EDMN Trader turns binary prediction-market books into auditable same-market complement-parity evidence: normalize, scan, stress, paper, risk-gate, reconcile, validate.
Important
Public scope is research, paper, and Kalshi Demo. Production endpoints, credentials, and real-money execution are intentionally absent.
| Layer | Contract |
|---|---|
| Inputs | Kalshi Demo read-only data, Polymarket US public data, and local fixtures. |
| Semantics | Canonical Decimal YES-side books and same-market complement candidates. |
| Stress | Explicit fees, slippage, latency, partial fills, and failed-leg reserves. |
| Controls | Source hashes, paper ledger, risk limits, kill switch, and manual approval. |
| Demo | Dry-run previews, a guarded submit boundary, and deterministic reconciliation. |
| Evidence | Daily reports plus a rolling 7/30/90-day validation framework. |
implied_yes_ask = 1 - best_no_bid
gross_edge = best_yes_bid + best_no_bid - 1
These formulas create research candidates, never executable instructions.
python3.12 -m venv .venv
source .venv/bin/activate
python -m pip install -e ".[dev]"
pytest
ruff check .
PYTHONPATH=src python scripts/01_replay_orderbook_fixture.pyEDMN Trader sits beside Equity Factor Research, the factor-research sister project; it is not a runtime dependency. EDMN owns event-market normalization through paper/Demo evidence. Any future real-money system remains a separately reviewed private layer behind the evidence gate.
See Risk Policy, Real-Data Boundary, and Private Live Gate.
Specification · Roadmap · Visual overview · Stage ledger · Current handoff
Research and engineering infrastructure only. Not investment advice, trading advice, a profitability claim, or a production trading system.