Sports Betting Bot

A 24/7 NBA betting analysis bot written in Go that identifies +EV opportunities on Kalshi by comparing prediction market prices against sportsbook consensus.

How It Works

1. Polls odds from 14+ sportsbooks via balldontlie.io (600 req/min)
2. Calculates consensus probabilities with vig removal and log-linear pooling
3. Filters for freshness — only uses odds updated within 30 seconds
4. Compares against Kalshi prices to find +EV opportunities
5. Rejects longshots — skips bets where true probability < 15% or > 85%
6. Sizes bets using Kelly criterion (quarter-Kelly default)
7. Optionally executes trades via Kalshi API with RSA authentication

Features

• Log-linear opinion pool consensus with winsorized outlier capping and Power method vig removal
• Spread/total line normalization using Student's t-distribution with context-dependent SD
• Fee-adjusted EV calculation (accounts for Kalshi's dynamic fee: 0.07 * price * (1-price), capped at $0.0175)
• Longshot filter: skips bets where true probability < 15% or > 85%
• Staleness filtering: rejects vendor odds older than 30 seconds (game and prop markets)
• Order book depth and slippage analysis
• Arbitrage detection on existing positions
• SQLite position tracking with hedge alerts
• Deduped console alerts (5-min cooldown)

Quick Start

# Clone
git clone https://github.com/mchen04/sports-betting-bot.git
cd sports-betting-bot

# Configure
cp .env.example .env
# Edit .env with your API keys

# Run
go run ./cmd/bot

Configuration

Variable	Default	Description
BALLDONTLIE_API_KEY	required	Ball Don't Lie API key
KALSHI_API_KEY_ID	optional	Kalshi API key ID
KALSHI_API_KEY_PATH	optional	Path to RSA private key (local dev)
KALSHI_PRIVATE_KEY	optional	RSA private key content (cloud deployment)
EV_THRESHOLD	0.03	Min EV to alert (3%)
KELLY_FRACTION	0.25	Fraction of full Kelly
MAX_ODDS_AGE_SEC	30	Max age of vendor odds in seconds
AUTO_EXECUTE	false	Auto-execute trades
MAX_BET_DOLLARS	0	Max bet size (0 = no cap)
KALSHI_DEMO	false	Use demo API

See docs/architecture.md for full configuration options.

Documentation

• Architecture - System design and configuration
• Math - EV calculation, Kelly criterion, vig removal
• Player Props - Props analysis and line interpolation

Project Structure

├── cmd/bot/           # Entry point
├── internal/
│   ├── api/           # balldontlie.io client
│   ├── kalshi/        # Kalshi API, orderbook, orders
│   ├── odds/          # Consensus, conversion, vig removal
│   ├── analysis/      # EV detection, Kelly sizing, longshot filter
│   ├── mathutil/      # Logit/sigmoid, t-distribution, beta functions
│   ├── config/        # Configuration loading & validation
│   ├── engine/        # Polling loop, trade execution, ticker mapping
│   ├── positions/     # SQLite tracking, hedge detection
│   └── alerts/        # Notification system
├── Dockerfile         # Multi-stage build
└── fly.toml           # Fly.io deployment

Deployment

Deploy to Fly.io:

fly launch
fly secrets set BALLDONTLIE_API_KEY=xxx KALSHI_API_KEY_ID=xxx
fly deploy

Data Sources

• balldontlie.io - GOAT tier ($39.99/mo) for 14+ sportsbook odds
• Kalshi - Prediction market prices and execution

License

MIT