H-007: BTC out of the market when 7-day funding is above its 1-year 80th percentile
Fail Family: flows · Status: rejected
Do high funding rates warn of Bitcoin drops? The idea did not beat simply buying the coin and holding it. It is rejected.
Result
oos: wins 0/1 (need 1), beats buy-and-hold 0/1
| Asset | Period | Sharpe | Hold Sharpe | CAGR | Hold CAGR | Max drawdown | Hold drawdown | In market | p-value | Beats hold | Counts as win |
|---|---|---|---|---|---|---|---|---|---|---|---|
| BTCUSDT | oos (2022-01-01 → 2026-10-13) | 0.24 | 0.50 | +0.2% | +13.1% | -66.9% | -66.9% | 83% | 0.849 | no | no |
Equity curves
Growth of 1 unit, log scale, fees included. Only days on which the rule is fully defined are counted.
The same rule on every coin
Exploratory check of this rule on all coins since 2022-01-01. It is not part of the verdict above.
- ✓works (passed the chance check)
- ↑better than holding, could be luck
- ✗worse than holding
- ·not enough data
- original coin of the hypothesis
| Coin | ✓ | Result | Sharpe: rule / holding | CAGR | From |
|---|---|---|---|---|---|
| BTC | ✗ | worse than holding | 0.23 / 0.48 | -0% / +12% | 2022-01-01 |
| ETH | ✗ | worse than holding | 0.01 / 0.20 | -18% / -9% | 2022-01-01 |
| SOL | ✗ | worse than holding | -0.10 / 0.35 | -35% / -10% | 2022-01-01 |
| BNB | ✗ | worse than holding | 0.21 / 0.42 | -2% / +8% | 2022-01-01 |
| XRP | ✗ | worse than holding | 0.29 / 0.51 | -4% / +10% | 2022-01-01 |
| ADA | ✗ | worse than holding | -0.24 / -0.01 | -39% / -32% | 2022-01-01 |
| LTC | ✗ | worse than holding | -0.28 / 0.11 | -32% / -18% | 2022-01-01 |
| LINK | ✗ | worse than holding | 0.10 / 0.30 | -20% / -10% | 2022-01-01 |
| DOGE | ✗ | worse than holding | 0.09 / 0.26 | -22% / -15% | 2022-01-01 |
| XLM | ✗ | worse than holding | 0.19 / 0.30 | -12% / -9% | 2022-01-01 |
| TRX | ✗ | worse than holding | 0.69 / 0.74 | +22% / +36% | 2022-01-01 |
| EVER | ✗ | worse than holding | -0.37 / -0.15 | -68% / -68% | 2022-03-19 |
| GRAM | ✗ | worse than holding | -0.33 / 0.13 | -35% / -17% | 2023-03-14 |
Why it might work
Crowded leveraged longs precede liquidations. Ported from F1MEBOT PREREG.md (H2).
Verdict history
- 2026-09-30: Fail (F1MEBOT (2026))
- 2026-10-14: Fail (this engine, data to 2026-10-13)
Pre-registered specification
kind: timing_rule
rule:
type: feature_vs_rolling_quantile
feature: funding_btc_7d
q: 0.8
long_when: not_above
lookback: 365
min_periods: 180
lag: 1
mode: long_flat
fee: 0.001
universe:
- BTCUSDT
periods:
- name: oos
start: '2022-01-01'
significance:
method: circular_shift
n: 2000
alpha: 0.0125
criterion:
min_wins:
oos: 1
require_significance: true