Graveyard / Positioning / Binance perps
Following whale positioning
Does tracking whale long/short ratios predict Bitcoin?
The idea
Binance publishes two long/short ratios for its top traders, one weighted by position size and one by account count. When the big accounts lean one way and the many lean the other, follow the big money.
How we tested it
| Data | Binance BTCUSDT perpetual, 5-minute top-trader ratios. Entry when the divergence z-score passes a threshold for two bars in a row. |
|---|---|
| Period | 21 June 2024 to 22 June 2026. |
| Costs | 0.11% per round trip. |
| Holdout | Results were split by regime: bull, bear, sideways. |
| Control | No random-entry control was run for this test. |
Result
| Correlation with next move | 0.034 |
|---|---|
| Same desk's working sentiment signal | 0.13 to 0.21 |
| Regimes with a profit | 0 of 3 |
Why it failed
The split between big and many carries no direction. Weighting by size washes out the signal that the plain ratio has. The divergence also changes slowly, so most signals repeat the previous one.
What would make it worth retesting
No. Every window and threshold tested came out flat.
Test id: whale-btcusdt-event-v1. Past results of a backtest say nothing certain about the future.