← Research

Why an edge should exist / hypothesis

Funding inefficiency around settlement should, in theory, yield several tradable patterns.

Construction

7 mechanics (mean-reversion, settlement arb, dump, spot-hedge, ADL front-run, time-arb, direction-signal) on 10 symbols.

Test data

Bybit OHLC + funding, 10 symbols, 6 months, cost model ~16 bps/trade.

Results

0/7. S6 (mean-reversion, Sharpe 1.53) = hidden directional long bias (144/144 LONG, Feb −35%). S5 (settlement arb) WR=0%.

Validation

BTC: 0 events >2bps in 6 months — majors are efficient.

Root cause

The funding edge lives only on illiquid microcaps (which Funding Capture already harvests via execution); on majors the market is efficient.

Verdict

Dead; confirmed the boundary — funding edge exists only on illiquids.

Source

scripts/research/exploit_7_strategies.py · RESEARCH.md 2026-04-22