawdown (e.g. 3 consecutive losses or 4 of the last 5 settled as losses in Supabase), it enters Slump Mode, automatically reducing the suggested stake sizes by 50% (0.5× multiplier) to preserve capital during high-variance periods.