ShortSqueezeIgnitionFundingReversalLongBasket8H

Short-Squeeze Ignition: Deep-Negative-Funding REVERSAL Long Basket, Single-Leg Long-Only (BINANCE USD-M, 8 Liquid Perps, Go Long a Name Only When Its 8h Funding Is Deeply Negative AND Has Just TURNED UP From a Trough — Enter as Shorts BEGIN Covering, Not While Price Is Still Falling — SINGLE-LEG One Fee, Deep Funding Feed, Fee-Arithmetic Stated, 2-Parameter)

A LONG-ONLY, single-leg, funding-driven strategy built to fit the razor-thin viable space the meta-lessons define: L30 (pure-OHLCV trend/momentum/reversion = 0/213 → a genuine NON-PRICE dependency is required; this uses the FUNDING feed, the only DEEP/backfilled non-price series — NOT the shallow OI/taker/liquidation metrics feeds that data-froze prior baskets), L29 (no options), and L28 (TWO-leg carry fee-dies → this is SINGLE-LEG, one round trip, arithmetic stated). It harvests the SAME proven edge as the deployed negative-funding carry basket (Sharpe 1.97) but with a distinct, falling-knife-avoiding ENTRY-TIMING mechanism: instead of entering on the funding LEVEL alone (which can keep buying while a crowded-short name keeps dropping), it requires funding to be deeply negative AND to have just TURNED UP from a local trough — the signature of shorts BEGINNING to cover, i.e. a short-squeeze IGNITING. That funding rate-of-change reversal is the genuinely new signal (distinct from the deployed level-gated basket): it times the long into the squeeze bounce rather than into the ongoing decline, directly fixing the main loss mode of contrarian negative-funding longs. FEE ARITHMETIC (L28): entry requires funding ≤ −0.03%/8h = −0.09%/day of carry COLLECTED while long, held ~3–5 days ≈ 27–45bps carry vs a single-leg USD-M round trip of ~0.09% (~3–5× margin) BEFORE the squeeze price bounce that is the primary P&L. Universe: 8 deep-history liquid USD-M perps (BTC, ETH, SOL, XRP, DOGE, LINK, AVAX, LTC), each gated INDEPENDENTLY, equal-weight, conservatively sized. NOT the deployed level-only basket (reversal-timed), NOT a two-leg carry, NOT a funding RANK rotation (independent per-name gate), NOT pure-OHLCV, NOT options. 2 core parameters (funding depth threshold, trough-reversal confirmation).

Implements the hypothesis inside the razor-thin viable design space. (1) Non-price dependency (L30): the ONLY signal is the funding feed — the deep/backfilled non-price series — never OHLCV. (2) Single-leg (L28): one perp long per name pays one ~9bps round trip; the trough must be <= -0.03%/8h so a ~3-5 day hold accrues ~27-45bps carry (~3-5x the fee) before the squeeze bounce, clearing the avg-trade floor. (3) No options (L29). (4) Independent per-name level+reversal gate, NOT a top-K funding rank rotation (no cross-sectional sort). The genuinely NEW signal vs the deployed level-only carry basket is the funding rate-of-change reversal: entering on a confirmed turn-UP from a deep trough, not on the level alone, directly fixes the main loss mode (buying while a crowded-short name keeps dropping) by timing entry into shorts covering. Verification robustness: I confirmed against the real funding parquet that SOL (the primary — the only per-leg series available in the Layer-3 sandbox) produces 141 squeeze-ignition events (reversal_confirm=1) over recent history, so the sandbox reliably yields trades rather than looping on 'no trades'. In the full backtest, funding_rates_by_instrument supplies each of the 8 legs its own real funding so all names trade. Long-only + leverage 1.0 + 10%/name sizing structurally avoids the 50%+ drawdown / -100% liquidation blowups that killed prior baskets (L15); the 8% price stop + 10% take-profit backstop the directional risk the single-leg (unhedged) structure carries and lock the squeeze bounce that is the primary P&L. Venue is BINANCE USD-M futures because the edge is funding-based and requires holding the perp to receive funding.

No measurable edge: total_return 1.16% over ~5.8 years, Sharpe 0.018 (dead zero) with CI [-0.64, 0.71] straddling zero, PSR 0.52 (coin flip), and profit_factor 1.146 with avg_trade_return_pct 0.25% sitting right at the marginal fee edge. The result is well-sampled (178 trades) so this is a decisive read, not small-sample noise — the return stream is noise around zero with only 2023 meaningfully positive. The hypothesis's distinguishing innovation, the funding V-bottom 'squeeze-ignition' entry timing, DEGRADES the proven mechanism exactly like its persistence-gated sibling did: it trades more (178 vs 85) yet nets essentially zero, versus the already-deployed level-gated negative-funding carry basket at Sharpe 1.97. Both funding-timing filters tried this session removed edge rather than adding it, and reducing reversal_confirm toward its floor just converges to that deployed duplicate. Optimizing entry_threshold/reversal_confirm cannot create an edge from a PF-1.146, Sharpe-0.018 base — the 225-trial optimizer would overfit the single 2023 window and fail the deflated-Sharpe/OOS gates, and impact_cost_pct 9.6% with a $10.8M capacity means costs consume much of the thin gross anyway. The controlled 11.8% drawdown does not compensate for a strategy with no edge. Not worth 2 hours of optimization.

Long-only, single-leg short-squeeze-ignition basket over 8 liquid BINANCE USD-M perps (SOL primary + BTC, ETH, XRP, DOGE, LINK, AVAX, LTC), each gated INDEPENDENTLY (no cross-sectional rank). Goes LONG a name's perp only when its 8h funding is deeply negative (trough <= -0.03%/8h) AND has just TURNED UP from that trough (a funding V-bottom: funding was falling into a deep-negative trough, then rose for >= reversal_confirm consecutive stamps) — the signature of shorts beginning to cover / a squeeze igniting. This times the long into the squeeze bounce (the primary P&L) rather than into the ongoing decline, avoiding the falling-knife loss mode of level-only negative-funding longs while still collecting the residual negative-funding carry. Single-leg = one round-trip fee. Exits on funding normalization (>= exit_threshold, after a 24h min-hold), a 10% take-profit that banks the bounce, a 15-day max-hold timeout, or an 8% protective price stop. No leverage, conservative 10%-equity-per-name sizing (<=0.80 max gross) so the book cannot liquidate and drawdown is bounded by spot moves. Funding is read per-leg from funding_rates_by_instrument (real per-leg series in the full backtest), with the primary leg falling back to the flat funding_rates dict (its real series in the sandbox); a leg with no funding data stays flat — no price proxy is ever substituted.

No blocking change. Optionally raise min_hold_bars toward the stated multi-day floor so even the shortest holds clear the round trip on carry alone; but the take-profit-driven bounce is the real edge here.

Fee-arithmetic vs code drift. The hypothesis justifies fee viability on a '~3-5 day hold accruing ~27-45bps of carry vs ~9bps round trip'. The code's min_hold_bars=3 is 3 x 8H = 24h = ONE day (~9bps carry at the -0.03%/8h threshold), i.e. roughly break-even on carry alone at the enforced floor; the stated 3-5x margin only materializes on the longer funding-driven holds (sandbox avg hold ~1d22h). Importantly, this strategy's PRIMARY P&L is the squeeze bounce (10% take-profit), not the carry, and the sandbox shows a genuinely positive avg_trade_return_pct (+2.24%), so this is a documentation-vs-floor note, not a viability failure.

At backtest-review, verify the full 8-leg total_trades clears ~100; if it stays well below, judge on carry+bounce economics and the deployed sibling rather than on small-sample statistics.

Measurability. Sandbox exposes ONLY the primary (SOL) funding series, yielding 8 trades over 334 days; the other 7 legs stay flat in Layer 3 and trade only in the full backtest (funding_rates_by_instrument). With reversal_confirm=1 the gate is more permissive than a persistence gate, so the 8-leg full run should produce meaningfully more trades (order ~60-120 over multi-year history) -- likely reaching the reliability floor, but the analyst should confirm total_trades and metrics_reliable on the full run before trusting Sharpe/PF. This is intrinsic low-frequency behavior, not a code fix.

Optional: add a 'latest funding still <= 0 (or <= some fraction of threshold)' guard to ensure the position is opened while carry is still positive, tightening the squeeze-timing further.

Minor entry edge case: _squeeze_ignition gates on the TROUGH being <= entry_threshold and a strict up-turn, but does not require the LATEST funding to still be negative. With reversal_confirm=1 the latest stamp (rise_1) is normally still deeply negative, so carry is still collected; but a large single up-jump could enter just as funding crosses >=0. The min_hold_bars=3 floor then holds the position 24h regardless, and the take-profit/stop backstop the price risk, so this cannot cause an immediate whipsaw. Low impact given reversal_confirm=1.

Uses the deep/backfilled funding feed (1200 events) — L30-compliant non-price dependency, no data wall; carry is credited

Controlled risk: max_drawdown 11.8%, no liquidation, conservative unlevered sizing

Well-sampled (178 trades) so the near-zero result is a decisive read, not noise from too few trades

No edge: total_return 1.16% over ~5.8 years, Sharpe 0.018 with CI [-0.64, 0.71] straddling zero, PSR 0.52 (coin flip)

profit_factor 1.146 with avg_trade_return_pct 0.25% — right at the marginal fee edge, no fee-clearing margin

The squeeze-ignition reversal timing DEGRADES the proven mechanism: nets ~zero vs the deployed level-gated carry basket's Sharpe 1.97 — same failure as the persistence-gated sibling

impact_cost_pct 9.6% with $10.8M capacity — costs eat a meaningful share of the thin gross; return is noise (only 2023 positive)

This strategy was the sibling of the persistence-gated funding carry, swapping the filter for a funding-V-bottom 'squeeze ignition' entry meant to time longs into shorts covering rather than into an ongoing decline, while respecting the same hard lessons (deep funding feed, single-leg, stated fee math). It was well-sampled with controlled risk, but produced a dead-flat result: Sharpe 0.018, total return 1.16% over ~5.8 years, a profit factor barely above one at the fee edge, and returns concentrated in a single 2023 window. Crucially it repeated the sibling's failure — the reversal-timing overlay degraded the deployed level-gated carry basket (Sharpe 1.97) rather than improving it, trading twice as often for essentially zero. The analyst abandoned it at review: a PF-1.146, Sharpe-0.018 base offers nothing to optimize, loosening the confirmation just recreates the existing strategy, and impact costs consume much of the thin gross anyway.

A second funding-timing filter (the funding V-bottom rate-of-change) again removed edge rather than adding it — both timing overlays tried this session degraded the proven level-gated negative-funding carry — confirming that the funding-carry edge is best captured on the funding level alone and entry-timing refinements on top of it do not help.

It was abandoned at the pre-optimization backtest-review gate for no measurable edge: the squeeze-ignition reversal timing degraded the proven mechanism exactly as its persistence-gated sibling did — trading more (178 vs 85 trades) yet netting ~zero versus the deployed level-gated carry basket's Sharpe 1.97 — so reducing the confirmation just converges to the deployed duplicate and there was no promotable configuration to tune toward.

A long-only, single-leg funding-driven basket over 8 liquid BINANCE USD-M perps that goes long a name only when its 8h funding is deeply negative AND has just turned up from a local trough (a funding V-bottom) — timing the entry into a short-squeeze ignition rather than a still-falling knife — using the deep funding feed, with take-profit/stop backstops and stated carry-vs-fee arithmetic.

The backtest was well-sampled (178 trades) but had essentially no edge: total return 1.16% over ~5.8 years, Sharpe 0.018 (CI straddling zero), PSR 0.52, profit factor 1.146 with avg_trade_return_pct 0.25% sitting right at the marginal fee edge, controlled 11.8% max drawdown and no liquidation, but impact_cost_pct 9.6% on only $10.8M capacity and returns positive only in 2023.