.agents/memory/2026-02-27-pm-kalshi-algorithm-explanation-task.md

2026-02-27 Session Notes

PM-Kalshi Algorithm Explanation Task

Nicholai requested an explanation of the pm-kalshi prediction market trading algorithm targeted at an investor audience familiar with Amazon ad optimization. The goal was to explain the technical details without skipping them, but using familiar concepts (budget allocation, bidding strategies, performance optimization) as analogies.

To prepare a detailed explanation, an Explore agent was launched to thoroughly analyze the pm-kalshi and pm-garden crates in the OpenMarketUI monorepo. The agent was tasked with understanding:

• Full pipeline flow (market sourcing, filtering, scoring, selection, execution)
• Individual scorer mechanics (momentum, mean reversion, volume, ensemble)
• Backtest and paper trading engines
• Risk management and position sizing
• Tunable configuration parameters

The session was in early discovery phase, with the agent beginning to map the codebase structure across pm-kalshi, pm-garden, and pm-core.