BTC Cross-Venue Funding Rate Differential Carry (Binance vs Hyperliquid, 1H Monitoring, Market-Neutral)

A delta-neutral CROSS-VENUE funding-rate-differential carry strategy on BTC perpetuals. Continuously monitors the normalized 8h-equivalent funding rate differential between BTCUSDT.BINANCE (funding accrues every 8h) and BTCUSD.HYPERLIQUID (funding accrues hourly). When the differential exceeds entry_threshold, opens opposing equal-dollar-notional positions: short the high-funding venue + long the low-funding venue, collecting the funding spread while remaining net-zero on BTC price exposure. Closes when differential mean-reverts. This is a structurally different mechanism from BtcInterExchangeBasisMR (which trades the PRICE basis between venues) — this trades the FUNDING RATE differential, which is a distinct mechanically-driven cash flow. Hyperliquid funding rates are documented as substantially more volatile than Binance funding because: (1) HL's funding rate has a higher cap (1.0% vs 0.75%) and recalculates hourly versus Binance's 8h cycle, (2) HL has thinner depth and retail-heavy flow producing more extreme funding regime shifts during sentiment swings, (3) HL's predicted_fundings stream provides forward-looking divergence signals. When extreme HL funding diverges from Binance, the spread reliably mean-reverts over 12-36h as cross-venue arbitrageurs close the gap. FILLS MULTIPLE UNDER-REPRESENTED PORTFOLIO BUCKETS: HYPERLIQUID (4.8% vs ≥20% target), cross-venue (6% vs ≥15% target), long_short market-neutral (12.3% vs ≤55% long-only target). KEY DESIGN DISCIPLINE LEARNED FROM RECENT FAILURES: only 4 hypothesis-declared tunable parameters, every other internal constant hard-coded to prevent parameter bloat. Symmetric reward/risk via small fixed stop on basis divergence (no asymmetric TP<SL). Mechanism class proven structurally sound (funding carry is one of the few mechanically-guaranteed cash flows in crypto), distinct from the failed mechanism classes this session (BB-MR, volume-confirmed breakout, multi-week trend following, drawdown accumulation cross-asset transfer, volume-surge momentum continuation).

Implements the hypothesis's cross-venue funding-rate-differential carry (distinct from the price-basis trade). I verified the data this session: HL BTC funding (hl_funding_rates) is hourly with history 2025-03→2026-05; Binance BTCUSDT funding (funding_rates) is 8h with history 2020→2026-04; HL BTCUSD 1-HOUR bars span 2025-10→2026-05, fully inside the dual-funding-coverage window — so the backtest runs where BOTH funding legs exist, avoiding the empty-HL-funding window that causes a 'only 1 entry / non-evaluable' result. From the real overlap I measured the 8h-equivalent differential: mean|diff|≈0.7bps, with |diff|≥0.0002 (the entry default) occurring ~4% of hourly bars — enough distinct mean-reversion episodes to be evaluable. Crucially I use the economically correct carry polarity: carry of (long A + short B) per period = rate_B − rate_A, so the strategy SHORTS the higher-funding venue (receives the rich funding) and LONGS the lower one; QA checks signal polarity, so the sign is what makes the carry positive rather than negative. Both legs are MARGIN futures venues, enabling genuine long+short and delta-neutrality via equal dollar notional. The funding-series parsers are module-level functions (not methods) so the Layer-2 proxy's MethodType rebinding can't strip their arity, and uses hl_funding_rates (historical) rather than hl_predicted_fundings (a forward snapshot with no history that would void all backtest trades). Two-leg execution overrides _submit_entry/_submit_exit (not on_bar), preserving the base class's correct position-lookup and warmup handling. Exactly 4 tunables with all operational constants hard-coded to prevent parameter bloat, and a symmetric combined-PnL stop (no asymmetric TP<SL). Leverage stays 1.0 (per-leg 0.30 → gross 0.60 < 1×) and is referenced consistently in sizing, avoiding the leverage_set_but_unused gate. A generous 5-day max hold lets the thin carry accumulate against the ~19bps round-trip taker cost; built-in coverage/differential diagnostic logs let the analyst confirm net fee viability.

Edge structurally below trading costs on BTC — NOT a funding-creditability issue (funding is credited: +$90.4 collected, visible in trade records). The cross-venue funding differential captures only ~$5.32/pair while the two-leg taker round-trip costs ~$57.12/pair (~11x), and the legs fail to hedge (total price_pnl -$1,279.7, a consistent one-way leak that also explains the 81.7% drawdown on a 'neutral' book). Net -$1,189 (+90 funding -971 commission -1,280 leakage); even a perfect hedge still loses -$881. Optimizing the 4 params cannot close an 11x fee-to-edge gap — raising entry_diff_8h to a fee-beating level would leave ~0 trades because BTC's cross-venue funding differential is kept tight by arbitrage. NOTE FOR RESEARCH LEAD: the funding-carry mechanism class is valid and credited; the failure is the TARGET, not the premise. To retry, reframe onto a high-funding ALTCOIN perp with persistent, large venue funding skew (where the 8h-equivalent differential routinely exceeds the round-trip fee), with far fewer round-trips (longer minimum holds so accrued carry >> fees) and ideally maker entries to cut the cost base. As-is on BTC the instance is dead.

Market-neutral cross-venue funding-rate-differential carry on BTC: equal-dollar-notional opposing perp positions on Binance USD-M (BTCUSDT.BINANCE, 8h funding) and Hyperliquid (BTCUSD.HYPERLIQUID, hourly funding), monitored on 1-HOUR bars. Each bar it reads both historical funding series from supplementary_data (funding_rates for Binance, hl_funding_rates for Hyperliquid — both flat {ns:rate} dicts), normalizes to an 8h-equivalent (binance_8h = rate; hl_8h = hl_rate*8), and computes diff_8h = binance_8h − hl_8h. When diff_8h ≥ +entry_diff_8h (Binance richer) it SHORTS Binance + LONGS HL; when diff_8h ≤ −entry_diff_8h (HL richer) it LONGS Binance + SHORTS HL — always shorting the higher-funding venue so the collected funding carry is positive while staying delta-neutral on BTC price. It holds (collecting the spread) until the differential mean-reverts below exit_diff_8h (after a min hold), a 5-day time stop, or a symmetric combined-PnL stop. If either funding series is missing at a bar it goes flat (no price-proxy fallback). 4 tunables: entry_diff_8h, exit_diff_8h, per_leg_pct, stop_loss_pct.