Strategy Validation Properties (Exhaustive)

This is the canonical running list of properties that validation enforces for strategies.

Framework contract reference: docs/framework.md

Validation Execution Flow

flowchart LR
    A["Input strategy + date range"] --> B["Contract checks"]
    B --> C["Backtest metrics"]
    C --> D["Leakage + strict diagnostics"]
    D --> E["Threshold checks"]
    E --> F["Validation pass/fail summary"]

Validation Property Checklist

1. [ ] Allocation span config is valid

2. STACKSATS_ALLOCATION_SPAN_DAYS must be an integer in [90, 1460] (default 365).

3. [ ] Strategy cannot bypass framework orchestration

4. Custom compute_weights overrides are rejected.

5. [ ] Strategy must implement an allowed hook path

6. At least one is required: propose_weight(state) or build_target_profile(ctx, features_df, signals).

7. [ ] transform_features output type is valid

8. Must return a pandas DataFrame.

9. [ ] build_signals output type is valid

10. Must return dict[str, pandas.Series].

11. [ ] Signal and profile series shape/index contract holds

12. Each series must be a pandas Series, with no duplicate index, ascending index, exact match to window index.

13. [ ] Signal and profile series numeric validity holds

14. Series values must be finite numeric (no NaN, inf, or non-numeric coercion failures).

15. [ ] propose_weight outputs are finite

16. Every proposal must be finite numeric.

17. [ ] Target profile mode is valid

18. Only preference and absolute modes are allowed.

19. [ ] Allocation index monotonicity holds for temporal counting

20. Allocation index used for n_past must be monotonic increasing.

21. [ ] Locked-prefix structural validity holds

22. locked_weights must be 1D, finite, values in [0, 1], and length <= n_past.

23. [ ] Locked-prefix budget feasibility holds

24. Running sum of locked weights must never exceed total budget 1.0.

25. [ ] Locked-prefix daily bounds hold on contract-length windows

26. When window length equals configured span, locked values must be within [1e-5, 0.1].

27. [ ] Per-day clipping obeys future feasibility constraints

28. For contract-length windows, daily clipping enforces both day bounds and future-budget feasibility.

29. Infeasible bounds must hard-fail.

30. [ ] Output weight vector structure is valid

31. Final weights must be 1D and finite.

32. [ ] Output weight values are valid

33. No negative weights; no values above 1.0 (within tolerance).

34. [ ] Output weights sum exactly to budget (within tolerance)

35. Final weight sum must be 1.0.

36. [ ] Contract-length day bounds hold on final weights

37. For span-length windows, each day weight must be within [1e-5, 0.1].

38. [ ] Daily bounds are globally feasible for span length

39. Span must satisfy feasibility: n_days * min_daily_weight <= 1.0 <= n_days * max_daily_weight.

40. [ ] Historical lock immutability is enforced in strict checks

41. Injected locked prefix must be preserved exactly when recomputing under perturbed future features.

42. [ ] Forward-leakage resistance: masked-future invariance

43. Prefix weights at probe date must match when all future features are masked to NaN.

44. [ ] Forward-leakage resistance: perturbed-future invariance

45. Prefix weights at probe date must match when future features are strongly perturbed.

46. [ ] Profile-only leakage resistance is enforced

47. For profile-hook strategies (without propose hook), prefix profile values must be invariant under masked/perturbed future inputs.

48. [ ] Strict mode forbids in-place feature mutation

49. Strategy must not mutate ctx.features_df during weight computation.

50. [ ] Strict mode forbids profile-build feature mutation

51. Strategy must not mutate ctx.features_df during transform/signal/profile build path.

52. [ ] Strict mode determinism holds

53. Repeated runs with identical inputs must produce exactly matching weights (within atol=1e-12).

54. [ ] Weight constraints hold across validation windows

55. Validation windows must not violate sum, negativity, or (when applicable) min/max day bounds.

56. [ ] Boundary saturation stays below strict threshold

57. In strict mode, boundary-hit rate (days at MIN or MAX) must be <= max_boundary_hit_rate (default 0.85).

58. [ ] Cross-fold robustness minimum is met in strict mode (when fold checks run)

59. Minimum fold win rate must be >= min_fold_win_rate (default 20.0).

60. Fold checks are skipped when there is insufficient date range for at least two valid folds.

61. [ ] Cross-fold instability is bounded in strict mode (when fold checks run)

62. Fold win-rate standard deviation must be <= max_fold_win_rate_std (default 35.0).

63. Fold checks are skipped when there is insufficient date range for at least two valid folds.

64. [ ] Shuffled-null robustness threshold is met in strict mode (when shuffled checks run)

65. Mean win rate on shuffled-price trials must be <= max_shuffled_win_rate (default 80.0) across shuffled_trials (default 3).

66. Shuffled checks are skipped when PriceUSD_coinmetrics is missing, the shuffled window is empty, or shuffled_trials <= 0.

67. [ ] Global win-rate threshold is met

68. Validation backtest win rate must be >= min_win_rate (default 50.0).

69. [ ] Validation date range must contain data

70. Empty requested validation range is an automatic validation failure.

71. [ ] Backtest path must generate windows

72. Validation relies on backtest windows; if none are generated, validation cannot pass.