sswadkar/kalshi-interface

Purpose and Entry Points

This repository is a lightweight Kalshi trading integration: a FastAPI service exposes status and trading operations, while shared Python logic authenticates Kalshi API calls and calculates position metrics. The operational entry point is server.py; core API/client and accounting logic resides in kalshi_positions.py. A minimal browser-facing page is available at static/index.html.

Service Runtime and Deployment

The service runs as Python/FastAPI, locally via python server.py or in a container defined by Dockerfile. Runtime configuration is environment-driven, selecting demo versus production credentials and a monitored event ticker. Dependencies are declared in requirements.txt.

Relevant paths: server.py, Dockerfile, requirements.txt.

Kalshi Authentication Boundary

Outbound requests to Kalshi pass through PSS/RSA signing logic in the core client layer. Private keys are environment-isolated, with distinct demo and production key locations. Key IDs and active environment are supplied externally through environment variables rather than embedded application logic.

Relevant paths: kalshi_positions.py, api_keys/demo/private_key.pem, api_keys/prod/private_key.pem.

Market and Position Polling

The server maintains real-time, in-memory state through two background polling loops: market/position refreshes run at roughly 0.5-second intervals, while resting-order refreshes run roughly every three seconds. This state is the source for the status API, creating a boundary between Kalshi’s remote API responses and locally served read status.

Relevant paths: server.py, kalshi_positions.py.

Trading API Flow

FastAPI endpoints accept buy, sell, and cancellation requests, delegate execution to the signed Kalshi client, and return the resulting order payload. Trading is an externally side-effecting boundary; orders are submitted to Kalshi rather than persisted in a local database.

Relevant paths: server.py, kalshi_positions.py.

Position Accounting

The core logic derives per-market liquidation value, realized PnL, and net exposure from fetched positions and market prices. It also applies the repository’s taker-fee model, 0.07 × P × (1 − P), when interpreting order costs and fees.

Relevant paths: kalshi_positions.py.

External Interfaces

Primary external dependency: Kalshi’s trading API, accessed with signed HTTP requests. Consumers can use the HTTP API on port 8000 or the included static status/UI asset. The repository has no indicated persistent datastore, queue, or streaming transport; process memory is the live state store.

Relevant paths: server.py, static/index.html, kalshi_positions.py.