Sentiment Market MM
Updated Mar 2026
PythonFastAPIPostgreSQLRedisReactTypeScriptDocker
this is a full-stack market-making system for prediction markets. it ingests real-time sentiment signals, prices two-sided quotes with dynamic spread adjustment, manages inventory exposure, and executes across multiple venues through a unified order management layer.
sentiment-weighted spread pricing that adjusts bid-ask width based on news flow, social volume, and order book imbalance
inventory risk engine with position limits, greeks-inspired exposure tracking, and dynamic skew to reduce directional risk
multi-venue execution layer with order lifecycle management across Kalshi and Polymarket
what i built
- fastapi backend with sentiment ingestion, spread pricing engine, order management, and position tracking
- postgres + redis schema for markets, quotes, orders, fills, positions, and real-time sentiment state
- react monitoring dashboard for live P&L, position exposure, spread visualization, and order flow
how it works
- 1ingest sentiment signals from news APIs, social feeds, and order book data into a scoring pipeline
- 2price two-sided quotes with spread width driven by sentiment confidence, volatility, and current inventory skew
- 3execute and manage order lifecycle across venues with fill tracking and position reconciliation
results
- ✓the whole market-making loop from signal ingestion through quote pricing to execution and risk tracking
- ✓gives me a sandbox for testing different spread strategies against real prediction market data
what's next
- add a backtesting harness to evaluate spread strategies against historical order book snapshots
- implement adaptive position sizing based on realized P&L and drawdown thresholds