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, 174 XRP-gate60 long days) | Hyperliquid XRP funding cache for CARRY+BASIS; XRP daily bars from validation.common; _single_coin_decide on XRP gated by BTC ROC60, validated in reports/universe_generalization.md used to compute the XRP-gate60 target.
Diagnostics: corr(carry_z, basis_z) = +0.759 | fraction of long days with both signals stretched long (0.5x bucket): 0.4% | fraction with both signals stretched short (1.25x bucket): 1.0%.
| Variant | Days | LongDays | Ret | Sharpe | PF | MaxDD |
|---|
| baseline (gate60, no overlay) | 788 | 174 | +323.16% | 1.52 | 1.69 | -31.71% |
| primary overlay (0.5/1.0/1.25) | 788 | 174 | +336.16% | 1.56 | 1.70 | -31.71% |
| Bucket | Days | Share |
|---|
| 1.0x (no signal) | 149 | 85.6% |
| 1.0x (mixed) | 14 | 8.0% |
| 1.25x (both stretched short) | 8 | 4.6% |
| 0.5x (both stretched long) | 3 | 1.7% |
Bootstrap p(overlay beats baseline, full window): 0.358
Bootstrap p(IS < 2025-01-01): 0.589
Bootstrap p(OOS >= 2025-01-01): 0.001
| Segment | Variant | LongDays | Ret | Sharpe | PF |
|---|
| IS | baseline | 63 | +234.20% | 3.00 | 2.43 |
| IS | primary overlay | 63 | +224.83% | 2.98 | 2.40 |
| OOS | baseline | 111 | +26.62% | 0.58 | 1.22 |
| OOS | primary overlay | 111 | +34.27% | 0.68 | 1.27 |
| look (h) | carry z_enter | basis z_enter | LongDays | Ret | Sharpe | PF | MaxDD |
|---|
| 84 | 1.0 | 1.0 | 174 | +419.20% | 1.80 | 1.85 | -29.79% |
| 84 | 1.0 | 1.5 | 174 | +422.66% | 1.77 | 1.84 | -29.79% |
| 84 | 1.0 | 2.0 | 174 | +371.71% | 1.68 | 1.78 | -29.84% |
| 84 | 1.5 | 1.0 | 174 | +369.20% | 1.64 | 1.76 | -29.79% |
| 84 | 1.5 | 1.5 | 174 | +346.27% | 1.61 | 1.73 | -29.79% |
| 84 | 1.5 | 2.0 | 174 | +315.37% | 1.54 | 1.69 | -29.84% |
| 84 | 2.0 | 1.0 | 174 | +369.58% | 1.64 | 1.75 | -29.79% |
| 84 | 2.0 | 1.5 | 174 | +366.97% | 1.64 | 1.75 | -29.79% |
| 84 | 2.0 | 2.0 | 174 | +334.40% | 1.57 | 1.71 | -29.84% |
| 168 | 1.0 | 1.0 | 174 | +429.98% | 1.78 | 1.82 | -29.84% |
| 168 | 1.0 | 1.5 | 174 | +424.13% | 1.74 | 1.82 | -29.84% |
| 168 | 1.0 | 2.0 | 174 | +374.68% | 1.65 | 1.76 | -31.71% |
| 168 | 1.5 | 1.0 | 174 | +378.58% | 1.64 | 1.75 | -29.84% |
| 168 | 1.5 | 1.5 | 174 | +361.62% | 1.61 | 1.73 | -29.84% |
| 168 | 1.5 | 2.0 | 174 | +336.16% | 1.56 | 1.70 | -31.71% |
| 168 | 2.0 | 1.0 | 174 | +353.92% | 1.60 | 1.73 | -29.84% |
| 168 | 2.0 | 1.5 | 174 | +353.92% | 1.60 | 1.73 | -29.84% |
| 168 | 2.0 | 2.0 | 174 | +336.16% | 1.56 | 1.70 | -31.71% |
| 336 | 1.0 | 1.0 | 174 | +501.03% | 1.84 | 1.89 | -29.45% |
| 336 | 1.0 | 1.5 | 174 | +418.67% | 1.73 | 1.81 | -29.84% |
| 336 | 1.0 | 2.0 | 174 | +375.38% | 1.65 | 1.76 | -31.71% |
| 336 | 1.5 | 1.0 | 174 | +373.69% | 1.62 | 1.74 | -31.71% |
| 336 | 1.5 | 1.5 | 174 | +349.15% | 1.58 | 1.72 | -31.71% |
| 336 | 1.5 | 2.0 | 174 | +328.65% | 1.54 | 1.69 | -31.71% |
| 336 | 2.0 | 1.0 | 174 | +341.42% | 1.57 | 1.71 | -31.71% |
| 336 | 2.0 | 1.5 | 174 | +341.42% | 1.57 | 1.71 | -31.71% |
| 336 | 2.0 | 2.0 | 174 | +328.65% | 1.54 | 1.69 | -31.71% |
- IS baseline Sharpe: 3.00, ret +234.20%, maxDD -24.45%
- IS overlay Sharpe: 2.98, ret +224.83%, maxDD -24.45%
- OOS baseline Sharpe: 0.58, ret +26.62%, maxDD -31.71%
- OOS overlay Sharpe: 0.68, ret +34.27%, maxDD -31.71%
- Pre-registered pass: OOS Sharpe +>= 0.10 AND p < 0.05 AND maxDD not +> 5pp
- Result: REJECT (sharpe ++0.10 (need +0.10))
- 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.