SolAvaxFourHourRatioZScorePairsMeanReversionLongShort

SOL-AVAX 4H Cointegrated Pairs Ratio Mean-Reversion Long-Short (BINANCE Futures, OHLCV-Only, 30-Day Z-Score)

Market-neutral 4-hour pairs ratio mean-reversion strategy on SOLUSDT and AVAXUSDT Binance USD-M futures. Mechanism: compute the SOL/AVAX price ratio on each closed 4H bar, then z-score that ratio against its 30-day (180-bar) rolling mean and standard deviation. When |z| > 2.0, take a delta-neutral pair position — long the under-performing side, short the over-performing side, sized so that USD notional of each leg matches at entry. Exit when z crosses back through 0 (mean reversion completed) or after a hard time-stop. SOL and AVAX are both Layer-1 smart-contract chains with shared sector exposure (DeFi TVL flows, retail alt-season flows, EVM-compatible ecosystem), so idiosyncratic ratio divergences should revert as broader sector flows dominate. EXPLICITLY DISTINCT from the BTC-ETH spread MR failures cataloged this session: (a) different asset pair with intra-sector cointegration relationship — SOL/AVAX are direct L1 competitors whereas BTC-ETH spans different crypto categories, (b) wider |z| > 2.0 threshold than prior tighter thresholds, (c) 4H decision frequency hits the empirical sweet spot between daily-breakout-OOS-empty and 1H-momentum-fee-eaten regimes, (d) ZERO funding-rate dependency, ZERO HL data dependency, ZERO fallback mode — pure OHLCV mechanism fully testable in current factory infrastructure with no blocked supplementary data pipeline issues. Deliberately MINIMAL 4-parameter design (z_entry_threshold, z_exit_threshold, lookback_bars, time_stop_bars) directly following analyst guidance that 9-parameter strategies cannot be reliably optimized at this trade frequency. Fills three under-represented portfolio buckets simultaneously: pairs scope (17.4% vs higher target), long_short direction (11.9% vs higher target), and explicitly distinct mechanism class from the dying daily/4H Donchian-breakout family.

Implements the hypothesis with the proven centralized multi-leg pattern and the repo's SyncedBuffer helper (used by landed pairs strategies like ADADOTPairsSpreadMR and BnbEthRatioZScoreLongSpreadPairMR). SyncedBuffer guarantees the ratio is always computed on contemporaneous SOL/AVAX closes — it commits an (a,b) sample only when both legs report the same ts_ns — eliminating the independent-array-trim alignment bug. The base cross-leg barrier (both legs are 4-HOUR) ensures calculate_signal fires only when the AVAX bar for the current SOL timestamp is present, so the buffer commits one aligned sample per bar. min_bars_required is 2 so the buffer starts filling immediately; zscore_last returns None until 180 aligned samples accrue, naturally gating the warmup without dead bars. Entry is dollar-neutral (each leg = equity*leg_frac/price, not equal qty) which is what makes the position delta-neutral so P&L tracks ratio reversion rather than either coin's direction; the entry is atomic — if either leg fails its qty/min-notional check, neither leg is opened, preventing a naked directional bet. Exits close both legs on z crossing the z_exit band toward 0 (short side at z<=z_exit, long side at z>=-z_exit) or a 60-bar time stop, with entry_sign re-inferred from live position side if state is ever lost. leg_frac=0.45 keeps gross ~90% within the leverage-1.0 futures margin ceiling (so leverage is not set >1 and the unused-leverage gate does not apply). The returned z-score is a genuinely continuous per-bar signal so Layer-2 frozen-signal detection passes while entries remain threshold-gated. Only 4 edge parameters per the analyst's low-parameter guidance for this trade frequency; the should_enter/should_exit/position_size hooks are inert because the base entry path manages only the primary instrument.

Do not optimize — the strategy is a money-loser whose core premise is falsified by its own backtest. Over 222 trades it returns -19.1% with profit_factor 0.976, expectancy -$60.6/trade, Sharpe 0.051 (sharpe_ci_low -0.88), and a 65.7% max drawdown on a supposedly market-neutral pair. That combination is the signature of a NON-stationary spread: the SOL/AVAX ratio trended (SOL structurally outperformed AVAX across 2023-2025) instead of mean-reverting, so the z-score logic kept shorting the winner and longing the loser, producing a left-tail-heavy P&L (return_skew -3.4, kurtosis 70.8, largest_loss -$44k vs largest_win +$27k). The hypothesis bet specifically on SOL/AVAX intra-sector cointegration; the data shows that relationship did not hold over the test period. This is a de-cointegration / dead-pair failure, not a parameter problem — no change to z_entry, z_exit, lookback, or time_stop makes a trending ratio revert, and trade count is healthy (222) so it is not sparsity. The pairs mean-reversion mechanism itself may be sound on a genuinely cointegrated pair, but THIS pair is structurally unsuitable; the idea should be abandoned (or, if the mechanism is to be pursued, re-pointed at a still-cointegrated pair via a fresh hypothesis) rather than spend 2 hours optimizing a non-stationary spread.

Market-neutral 4H pairs ratio mean-reversion on SOLUSDT (leg A) and AVAXUSDT (leg B) Binance USD-M futures. On each timestamp-aligned 4H bar it commits the (SOL, AVAX) close pair to a SyncedBuffer, computes the SOL/AVAX ratio, and z-scores it over a 180-bar (30-day) rolling window. When z >= +2.0 it shorts SOL and longs AVAX; when z <= -2.0 it longs SOL and shorts AVAX, each leg sized to equal USD notional (delta-neutral). It closes both legs when z reverts back through the z_exit band toward 0, or after a 60-bar time stop. Pure OHLCV, no funding/HL/supplementary dependency, leverage 1.0 with ~90% gross (45% per leg). Four edge parameters.

Clean market-neutral construction (beta -0.02, benchmark_correlation -0.08) and healthy trade count (222) — the mechanism executes as designed

Pure OHLCV, no supplementary-data dependency; minimal 4-parameter design

Loses money: total_return -19.1%, PF 0.976, expectancy -$60.6/trade, Sharpe 0.051 with sharpe_ci_low -0.88 (CI straddles 0)

65.7% max drawdown on a supposedly market-neutral pair — the spread is not stationary; the SOL/AVAX ratio trended (SOL structurally outperformed AVAX) rather than reverting

Left-tail risk profile (return_skew -3.4, kurtosis 70.8, largest_loss -$44k vs largest_win +$27k) — the MR logic repeatedly faded a persistent trend

Core cointegration premise is falsified by the data; no parameter change fixes a de-cointegrated pair

SolAvaxFourHourRatioZScorePairsMeanReversionLongShort bet that SOL and AVAX, as intra-sector L1 competitors, would exhibit a cointegrated ratio whose z-score divergences revert, filling under-represented pairs and long/short portfolio buckets with a clean minimal 4-parameter design. The mechanism executed exactly as intended — genuinely market-neutral with 222 trades — but it lost 19.1% with a profit factor below 1 and a 65.7% drawdown, because the SOL/AVAX ratio trended rather than reverting as SOL structurally outperformed AVAX. That de-cointegration produced a left-tail-heavy P&L from repeatedly fading a persistent trend, a structural failure no parameter tuning can fix. The analyst abandoned the arm at the backtest-review gate, noting the mean-reversion mechanism might still work if re-pointed at a genuinely cointegrated pair via a fresh hypothesis.

A pairs mean-reversion strategy lives or dies on the pair actually being cointegrated — no change to z-entry, z-exit, lookback, or time-stop can make a trending, de-cointegrated ratio revert, so the asset-pair selection must be validated for stationarity before the mechanism can work.

The analyst abandoned it at the backtest-review gate before optimization because its core cointegration premise was falsified by its own data: the 65.7% drawdown on a supposedly market-neutral pair signaled a non-stationary spread — the SOL/AVAX ratio trended (SOL structurally outperformed AVAX across 2023-2025) rather than reverting, so the z-score logic repeatedly shorted the winner and longed the loser.

A market-neutral, OHLCV-only 4H pairs mean-reversion long/short on SOLUSDT and AVAXUSDT Binance futures — computing the SOL/AVAX price ratio, z-scoring it against a 30-day (180-bar) rolling mean/std, and taking a delta-neutral position (long the underperformer, short the overperformer) when |z| > 2.0, exiting when z crosses back through 0 or on a time stop.

The mechanism executed cleanly as market-neutral (beta -0.02, benchmark correlation -0.08) over a healthy 222 trades (111 long, 111 short, 53% win rate), but it lost money: -19.1% total return, profit factor 0.976, per-trade expectancy -$60.6, Sharpe 0.051 (CI low -0.88), and a 65.7% max drawdown with a heavy left tail (skew -3.4, kurtosis 70.8, largest loss -$44k vs largest win +$27k).