Pixie Dust scans prediction markets for statistical edges and trades them automatically. Weather, politics, NBA, NHL, MLB, CBB, tennis — all from your desktop.
Purpose-built scanners analyze real-time data feeds for statistical mispricings across prediction market categories.
Compares NWS forecast distributions against Kalshi contract prices for 20+ cities. Temperature, precipitation, wind speed, snowfall.
NOAA + NWS feedsMonitors Congressional activity, SCOTUS decisions, and Cabinet markets. Sentiment-scored social signals feed into probability models.
Real-time sentimentPitcher matchups, bullpen fatigue, park factors, and lineup data synthesized into game outcome probabilities. Updated through first pitch.
Live game dataLive moneyline scalping with zone-gated entries, resting stop orders, and tiered take-profit. Plus over/under totals from live game flow.
ESPN + live scoresHockey moneyline and puck line scanning with multi-factor win probability models. Goalie matchups, power play efficiency, and recent form.
ESPN + NHL APIMarch Madness and NIT markets. DraftKings odds, team stats, home court advantage, and scoring models for moneyline edge detection.
ESPN + DraftKingsATP and WTA match scanning with multi-source fallback (API-Tennis, Sofascore, ESPN). Surface-adjusted Elo ratings and head-to-head data.
Multi-source feedsReal-time portfolio tracking, signal history, and scanner status in a native desktop app.
From data ingestion to order execution in milliseconds, fully automated.
Scanners pull live data from government APIs, news feeds, and sports databases on a continuous loop.
When a scanner detects a statistical edge above your configured threshold, it emits a trade signal with confidence and sizing data.
Signals pass through Kelly criterion sizing and risk guardrails, then execute on Kalshi automatically. Or flag for manual approval.
No cloud dependency. Your keys stay on your machine. The engine runs locally for maximum speed and privacy.
Signal generation to order placement in under 50ms. Native speed, no browser overhead.
PEM-based authentication. Private keys stored locally, never transmitted to any server.
Fractional Kelly with configurable fraction, daily stop loss, and per-category price bands.
Live P&L, position tracking, signal history, and scanner status from a Tauri-powered desktop app.
Every trade passes through multiple risk checks before execution. Configurable limits you control.
Engine halts all trading when daily P&L hits your configured loss limit
Maximum contracts per trade and concurrent open position caps
Minimum statistical edge required before any signal is generated
Trading pauses automatically when account balance drops below your minimum
Professional-grade trading infrastructure. Choose the level that matches your edge.
Weather scanner with manual trade approval. See what the bot sees.
Weather + political scanners with moderate limits. Signal-driven edge.
All scanners, moneyline scalper, 4Q system, auto-trading, API access, and no limits.
Get notified when the next release drops, or download the latest build now.
Android companion requires the engine running on your PC, Mac, or server. Pair via QR code or manual connect.
mkdir -p /opt/pixiedust/keys /opt/pixiedust/datacd /opt/pixiedustcurl -Lo kalshi-engine https://pixiedustbot.com/downloads/kalshi-engine-linux-x86_64curl -Lo config.toml https://pixiedustbot.com/download/config.toml.examplechmod +x kalshi-engine
config.toml with your Kalshi API key ID and copy your .pem private key file into the keys/ folder:nano config.tomlcp ~/your-kalshi-key.pem keys/kalshi-private.pem
./kalshi-enginehttp://your-server-ip:3847/pair in any browser to get a scannable QR code and pairing token. Or check the engine logs:grep "token" /opt/pixiedust/engine.log
http://your-tailscale-ip:3847 from anywhere — no port forwarding needed.
curl -Lo docker-compose.yml https://pixiedustbot.com/download/docker-compose.ymlcurl -Lo .env.example https://pixiedustbot.com/download/.env.examplecp .env.example .env && nano .env — paste your Tailscale auth keydocker compose up -d