BtcCrossVenueFundingSpreadArbLongHlShortBinance1H

BTC Cross-Exchange Funding Rate Spread Arbitrage: Long Hyperliquid Perp + Short Binance Perp When Funding Spread > 0.025%/8h

A delta-neutral cross-venue funding-spread arbitrage that captures structural differences between Binance and Hyperliquid perpetual funding rates on BTC. When Binance 8h funding exceeds Hyperliquid 8h funding by ≥ 0.025% (annualized ~27%), simultaneously go LONG BTCUSDT on Hyperliquid (receive higher funding from the longs that are crowded on the cheaper venue) and SHORT BTCUSDT on Binance (collect the elevated funding the Binance longs are paying). The two perp legs cancel net delta to <1%, so the trade is directionally hedged — P&L comes purely from the funding-rate differential plus minor basis-convergence dynamics. This fills FIVE under-represented portfolio buckets simultaneously: cross-venue (1.4% → boosts toward 15% target), Hyperliquid (7.5% → toward 20%), long-short direction (13.5% → toward 45%), non-Binance dominance (Binance is 86.1% — this trade reduces single-venue concentration), and mid-term-horizon market-neutral. Critically the mechanism is FUNDAMENTALLY different from every failure this session: it does NOT rely on OHLCV pattern recognition, it does NOT depend on BTC's 2024-2026 microstructure regime (which has decayed every directional BTC strategy), and it has only TWO core parameters (entry-spread threshold, exit-spread threshold) which makes walk-forward overfitting structurally impossible. Position sizing: 25% of equity per leg (50% total margin utilization across both exchanges). Risk per trade: ~1.5% of equity in worst-case basis-divergence + funding-flip scenario.

Directly implements the hypothesis: long Hyperliquid + short Binance when the Binance-minus-Hyperliquid 8h funding spread exceeds 0.025%/8h (~27% APR). The signal is the raw per-8h spread in percent (continuous, recomputed every bar) and is thresholded in should_enter against the matching-unit entry_spread_pct, so the comparison units are consistent. Binance is the primary leg so the backtest's funding_rates feed resolves to Binance USD-M and hl_funding_rates resolves to Hyperliquid by coin, giving both legs correct funding. Hyperliquid's hourly cadence is normalized to an 8h-equivalent via hl_period_scale=8 so the two venues are compared on the same footing the hypothesis specifies. Equal-notional two-leg sizing at 25% equity per leg (50% total margin) makes the pair delta-neutral so P&L comes from the funding differential, not BTC direction; the atomic-hedge entry prevents naked-leg exposure. Only two core edge parameters (entry/exit spread) keep walk-forward overfitting structurally unlikely. Leverage left at 1.0 and unused in sizing, so the leverage gate is satisfied. The strategy fills the cross-venue, Hyperliquid, long-short, non-Binance-dominance, and mid-term market-neutral buckets, and its funding-mechanism edge is independent of OHLCV pattern recognition and BTC's directional regime.

Cross-venue BTC funding-spread arb is structurally fee-dominated and signal-starved, not an optimization candidate. The Binance-vs-Hyperliquid BTC 8h funding differential exceeds the 0.025%/8h entry threshold only 9 times across 56,136 bars (both venues are deeply liquid and tightly arbitraged on BTC), and all 9 events lose: PF 0.80, expectancy -$27.7, Sharpe -13.7, avg_win ($244) ≈ avg_loss ($245). The economics cannot clear: a 4-leg round trip costs ~0.19%+ (Binance ~0.10% RT + HL ~0.09% RT), but the ~9.9h convergence-exit hold accrues only ~0.03% of the targeted differential, so commission is 21.8% of gross and every cycle bleeds. This is NOT the 'engine doesn't credit funding' false reason (funding is credited and both legs read correctly) and NOT a zero-trade wiring bug (9 trades fired) — it is a genuine fee/sparsity wall specific to BTC. Only 9 events is also far too thin to optimize two thresholds without fitting noise, and lowering the threshold harvests more noise at the same fee disadvantage. The mechanism is sound but the target is wrong: funding arb needs an instrument with persistent, large funding skew (e.g. a high-funding altcoin perp or a venue pair with a structural funding gap), not BTC where the cross-venue spread is tiny and rare. Recommend the Research Lead reframe onto a high-funding-skew instrument and require a hold-through-multiple-funding-settlements exit so accrued funding can exceed the 4-leg fee load.

Delta-neutral cross-exchange funding-rate spread arbitrage on BTC perpetuals. Computes the per-8h funding differential between Binance USD-M (8h funding, read from supplementary funding_rates with Binance as primary leg) and Hyperliquid (hourly funding scaled x8 to an 8h footing, read from hl_funding_rates). When Binance funds richer than HL by >= 0.025%/8h it SHORTS Binance and goes LONG Hyperliquid (equal notional, ~delta-neutral); when HL funds richer the legs flip (long Binance / short HL). Both legs are entered atomically on the same bar after validating markets, harvesting the funding differential plus minor basis convergence while BTC price risk cancels. Exits when the differential collapses inside the exit band or after a max hold. Fixed 25%-of-equity notional per leg, leverage 1.0, no price-only fallback.

Mechanism is structurally sound: delta-neutral two-leg hedge, atomic both-legs-validated entry, low parameter count, funding correctly read on both venues (funding_rates + hl_funding_rates keys), no price-only fallback edge

Genuinely diversifying intent (cross-venue, market-neutral, Hyperliquid exposure)

Only 9 entry events across 56,136 bars — the BTC Binance-vs-HL funding spread almost never exceeds the 0.025%/8h threshold because both venues are tightly arbitraged; far too few trades to optimize without fitting noise

Every trade loses: profit_factor 0.80, expectancy -$27.7, Sharpe -13.7, avg_win ≈ avg_loss (noise + negative fee drift)

Fee-dominated: 4-leg round trip ~0.19%+ vs ~0.03% funding accrued over the ~9.9h avg hold; commission is 21.8% of gross

Convergence exit closes positions in ~10h, so little funding accumulates to offset the entry/exit fees on two legs across two venues

positive

BtcCrossVenueFundingSpreadArbLongHlShortBinance1H harvested the Binance-vs-Hyperliquid BTC funding differential as a delta-neutral two-leg pair, shorting the richer venue and longing the cheaper, on a deliberately minimal two-threshold design. The implementation was clean — atomic hedged entry, funding read correctly on both venues, no price-only fallback — but the spread cleared the entry threshold only 9 times in 56,136 bars and every cycle lost, with commission eating 21.8% of gross because the ~10h convergence hold accrued far less than the four-leg round-trip cost. The analyst ruled it a structural fee/sparsity wall specific to BTC rather than a code or crediting bug, too thin to optimize, and recommended reframing onto a persistently high-funding instrument with a longer hold. It ended after one iteration as abandoned, never advancing to optimization or risk review.

The mechanism is sound and funding is credited correctly, but BTC is the wrong target: the cross-venue funding spread rarely clears its threshold and the four-leg round-trip fee swamps the thin carry — funding arb needs an instrument with persistent, large funding skew (a high-funding altcoin perp or a structurally gapped venue pair) and a hold-through-multiple-settlements exit so accrued funding can exceed the fee load.

It was abandoned at the pre-optimization backtest-review gate (verdict: abandon) as a structurally fee-dominated, signal-starved trade — the BTC cross-venue funding spread is tiny and rare because both venues are deeply liquid and tightly arbitraged, so 9 events is far too thin to optimize and lowering the threshold only harvests more noise at the same fee disadvantage — and optimization plus all later stages were never reached.

A delta-neutral cross-venue funding-spread arbitrage on BTC perps — when the Binance USD-M 8h funding exceeded the Hyperliquid 8h-equivalent funding by ≥0.025%/8h, shorting the richer venue and longing the cheaper one in equal notional (and flipping symmetrically when HL funded richer) to harvest the funding differential while staying price-hedged, on just two thresholds with atomic both-legs-validated entry.

The spread cleared the 0.025%/8h threshold only 9 times across 56,136 bars (18 leg-trades), and every event lost: profit factor 0.80, expectancy -$27.7/trade, Sharpe -13.7 (CI [-23.4, -8.5]), win rate 44%, avg_win ($244) ≈ avg_loss ($245), with commission at 21.8% of gross — a ~0.19% four-leg round trip against only ~0.03% of differential accrued over the ~9.9h convergence hold.