🎯
Welcome to Nostrabot
Automated dutching & arbitrage on prediction markets.
🔍 Scans Continuously
Polls Kalshi & Polymarket every few seconds for mispriced multi-outcome events.
💰 Sizes with Kelly
Quarter-Kelly position sizing keeps bets proportional to edge and bankroll.
🌡️ Triages Live
An NWS-driven agent reviews open weather positions and recommends exits.
📊 Tracks Everything
Every scan, decision, near miss, fill, and P&L is recorded for review.
The next 3 slides explain the two strategies and how the pipeline works. Click Teach me in the header anytime to revisit a guided tour.
🎲 Strategy 1: Dutching
Buy YES on multiple outcomes of a mutually-exclusive event so that you profit no matter which one wins — as long as the combined cost is below the $1.00 payout.
Example: Atlanta high temperature today
Total cost per contract:
$0.90
Payout if any wins:
$1.00
Locked-in profit (per contract):
+$0.10 (~11%)
Wins when one of your covered brackets is the actual high.
Loses the cost only if the high lands outside your covered range.
Legs must be contiguous — we never buy 82–83 and 86–87 leaving the middle uncovered.
⚖️ Strategy 2: Cross-Exchange Arbitrage
When the same event trades on both Kalshi and Polymarket, prices can drift apart. Buying YES on one side and NO on the other for less than $1.00 total locks in risk-free profit.
Example: “Will candidate X win?”
POLYMARKET
Buy NO @
$0.55
Combined cost:
$0.97
Guaranteed payout:
$1.00
Risk-free edge:
+$0.03 (~3%)
How matches are found: A semantic-embedding pipeline pairs Kalshi and Polymarket markets that describe the same outcome, then a Gemini verifier confirms the pair before it’s eligible for arbitrage.
🔁 The Pipeline
Every scan cycle (configurable per profile) flows through six stages. You can inspect the result of each stage on the Report tab.
1. Scan
Pull active events from Kalshi/Polymarket REST + WebSocket feeds.
↓
2. Filter
Apply rules: liquidity, time-to-close, category, weather-bracket adjacency.
↓
3. Evaluate
For every K-leg subset compute coverage cost, EV, edge, and Kelly size.
↓
4. Decide
Pick best subset by edge or coverage; record near-misses for borderline events.
↓
5. Execute
Sim profiles log virtual fills; live profiles place real Kalshi/Polymarket orders.
↓
6. Triage & Settle
Monitor open positions (NWS forecasts for weather), recommend exits, settle P&L.