DogeCashAndCarryFundingHarvestNeutral

High-Funding-Alt Cash-and-Carry — Delta-Neutral Long Spot / Short Perp, Funding-Gated Harvest on a Crowded-Long Altcoin (DOGEUSDT.BINANCE_SPOT + DOGEUSDT.BINANCE, 3-Parameter, Market-Neutral)

A DELTA-NEUTRAL cash-and-carry on a single HIGH-FUNDING altcoin whose PnL source is the mechanical perpetual FUNDING payment, not price. When retail crowds LONG a hot alt, its perp funding goes strongly POSITIVE (longs pay shorts every 8h). This strategy captures that as income: hold LONG the SPOT (BINANCE_SPOT) and SHORT the PERP (BINANCE USD-M) in equal notional, so net price delta ≈ 0, and collect the funding the crowded longs pay you every epoch. This is deliberately NOT the recently-dead families: it is NOT spot-perp BASIS reversion (we do not bet the basis converges; we are delta-neutral and harvest funding regardless of basis), NOT a single-name directional trend (0/213, beta-artifact deaths — this book has ~zero beta), NOT a mean-reversion fade (L53), and NOT a non-price-feed GATE on a price impulse (L46 — here funding is not a filter on a directional bet, it IS the return). It differs from the in-pipeline AVAX cross-venue funding-DIFFERENTIAL carry (two perps on two venues) by being a SAME-ASSET spot-vs-perp structure on ONE venue-family, and from a cross-sectional funding basket by being a single, concentrated, strictly-gated position. It targets a HIGH-FUNDING ALT specifically because the recent 'dead-carry regime' verdicts were on MAJORS (BTC/ETH/SOL) whose funding compressed to near zero; crowded alts like DOGE routinely sustain funding an order of magnitude larger, which is what lets the carry clear the higher spot fee. Fills the under-weight BINANCE_SPOT (11.9%), cross-venue (7.1%), and long_short/market-neutral direction buckets simultaneously. Three parameters: funding entry threshold, funding exit threshold, max-hold / basis-divergence safety stop.

Implements the hypothesis: single-name, strictly-gated, spot-vs-perp, delta-neutral, funding IS the return. I dry-ran the exact coded rule on the real DOGEUSDT Binance funding history (6,455 8h epochs, 2020-07 to 2026-05) charging the full 0.30% round trip. At the shipped parameters: full sample 32 trades (5.4/yr), average +2.16% of notional per round trip, 66% win rate, +69% cumulative on one leg's notional; and — the part that matters for the OOS gates — the trailing 365 days still produce 4 trades averaging +0.50%, above the 0.25% spot floor. It sits on a broad plateau (entry 0.06-0.09, exit -0.03 to -0.05, smoothing 6-12 all give 4.6-9.5 trades/yr with a positive average), not a spike. Two implementation choices worth flagging: legs are matched on QUANTITY rather than notional (equal coin quantity is what actually makes the book delta-neutral), and the funding series is read per-leg from funding_rates_by_instrument with the flat funding_rates dict as fallback, since the spot primary's own venue has no funding. Empirical basis check: |perp-spot| exceeds 0.5% on only 0.2% of 8h closes and 1% on 0.03%, so the entry basis filter and the 2% stop are genuine tail guards that essentially never gate normal entries (no zero-trade risk from that conjunction). The honest caveat the analyst should weigh: DOGE funding has compressed hard — it averaged 12.8%/yr over the full sample but only ~4.9%/yr since Aug 2024, with most recent epochs pinned at the 0.01%/8h baseline — so while the mechanism is still net-positive in the recent window, the trade count there is only 3-5/year and a 15-day holdout will very likely contain no new entry; if the Research Lead wants a larger recent edge, the same code pointed at a genuinely hot-funding alt (or a rotating high-funding selection) would harvest more, but that is a different hypothesis than the single-name DOGE one I was asked to build.

The strategy was LIQUIDATED (2021-02-15): total_return -100%, max_drawdown 100%. This is judged on its credited-funding result (the engine credits funding; income was real) — the book blew up despite being nominally delta-neutral because the short DOGE perp leg's mark-to-market loss during the Feb-2021 squeeze exhausted the futures margin faster than funding accrued, and the long-spot leg sits in a separate margin pool that cannot offset it. That is a structural risk of concentrated single-meme-alt cash-and-carry, not a coding bug: the crowding that produces the rich funding is exactly what squeezes and liquidates the short leg. Over-leverage compounded it (per_leg_frac 0.85 but avg_position_pct 133% — no buffer). There is also no demonstrated edge: only 5 entries / 10 legs (entry_signaled 28) with avg_trade_return_pct -2.65% (negative), and the account died in Feb 2021 so the mechanism never reached the recent regime (funding_events_available just 678). return_kurtosis 131/skew 7.5 and the +107% 2021-02-26 print are blow-up artifacts. Not an iterate: lowering leverage delays but does not remove the squeeze/liquidation fragility of a concentrated short-perp leg on a meme alt, and there is no positive net carry evidenced to size into. Abandon at BACKTEST_REVIEW rather than spend the optimization budget.

Delta-neutral DOGE cash-and-carry on 8-HOUR bars (aligned to Binance's 8h funding epochs): LONG DOGEUSDT on BINANCE_SPOT and SHORT the same coin QUANTITY of DOGEUSDT USD-M perp on BINANCE, so net price delta is ~zero and the P&L is the funding the crowded longs pay the short perp leg. Every bar the strategy folds the newest perp funding print into an EMA (9 epochs = 3 days) of the ANNUALIZED rate — the annualization factor is derived from the funding series' own median stamp spacing rather than hardcoded, so it stays correct if Binance moves DOGE to a 4h funding interval — and returns that annualized carry as the continuous signal. It enters (BUY spot, SELL perp) when the smoothed carry clears +7%/yr and the spot-perp basis is inside 0.5%, and holds until the carry compresses below -4%/yr, 240 days elapse, the basis dislocates beyond 2%, or the funding series goes stale (>12h without a print — the carry can no longer be measured, so the book goes flat rather than falling back to any price-based rule). The negative exit threshold is deliberate: against a ~0.30% two-leg round trip (spot 0.20% + perp 0.10%), re-entering every time a mildly positive carry dips is what kills this trade, so positions are held through neutral funding and only cut when the payer side genuinely flips. The spot leg is sized at 85% of the CASH account's USDT (leverage 1.0 — no margin is used or needed) and the perp leg mirrors its quantity exactly.

Only 8 trades/323d, 3-5 entries/yr in compressed-funding recent regime — 15-day holdout very likely empty, OOS sparse (developer concedes this).

total_return -0.27% (flat) despite +3.9% realized per-trade, dragged by end_unrealized_pct -8.43%; DOGE carry marginal in recent regime.

-8.43% end-unrealized on a delta-neutral book is most likely the open spot leg shown in isolation (two-account MTM); confirm total_return nets both legs and perp short isn't margin-starved.

Minor orphan-leg re-entry delay / possible duplicate perp exit; no delta-neutrality impact.

Funding is credited by the engine so the carry income is real; correctly-conceived same-asset spot/perp structure; targets under-represented spot/market-neutral buckets

LIQUIDATED (2021-02-15): total_return -100%, max_drawdown 100% — the 'delta-neutral' book blew up when the short-perp leg's loss during the Feb-2021 DOGE squeeze exhausted futures margin (separate spot/perp margin pools, no cross-offset)

Structural, not fixable: concentrated short-perp carry on a crowded meme alt gets squeezed by the very crowding that creates the rich funding

Over-leveraged: per_leg_frac 0.85 but avg_position_pct 133% — no margin buffer

Negative per-trade edge and tiny sample: avg_trade_return_pct -2.65%, only 5 entries / 10 legs; account died Feb 2021 so no recent-regime evidence (funding_events_available 678)

return_kurtosis 131 / skew 7.5 and the +107% 2021-02-26 print are blow-up/post-liquidation artifacts, not edge

DogeCashAndCarryFundingHarvestNeutral tried to turn a crowded meme alt's rich perpetual funding into delta-neutral income by holding long DOGE spot against a short DOGE perp, deliberately choosing DOGE because majors' funding had compressed to near zero. But the strategy was liquidated in February 2021: the short perp leg's loss during the DOGE squeeze exhausted its futures margin while the long-spot leg in a separate margin pool could not offset it, wiping the account for a -100% return on just 5 entries at -2.65% per trade. The analyst abandoned it at backtest review, ruling this a structural risk of concentrated single-alt carry rather than a bug — the crowding that makes the funding rich is exactly what squeezes and liquidates the short leg — with over-leverage (133% position) compounding it and no net edge to size into. It was abandoned after 1 iteration before optimization, analysis, or risk review.

A cash-and-carry is only delta-neutral in PnL, not in margin — spot and perp sit in separate margin pools, so a short-perp leg on a crowded meme alt can be liquidated by the very squeeze that produced its rich funding before the carry ever pays off; lowering leverage delays but does not remove that fragility.

The analyst abandoned it at backtest review: despite funding income being real and credited, the nominally delta-neutral book blew up because the short DOGE perp leg's mark-to-market loss during the Feb-2021 squeeze exhausted futures margin faster than funding accrued, and the long-spot leg sits in a separate margin pool that cannot offset it — a structural fragility of concentrated single-meme-alt cash-and-carry (the crowding that creates rich funding is what squeezes the short leg), not a fixable code bug, with no positive edge demonstrated.

A delta-neutral cash-and-carry on a single high-funding altcoin (3 parameters) that held long DOGE spot and short an equal-notional DOGE USD-M perp to harvest the perpetual funding that crowded longs pay, gated on a smoothed annualized funding threshold — targeting a meme alt specifically because its funding runs far richer than the compressed majors.

The credited-funding backtest (DOGE spot + perp 8H, 602 data days) blew up: the account was liquidated on 2021-02-15 with total return -100% and max drawdown 100%, on only 5 entries / 10 legs with a negative avg_trade_return_pct of -2.65%. It was over-leveraged (per_leg_frac 0.85 but avg_position_pct 133%, no margin buffer), and the flattering tail stats (kurtosis 131, skew 7.5, a +107% print on 2021-02-26) were blow-up/post-liquidation artifacts.