ε EPSILON

Participate / no account required

Question a conclusion.
Leave something useful.

Complete a short interpretation exercise or tell us about a confusing step. No coding, trading account or market-data access required.

Choose your contribution

The recorded case

Case 001: SPY and QQQ, Momentum (2%), September 1, 2025–February 27, 2026. The rule requires positive net return in all five predefined perturbations. The baseline was already negative. These are the September 13 maintainer's reported historical results, not a new run or an independent recomputation.

Reported net total returns for fixed case 001
RunChanged inputNet return
BaselinePublished configuration−7.75%
Higher feeFee 0.0001 → 0.0005 (×5)−8.67%
Higher slippage$0.01 → $0.05 per share (×5)−7.89%
Shifted windowBoth dates +30 calendar days; overlaps baseline−8.35%
Narrower universeSPY only, instead of SPY + QQQ−7.22%
Joint stressFee ×5 + slippage ×5 + dates +30 days−9.24%

All five variations miss the positive-return threshold: 0/5 pass. Their overlapping dates and selected surviving ETFs limit what this one case can establish.

Read the original maintainer report ↗

A second run on this hosted site repeats the same pipeline. To independently recompute the result, another researcher would need to calculate it separately from equivalent source market data and compare inputs and outputs. Full case and reproduction limits ↗

Judge four statements

Use these meanings for every statement:

  • supported: The recorded evidence supports the statement as written.
  • contradicted: The recorded evidence or stated method directly conflicts with it.
  • not established: The evidence does not justify the statement, but does not directly disprove it.
  • unclear: The wording or available information is insufficient to decide.
1. The reported case failed its stated five-perturbation pass rule.
2. A profitable baseline became unprofitable after assumptions changed.
3. These results establish that momentum strategies generally do not work.
4. Another hosted run would establish independent recomputation from source data.

This is a guided interpretation activity, not an independent reproduction or a validated learning assessment.

Do not include names, emails, credentials, account details or restricted datasets. Only the project maintainer reviews submissions. Nothing is published automatically. Records remain stored for review; to request deletion, contact the maintainer through the repository and provide your receipt ID, not its secret link.

Prefer a public discussion? Use GitHub Discussions ↗