BtcCrossVenueFundingCarryLongBinanceSpotShortHlPerpNeutral

Cross-Venue Delta-Neutral Funding Capture: Long Binance Spot BTC / Short Hyperliquid BTC Perp on Extreme HL Funding (Hedged Carry, Market-Neutral)

A MARKET-NEUTRAL, CROSS-VENUE hedged funding-carry strategy that harvests Hyperliquid's perpetual funding premium while staying delta-neutral via a Binance spot long. Hyperliquid funds HOURLY and its perp funding regularly runs richer and more persistent than Binance USD-M (newer venue, directional retail flow, thinner arb), so its positive-funding extremes are a less-arbitraged cash flow than the Binance-majors dispersion that the prior cross-sectional carry book found decayed to ~0. This is the analyst-suggested reframe of the funding-carry mechanism: same mechanism, moved off decayed Binance majors onto a venue that genuinely carries persistent funding skew, and run as a SINGLE-PAIR threshold-gated trade rather than cross-sectional rotation (which produced the COVID-era artifacts). It simultaneously fills three of the most under-represented portfolio buckets — HYPERLIQUID (6.2% vs 20% target), cross-venue (5.2% vs 15%), and long/short-market-neutral direction (14.4% vs target) — while avoiding every documented landmine: it uses only bars + funding (both of which exist and are clean for BTC on both venues), NOT the HL OI/liquidation/tick microstructure data that does not exist historically; it is NOT a pure-OHLCV breakout/fade (the failed HL family); and it is NOT a Deribit rolling-option (broken harness). Primary leg: BTCUSDT.BINANCE_SPOT (long, CASH, allowed). Hedge leg: BTCUSD.HYPERLIQUID perp (short, receives funding when funding>0). Net delta ~0, so PnL = funding collected + any spot-vs-perp basis convergence, minus fees. Only 3-4 parameters to keep overfitting risk low.

Harvests Hyperliquid perpetual funding, which regularly runs richer and more persistent than Binance USD-M (newer venue, directional retail flow, thinner arb capital, HOURLY funding), so its positive-funding extremes are a less-arbitraged cash flow than the Binance-majors dispersion that the prior cross-sectional carry book found decayed to ~0. This is the analyst-suggested reframe of the funding-carry mechanism: same mechanism, moved off decayed Binance majors onto a venue that genuinely carries persistent funding skew, and run as a SINGLE-PAIR threshold-gated trade rather than cross-sectional rotation (which produced COVID-era artifacts). Leg assignment respects venue constraints: Binance spot is CASH/long-only so it is the BUY-only primary, the shortable HL perp is the hedge; equal USD notional makes the book delta-neutral on BTC price, leaving funding accrual (plus minor basis convergence) as the PnL source. Fills three of the most under-represented buckets at once: HYPERLIQUID, cross-venue, and market-neutral direction. Uses only bars + funding (both clean for BTC on both venues), NOT HL OI/liquidation/tick microstructure (which lacks history), NOT a pure-OHLCV breakout/fade, NOT a Deribit rolling option. 3 edge parameters keep overfitting risk low. leverage=1.0 (spot cap) and sizing uses capital_frac not leverage, so no unused-leverage gate triggers.

Three independent, optimization-unfixable failures on the BTC instance. (1) NO OPTIMIZATION BASIS: only 2 entry cycles fired in 8 years (entry_signaled=2) — smoothed HL BTC funding essentially never clears funding_entry=3e-5/hr; the optimizer would fit to 2 trades = noise. (2) FEES STRUCTURALLY DOMINATE THE CARRY (decisive): trade-level detail shows per-cycle commission ~$130 against funding collected of only +$6.38 (cycle 1) and +$67.68 (cycle 2) — even the healthy cycle's carry loses to its fees, commission_pct_of_gross 77%. Lowering the threshold to add trades yields more, SMALLER carries against the same fixed Binance-spot (0.10% RT) + HL-perp (0.09% RT) + impact, so density worsens the fee ratio rather than fixing it; BTC HL funding is too thin to clear a two-venue round-trip. (3) HEDGE NOT NEUTRAL: equal-quantity legs (0.398/0.398) with identical entry/exit times show non-offsetting price-PnLs (short +$309 vs long -$514 in cycle 1; -$97 vs +$22 in cycle 2) — a symptom of cross-venue leg misalignment (mechanism not independently verified) that means the 'market-neutral' book carries directional risk. This is the documented funding-carry fee-fragility pattern compounded by a second venue's fees: the analyst-suggested reframe onto HL does not rescue it because BTC's HL funding, even at its triggering extremes, does not exceed the cross-venue cost. REDIRECT (new hypothesis for Research Lead, not an optimization of this one): the only way this mechanism clears costs is a HIGHER-FUNDING HL ALT perp where per-cycle carry >> two-leg fees AND the funding distribution produces 50+ cycles — and the hedge sizing/pricing must be corrected so the legs actually offset before any carry can be judged.

Market-neutral cross-venue hedged funding carry on BTC. Long leg = BTCUSDT.BINANCE_SPOT (CASH, long-only, primary); hedge leg = BTCUSD.HYPERLIQUID perp (extra). The signal is the smoothed historical Hyperliquid BTC perp funding rate, read from the hl_funding_rates supplementary series (flat {ns: rate}; HL funds HOURLY) at each 1-HOUR bar timestamp and averaged over the last smooth_window readings -- a continuous, time-varying carry signal independent of price. When smoothed HL funding >= funding_entry the strategy opens the hedged book (BUY Binance spot + SELL equal-USD-notional Hyperliquid perp), so the perp short receives funding each hour while the spot long neutralizes BTC price exposure; net delta ~0 so PnL = funding collected + any spot-vs-perp basis convergence minus fees. It flattens BOTH legs when smoothed funding decays below funding_exit, with frozen infra backstops (max-hold and a combined-PnL kill switch). Single-pair, threshold-gated -- NOT cross-sectional rotation. 3 edge parameters (funding_entry, funding_exit, capital_frac); smoothing window and backstops frozen. If the HL funding series is absent the signal stays 0 and no position is ever opened -- no price-pattern fallback.

Uses the correct supplementary key (hl_funding_rates) — data loads, signal fired (not the wrong-key zero-trade bug); the cross-venue plumbing and two-leg order submission work.

Targets genuinely under-represented portfolio buckets (Hyperliquid, cross-venue, market-neutral) and is a clean single-pair threshold design (only 3 edge params).

Only 2 entries (4 legs) across 2018-2026 — entry_signaled=2 over 5316 evaluated bars; far too few for optimization to be anything but noise-fitting.

Fees structurally dominate the BTC HL carry: per-cycle commission ~$130 vs funding collected of only +$6.38 (cycle 1) and +$67.68 (cycle 2); commission_pct_of_gross 77.15%. Even the best cycle's funding loses to its fees, and lowering funding_entry produces more, smaller sub-fee carries — worse, not better.

The hedge is not actually delta-neutral: equal-quantity legs with identical entry/exit times show non-offsetting price-PnLs (+$309/-$514 and -$97/+$22), so the book carries unhedged directional risk and/or cross-venue leg misalignment.

Net result negative (-9.3% total, Sharpe -6.07, PF 0.62, expectancy -$51.5/trade); metrics_reliable=false on a 2-trade sample.

carry $6-68 vs commission $130/cycle; 77% of gross

carry >> two-venue RT fees

leg price-PnLs don't offset (+309/-514)

legs ~cancel

>=50 cycles

BtcCrossVenueFundingCarryLongBinanceSpotShortHlPerpNeutral reframed the decayed funding-carry mechanism onto Hyperliquid, longing Binance spot and shorting the HL perp to harvest HL's richer hourly funding delta-neutrally. The plumbing worked and data loaded, but BTC HL funding almost never cleared the entry threshold — just 2 cycles in 8 years — and fees dominated what little carry there was (77% of gross), while the legs failed to offset, leaving residual directional risk. The analyst ruled it three independent, un-tunable failures and redirected the family to a higher-funding HL altcoin perp with a corrected hedge. It ended after one iteration as abandoned, never advancing to optimization or risk review.

Moving funding carry onto Hyperliquid does not rescue it on BTC: even at its triggering extremes, BTC HL funding is too thin to clear a two-venue round-trip (Binance spot 0.10% + HL perp 0.09% + impact), and lowering the threshold worsens the fee ratio — the mechanism needs a higher-funding HL ALT perp where per-cycle carry far exceeds two-leg fees and produces 50+ cycles, with the cross-venue hedge sizing/pricing corrected so the legs actually offset.

It was abandoned at the pre-optimization backtest-review gate (verdict: abandon) on three optimization-unfixable failures — no optimization basis (only 2 cycles in 8 years), fees structurally dominating the thin BTC HL carry (even the best cycle loses to its fees, and lowering the threshold only adds smaller sub-fee carries), and a hedge that isn't actually neutral (non-offsetting leg PnLs) — so optimization and all later stages were never reached.

A market-neutral cross-venue hedged funding carry on BTC — longing BTCUSDT.BINANCE_SPOT and shorting an equal-USD-notional BTCUSD.HYPERLIQUID perp when smoothed Hyperliquid hourly funding hit a positive extreme, holding to collect the funding while the spot long cancels price exposure and flattening on funding decay — the analyst-suggested reframe of the decayed funding-carry mechanism onto Hyperliquid's richer, less-arbitraged funding, run as a single-pair 3-parameter threshold trade.

It barely traded: only 2 entry cycles (4 legs) fired across 2018-2026 (entry_signaled=2 over 5,316 evaluated bars), with a negative result — total return -9.3%, Sharpe -6.07, profit factor 0.62, expectancy -$51.5/trade, commission 77.15% of gross — where per-cycle commission ~$130 swamped funding collected of only +$6.38 and +$67.68, and the supposedly delta-neutral legs failed to offset (price PnLs +$309/-$514 and -$97/+$22); metrics_reliable=false on the 2-trade sample.