メインコンテンツまでスキップ

Experiment Template

0. Metadata​

  • Strategy name:
  • Owner:
  • Date:
  • Code version (commit):
  • Dataset range:

1. Hypothesis​

  • What market behavior is targeted?
  • Why should this produce alpha after exposure matching?

2. Event Inputs​

  • Consumed event types:
  • Coalescing window:
  • Cooldown:

3. Feature Set​

  • Required core features:
  • Additional custom features:
  • Feature latency constraints:

4. Decision Logic​

  • Target position mapping:
  • Entry conditions:
  • Exit conditions:
  • Emergency flatten conditions:

5. Risk Constraints​

  • max_abs_position_eth:
  • max_order_rate_per_sec:
  • spread_guard_bps:
  • kill-switch thresholds:

6. Evaluation Plan​

  • Backtest windows:
  • Replay windows:
  • Cost/fill assumptions:
  • Primary metrics:
  • Secondary diagnostics:

7. Results​

  • Alpha Sharpe:
  • Alpha MaxDD:
  • Turnover:
  • Fill rate:
  • Adverse selection:

8. Failure Analysis​

  • Where did it lose?
  • Market regimes where it degraded:
  • Operational risks observed:

9. Decision​

  • Promote to replay
  • Promote to paper
  • Reject
  • Rework

10. Next Action​

  • Single concrete change for the next iteration: