BtcCrossVenueFundingDivergenceCarryNeutral1H

BTC Cross-Venue Funding-Rate Divergence Carry (Long Cheaper-Funding Perp / Short Richer-Funding Perp)

A market-neutral cross-venue pairs strategy that exploits funding-rate divergences between BTCUSDT.BINANCE (USD-M perp) and BTCUSD.HYPERLIQUID (perp). When the two venues' funding rates diverge meaningfully — one venue is paying longs to hold (negative funding) while the other is charging longs (positive funding), OR the absolute spread crosses a threshold — the strategy goes LONG the perp on the venue paying cheaper-to-hold side (negative or lower-positive funding) and SHORT the perp on the venue charging more (higher-positive funding). Both legs are sized to equal USD notional so the position is delta-neutral against BTC price moves. P&L is captured through the funding-rate differential paid every 8h on each venue, while the price exposure mostly cancels out. This addresses ALL FOUR major quota gaps simultaneously: cross-venue (currently 3.5% vs 15% target), Hyperliquid (5.8% vs 20% target), long_short direction (12% vs 45% goal), AND implements analyst-recommended untried mechanism family (b) funding-rate divergence. The mechanism is structurally durable — unlike breakout/reversion edges that decay with regime change, funding-rate divergences arise from cross-venue liquidity and positioning differences that persist across regimes.

Implements the hypothesis directly: a delta-neutral cross-venue pairs trade between Binance USD-M and Hyperliquid BTC perps that goes long the cheaper-funding venue and short the richer-funding venue. Binance is the primary leg so the engine resolves `funding_rates` to Binance USD-M and `hl_funding_rates` (historical HL funding, the correct flat {ns: rate} series — not the forward-looking hl_predicted_fundings snapshot) to Hyperliquid, giving both legs correct funding. The signal normalizes the venues' different funding cadences (Binance 8h, HL 1h -> x8) before differencing, so the per-8h spread and the entry/exit thresholds share units (percent-per-8h); 0.025%/8h ~= 27% annualized clears round-trip taker fees (~0.10% per leg) over a multi-period hold. Equal-NOTIONAL (not equal-quantity) hedging keeps the dollar exposure cancelling despite the two perps' slightly different prices. Futures venues are required (shorting + funding-based edge), leverage stays 1.0 with fixed 25%-equity-per-leg sizing for bounded exposure and overfit resistance. This addresses the cross-venue, Hyperliquid, long_short, and funding-divergence quota gaps simultaneously.

Structurally negative unit economics on BTC cross-venue funding carry, shown directly at the trade level (funding IS credited — this is a fair credited-result judgment, not a 'funding invisible' claim). Per round-trip pair, captured funding differential is ~$2-14 while two-leg commission is ~$47, so the carry does not clear fees even with a perfect hedge — the best pair (2026-01-20, $11.62 total funding) still nets -$35 against commissions before any price effect. On top of that, the 'delta-neutral' legs leave a large residual directional P&L from the Binance-vs-HL basis (net -$62 to -$108 per pair, e.g. 2026-04-05 HL +768.3 / Binance -876.2), which swamps the funding carry by 5-50x. The failure is not tunable: lowering entry_spread_pct (the only knob that would raise the trade count from 9) enters on SMALLER divergences and captures LESS funding for the same fees, making it worse; the current threshold already selects the widest divergences in the data and even those don't pay. This is structural to BTC specifically — the most-arbitraged perp has tiny, short-lived cross-venue funding divergences and a basis that moves more than the differential pays, falsifying the hypothesis's 'large and durable BTC divergence' premise. Distinct from this session's ETH cross-venue funding strategy (a fixable over-trading/churn bug that masked the economics): here holds are proper (3-17h) and the economics are exposed as net-negative. Also un-optimizable on 9 trades / ~7 months of overlapping HL data. FAILURE PATTERN: market-neutral cross-venue (Binance vs Hyperliquid) funding-divergence carry on BTC fails because BTC's cross-venue funding differential (~0.01-0.06%/8h of notional) is an order of magnitude too small to clear the two-leg round-trip fee (~0.19%) and is dwarfed by the Binance-HL basis/tracking noise (~0.4%/trade); lowering the entry threshold to generate trades only reduces the captured carry. Cross-venue funding carry needs a venue/asset with a persistent, large funding skew (a high-funding alt), not BTC.

Market-neutral cross-venue funding-rate divergence carry on BTC perpetuals. Holds equal-and-opposite USD notional in BTCUSDT.BINANCE (USD-M perp) and BTCUSD.HYPERLIQUID (perp), so directional BTC price risk cancels and the strategy harvests the funding-rate DIFFERENTIAL paid every 8h. Computes a per-8h funding spread (binance_8h - hl_8h, in percent, with the HL hourly rate scaled x8 to a common footing). When Binance funds richer than HL beyond entry_spread_pct it SHORTs Binance / LONGs Hyperliquid; when Binance funds cheaper it flips (LONG Binance / SHORT HL), filling the long_short bucket symmetrically. Exits on convergence inside exit_spread_pct or a max-hold cap. Entry is atomic two-leg (both legs validated before either order). Fixed equal-notional sizing at 25% of equity per leg, leverage 1.0. No price-only fallback: if either funding series is missing the signal is 0.0 and it stays flat.

Mechanism is cleanly implemented and now genuinely backtestable: funding is credited on both legs (HL and Binance funding both non-zero in the trades), atomic two-leg hedge fires, no naked legs, fills under-quota cross-venue/HL/long-short buckets.

Holds are sensible (3-17h, not churning) — so the trade-level economics are a fair, direct test of the edge rather than a fee-churn artifact.

Funding carry per round trip ($2-14) is structurally SMALLER than the two-leg commission (~$47) — the edge does not clear fees even before price effects. Best pair (2026-01-20, $11.62 funding) still nets -$35 vs fees.

Even with a PERFECT hedge it loses: funding < commission per pair, so fixing neutrality cannot make it profitable.

The 'delta-neutral' legs leave large residual directional P&L (net -$62 to -$108/pair from HL-vs-Binance basis moves) that swamps the ~$2-14 funding carry by 5-50x.

Lowering entry_spread_pct to trade more captures even LESS funding for the same fees — the obvious tuning knob makes it worse, so there is no robust region to optimize toward.

Only 9 entries over ~7 months of overlapping HL data — not optimizable regardless (would fit noise). BTC is the most-arbitraged perp, so cross-venue funding divergences are structurally tiny — the hypothesis's 'large and durable BTC divergence' premise is empirically false.

enough for stable optimization

net basis residual $50-110/pair

~0 residual

$2-14 funding vs ~$47 commission

funding > round-trip fees

BtcCrossVenueFundingDivergenceCarryNeutral1H harvested the Binance-vs-Hyperliquid BTC funding differential as a delta-neutral pair, longing the cheaper-funding venue and shorting the richer, to fill three under-quota buckets. The implementation was clean and genuinely backtestable with funding credited on both legs and sensible 3-17h holds, but the economics were exposed as hopeless: the spread fired only 9 times, captured ~$2-14/pair against ~$47 commission, and left a -$62 to -$108/pair basis residual that dwarfed the carry. The analyst ruled it structurally negative on BTC specifically — the most-arbitraged perp has tiny, short-lived divergences and a basis that moves more than the carry pays — distinct from the session's ETH churn bug, and un-optimizable on 9 trades. It ended after one iteration as abandoned, never advancing to optimization or risk review.

BTC is the wrong target for cross-venue funding-divergence carry: as the most-arbitraged perp its cross-venue differential (~0.01-0.06%/8h) is far too small to clear round-trip fees and is overwhelmed by Binance-HL basis/tracking noise, and the one tuning knob (lowering the entry spread) makes it worse — the mechanism needs a venue/asset with persistent large funding skew (a high-funding altcoin), not BTC.

It was abandoned at the pre-optimization backtest-review gate (verdict: abandon) as structurally negative unit economics shown at the trade level — funding is credited but is an order of magnitude too small to clear the ~0.19% two-leg fee and is dwarfed by Binance-HL basis noise, even a perfect hedge still loses, and lowering the threshold to add trades only captures less funding — so with no tunable region and just 9 trades, optimization and all later stages were never reached.

A market-neutral cross-venue funding-rate divergence carry on BTC perps — when the per-8h funding differential between BTCUSDT.BINANCE (USD-M) and BTCUSD.HYPERLIQUID crossed a threshold, longing the cheaper-funding venue and shorting the richer one in equal USD notional to harvest the funding spread while staying delta-neutral on BTC price, with atomic both-legs-validated entry and no price-only fallback, filling the cross-venue, Hyperliquid, and long_short quota gaps.

The spread cleared the threshold only 9 times (18 leg-trades over ~7 months of overlapping HL data), and the unit economics were structurally negative: profit factor 0.80, Sharpe -13.7 (CI [-23.4, -8.5]), expectancy -$27.7/trade, win rate 44%, with commission at 21.8% of gross — captured funding of only ~$2-14/pair against ~$47 two-leg commission, plus a large residual directional basis PnL of -$62 to -$108/pair that swamped the carry by 5-50x.