Kalshi NFL workflow

kalshiNFL.md

Full Order Pipeline Overview

This pipeline turns market data and probabilistic forecasts into executable orders by fusing rating inputs, Monte Carlo simulations, and Kalshi order book intelligence into a single automated run.

1. Assemble Inputs – Pull team power ratings, load the win distribution tables published by Inpredict NFL to anchor game-level priors, and ingest the latest Kalshi win-total, exact-win, and futures order books.
2. Run Season Simulations – nfl/scripts/bash/simulate_win_totals_and_exact_wins_and_futures runs a league-wide Monte Carlo that samples the Inpredict odds, applies custom tie-breaker logic for division and conference standings, and stores outcome frequencies for every team and market tenor.
3. Summarize Outputs – The simulator exports distributions for season win totals, exact win counts, and futures probabilities, which become the quantitative inputs for pricing edges against Kalshi markets.
4. Refresh Position State – scripts/bash/run_positions_pipeline.sh snapshots existing exposures, outstanding shares, and bankroll constraints so recommendations respect live risk limits.
5. Screen Markets – scripts/generate_skip_list.py computes a MAD-based filter that marks teams or contracts where projections are unstable or books are illiquid, creating a skip list used by downstream logic.
6. Generate Trade Ideas – scripts/generate_recommended_orders.py compares simulated fair prices to Kalshi bid/ask ladders, net of the skip list and risk tolerances, and writes structured recommendations to outputs/recommended_orders/*.json.
7. Validate Artifacts – The wrapper confirms the recommendation file path, ensuring a fresh JSON exists before attempting execution.
8. Submit Orders – scripts/place_recommended_orders.py --input <file> sends sized orders back to Kalshi, targeting the identified overlays while honoring current book depth and the skip constraints.