Prediction Market Lab
A paper-execution testbed for strategies running against live prediction-market data.
Personal research project
Source unavailable publicly
Strategy logic needs an observable test environment before live execution. The challenge is keeping market observation, simulated state and reporting consistent.
An observable paper testbed
- Strategy scanners against live Kalshi market information.
- An in-process paper executor and virtual-trader state.
- A web dashboard, optional Telegram digests and saved per-strategy results.
One runner, several views
- The observation layer fetches public market data and scans markets by volume.
- The paper runner hosts simulated execution and reporting in one process.
- Dashboard and digest summaries read the same in-memory state; periodic saves and shutdown saves preserve experiment output.
Simulated fills stay separate
- Run one paper-state owner instead of independent replicas with divergent balances.
- Use the same summary path for dashboard and digest output.
- Keep the paper runner distinct from the repository’s separate live-executor entry point.
Context and limits
This case study covers simulated fills only. The repository also contains a separate live-execution entry point; no claim of live trading results is made, and no account or position data is published.