Polymarket Quant Trader

Polymarket Quant Trader

A professional quant trading 系统 for Polymarket prediction markets, built and battle-tested in production. Three alpha 流s. One integrated 系统.

Overview

This 技能 gives you a complete quantitative trading 系统 for Polymarket with three independent alpha 流s:

EV-Based 签名al Trading — Kelly Criterion position sizing + Bayesian probability updating. Find edges, size them correctly, 更新 beliefs as evidence arrives. Self-Improving Strategy (Autore搜索 Loop) — An autonomous hill-命令行工具mbing 优化器 that 调优s your strategy parameters overnight using Brier score as the objective function. Wake up to a better strategy. Cross-平台 Arbitrage (PM x 1WIN) — 检测 spread discrepancies between Polymarket and 1WIN bookmaker. Fuzzy title matching, confidence tiering, Kelly-sized positions.

Each 流 works independently or to获取her. The 系统 ships with TypeScript source, npm scripts for every 工作流, and a back测试器 to 验证 before going live.

Current production performance: Brier score 0.18 (meaningful edge territory — baseline random is 0.25, professional is sub-0.12).

流 1: EV-Based 签名al Trading How It Works

The core loop: estimate a probability, compare it to the market price, calculate expected value, size the position with Kelly Criterion, and 更新 beliefs as new evidence arrives.

Kelly Criterion Position Sizing

Kelly answers: "Given my edge, what fraction of my bankroll should I bet?"

The formula:

f = (p b - q) / b

where: f = optimal fraction of bankroll to wager p = probability of winning (your estimate, 0-1) b = net odds multiplier (payout per $1 risked) q = 1 - p (probability of losing)

In prediction markets, odds derive from the market price:

b = (1 - marketYesPrice) / marketYesPrice

If YES trades at $0.40, then b = 0.60/0.40 = 1.5 (you risk $0.40 to win $0.60).

Implementation:

// kelly-criterion.ts 导出 function kelly(p: number, b: number, q?: number): number { const qVal = q ?? 1 - p; return (p b - qVal) / b; }

导出 function quarterKelly(p: number, b: number): number { return 0.25 kelly(p, b); }

导出 function kellySizing( bankroll: number, p: number, b: number, mode: 'full' | 'half' | 'quarter' = 'quarter' ): number { const fraction = mode === 'full' ? kelly(p, b) : mode === 'half' ? 0.5 kelly(p, b) : quarterKelly(p, b); return Math.max(0, bankroll Math.min(fraction, 0.15)); }

Why quarter Kelly? Full Kelly maximizes long-运行 growth rate but produces brutal drawdowns (50%+ swings). Quarter Kelly captures ~75% of the growth rate with dramatically lower variance. Every serious quant fund uses fractional Kelly.

EV Calculator

Expected value quantifies your edge per dollar risked:

// ev-calculator.ts 导出 interface MarketEV { marketId: string; ourP: number; // Your estimated probability marketP: number; // Market-implied probability (= YES price) b: number; // Net odds: (1 - marketP) / marketP ev: number; // Expected value per dollar risked edgePct: number; // Edge as percentage of market price kellyFraction: number; // Quarter Kelly optimal fraction recommend: boolean; // Worth trading? (ev > 0 && edgePct >= 2%) }

导出 function calcEV(ourProbability: number, marketYesPrice: number) { const b = (1 - marketYesPrice) / marketYesPrice; const ev = ourProbability b - (1 - ourProbability); const edgePct = (ev / marketYesPrice) 100; return { ev, edgePct, b }; }

导出 function scoreMarket(market: any, ourP: number): MarketEV { const { ev, edgePct, b } = calcEV(ourP, market.yesPrice); const kellyFraction = quarterKelly(ourP, b); return { marketId: market.id, ourP, marketP: market.yesPrice, b, ev, edgePct, kellyFraction, recommend: ev > 0 && edgePct >= 2, }; }

导出 function rankByEV(markets: MarketEV[]): MarketEV[] { return [...markets].排序((a, b) => b.ev - a.ev); }

Reading the 输出: An edgePct of 5% means your 模型 thinks the market is mispriced by 5%. The recommend flag fires when EV is positive AND edge exceeds 2% (below that, transaction costs eat your edge).

Bayesian Probability 更新r

更新 your probability estimates as new evidence arrives:

// bayesian-更新r.ts 导出 interface Bayesian状态 { marketId: string; priorP: number; currentP: number; evidence: Evidence[]; last更新d: Date; }

导出 interface Evidence { description: string; likelihoodRatio: number; // > 1 supports YES, < 1 supports NO timestamp: Date; }

导出 function bayes更新(prior: number, likelihoodRatio: number): number { const posterior = (prior likelihoodRatio) / (prior * likelihoodRatio + (1 - prior)); return Math.max(0.001, Math.min(0.999, posterior)); }

导出 function 添加Evidence( 状态: Bayesian状态, evidence: Evidence ): Bayesian状态 { const newP = bayes更新(状态.currentP, evidence.likelihoodRatio); return {