Pre-registered proposal: reports/research_proposal_carry_basis_overlay.md. Defaults: look=168h, carry|z|>=1.5, basis|z|>=2.0, multipliers {0.5, 1.0, 1.25}. Window: 2024-06-04 .. 2026-07-31 (788 days, 139 ADA-gate60 long days) | Hyperliquid ADA funding cache for CARRY+BASIS; ADA daily bars from validation.common; _single_coin_decide on ADA gated by BTC ROC60, validated in reports/universe_generalization.md used to compute the ADA-gate60 target.
Diagnostics: corr(carry_z, basis_z) = +0.787 | fraction of long days with both signals stretched long (0.5x bucket): 0.4% | fraction with both signals stretched short (1.25x bucket): 0.4%.
| Variant | Days | LongDays | Ret | Sharpe | PF | MaxDD |
|---|
| baseline (gate60, no overlay) | 788 | 139 | +70.01% | 0.75 | 1.30 | -43.69% |
| primary overlay (0.5/1.0/1.25) | 788 | 139 | +120.31% | 1.00 | 1.42 | -33.80% |
| Bucket | Days | Share |
|---|
| 1.0x (no signal) | 110 | 79.1% |
| 1.0x (mixed) | 23 | 16.5% |
| 1.25x (both stretched short) | 3 | 2.2% |
| 0.5x (both stretched long) | 3 | 2.2% |
Bootstrap p(overlay beats baseline, full window): 0.004
Bootstrap p(IS < 2025-01-01): 0.129
Bootstrap p(OOS >= 2025-01-01): 0.015
| Segment | Variant | LongDays | Ret | Sharpe | PF |
|---|
| IS | baseline | 54 | +107.58% | 2.15 | 1.92 |
| IS | primary overlay | 54 | +120.90% | 2.21 | 2.00 |
| OOS | baseline | 85 | -18.10% | -0.15 | 0.95 |
| OOS | primary overlay | 85 | -0.27% | 0.16 | 1.06 |
| look (h) | carry z_enter | basis z_enter | LongDays | Ret | Sharpe | PF | MaxDD |
|---|
| 84 | 1.0 | 1.0 | 139 | +177.53% | 1.19 | 1.56 | -29.08% |
| 84 | 1.0 | 1.5 | 139 | +155.38% | 1.12 | 1.51 | -29.43% |
| 84 | 1.0 | 2.0 | 139 | +124.88% | 1.03 | 1.43 | -30.65% |
| 84 | 1.5 | 1.0 | 139 | +146.06% | 1.10 | 1.49 | -29.08% |
| 84 | 1.5 | 1.5 | 139 | +135.64% | 1.06 | 1.46 | -31.12% |
| 84 | 1.5 | 2.0 | 139 | +120.85% | 1.02 | 1.43 | -31.90% |
| 84 | 2.0 | 1.0 | 139 | +132.54% | 1.05 | 1.46 | -32.02% |
| 84 | 2.0 | 1.5 | 139 | +135.64% | 1.06 | 1.46 | -31.12% |
| 84 | 2.0 | 2.0 | 139 | +120.85% | 1.02 | 1.43 | -31.90% |
| 168 | 1.0 | 1.0 | 139 | +152.52% | 1.11 | 1.50 | -29.38% |
| 168 | 1.0 | 1.5 | 139 | +135.40% | 1.06 | 1.46 | -31.19% |
| 168 | 1.0 | 2.0 | 139 | +120.31% | 1.00 | 1.42 | -33.80% |
| 168 | 1.5 | 1.0 | 139 | +147.21% | 1.11 | 1.49 | -29.08% |
| 168 | 1.5 | 1.5 | 139 | +134.48% | 1.05 | 1.46 | -31.46% |
| 168 | 1.5 | 2.0 | 139 | +120.31% | 1.00 | 1.42 | -33.80% |
| 168 | 2.0 | 1.0 | 139 | +140.04% | 1.08 | 1.48 | -27.97% |
| 168 | 2.0 | 1.5 | 139 | +127.86% | 1.03 | 1.45 | -31.53% |
| 168 | 2.0 | 2.0 | 139 | +120.31% | 1.00 | 1.42 | -33.80% |
| 336 | 1.0 | 1.0 | 139 | +150.45% | 1.10 | 1.49 | -32.64% |
| 336 | 1.0 | 1.5 | 139 | +140.75% | 1.06 | 1.48 | -32.00% |
| 336 | 1.0 | 2.0 | 139 | +129.82% | 1.04 | 1.45 | -30.90% |
| 336 | 1.5 | 1.0 | 139 | +134.73% | 1.06 | 1.46 | -31.21% |
| 336 | 1.5 | 1.5 | 139 | +130.81% | 1.04 | 1.45 | -31.21% |
| 336 | 1.5 | 2.0 | 139 | +130.50% | 1.04 | 1.45 | -30.04% |
| 336 | 2.0 | 1.0 | 139 | +129.27% | 1.04 | 1.45 | -31.21% |
| 336 | 2.0 | 1.5 | 139 | +129.27% | 1.04 | 1.45 | -31.21% |
| 336 | 2.0 | 2.0 | 139 | +130.50% | 1.04 | 1.45 | -30.04% |
- IS baseline Sharpe: 2.15, ret +107.58%, maxDD -31.25%
- IS overlay Sharpe: 2.21, ret +120.90%, maxDD -31.25%
- OOS baseline Sharpe: -0.15, ret -18.10%, maxDD -38.94%
- OOS overlay Sharpe: 0.16, ret -0.27%, maxDD -29.08%
- Pre-registered pass: OOS Sharpe +>= 0.10 AND p < 0.05 AND maxDD not +> 5pp
- Result: PASS
- Inert signal risk: if <5% of long days trigger either 0.5x or 1.25x, the overlay is a no-op and the experiment is a wash. Reported as INERT in the verdict.
- Threshold tuning: the proposal pre-registers look=168h, carry_enter=1.5, basis_enter=2.0. The robustness grid is REPORTED, not selected. Picking the IS winner would be the look-ahead bug class that bit R10-13.
- Hyperliquid only: Binance/Bybit/OKX funding are geo-blocked from this IP (451/403). Hyperliquid started in 2023; the cache is the longest available free source. Binance USDT-M funding IS in the metrics cache but is perpetual-rate only, not the premium index; the BASIS leg specifically needs premium.
- Hyperliquid is one venue: cross-venue funding arbitrage (HL vs Binance) is not what this study tests; this study asks whether HL alone is informationally useful as a sizing overlay on a slow-trend signal.
- CARRY and BASIS are correlated: corr reported above. If corr > 0.7, the composite is one-dimensional and the two-signal framing is misleading; the overlay is then a single-signal gate with two thresholds.
- Multipliers are heuristic: 0.5x and 1.25x are the natural ±50% / ±25% asymmetry, not the result of an optimization. An optimization step would re-introduce the threshold-tuning look-ahead; the honest path is to report the asymmetry you chose and let the verdict be what it is.