SolZeroParamAlwaysInCompositeLS

SOL Zero-Free-Parameter 5-Factor Composite (Always-In) — Long-Short, Winning Composite Form on the Highest-Volatility Liquid Major, Clean Single-Asset Test (SOLUSDT.BINANCE, Daily Bars, 0 Tunable Parameters)

A LONG-SHORT, single-instrument directional strategy on SOLUSDT.BINANCE (USD-M perpetual), daily bars, applying the session's WINNING composite form — the zero-parameter, equal-weight, always-in 5-factor daily composite — to the highest-volatility liquid major as a clean single-asset completeness test. HONEST framing up-front: the composite family is now DEFINITIVELY falsified as a promotion path — the zero-param BTC composite came in at deflated Sharpe 0.847 (<0.95) and did NOT generalize in walk-forward; more factors, more assets (multi-major, outlier-driven), gating, and horizon-diversification all made it worse or equal. So this is NOT expected to promote; it is a clean diagnostic of whether SOL's much higher volatility (larger per-trade moves, higher raw signal amplitude) lifts the composite's deflated Sharpe above BTC's 0.85 ceiling — the one asset-volatility dimension untested cleanly (the multi-major basket that included SOL was contaminated by early-SOL thin-liquidity outliers, kurtosis 18.5; a standalone recent-history SOL test isolates that). It uses the exact winning form and avoids all falsified levers (no gating, no addition, no parameter search). NOT the confluence template (L56 — 5-factor composite), NOT a single-signal probe, NOT gated/selectivity (falsified), NOT convex/regime-decayed, NOT cross-sectional (L52 — single asset), NOT reversion (L53), NOT microstructure/carry/basis/options (dead/infra). Fills long-short (13.5%). Risk: vol-normalized sizing capped 20% of equity*leverage; 2x leverage cap (reads self.config.leverage).

Implemented exactly as briefed: the winning form (daily, equal-weight 5-factor, always-in, zero parameter search), byte-identical factor windows and sizing constants to the BTC and ETH versions, none of the falsified levers, on SOL alone. Zero free parameters is mechanically enforced - SensitivityAnalyzer.generate_variations skips names starting with '_' and the walk-forward objective passes underscore/dict values through instead of calling suggest_int/suggest_float - so a parameters dict holding only '_param_bounds' (EMPTY), '_fixed' and '_zero_free_parameters' yields zero sensitivity variations and identical Optuna trials. THE DIAGNOSTIC ALREADY HAS ITS ANSWER, AND IT IS NO. Dry-run on the real SOLUSDT daily catalog (2020-09-14 to 2026-08-05, fills at bar close, 0.05% taker each side, modelling the framework's real close-then-re-enter-next-bar flip semantics): 206 trades (35/yr), Sharpe 0.752, profit factor 1.63, max drawdown 13.9%, +60% cumulative - BELOW BTC's 1.088 on the identical construction, so SOL's higher volatility does not lift the composite. Worse, it makes it far more outlier-dependent, which is the specific contamination the hypothesis wanted isolated: trade-return kurtosis is 66.8 and the TOP TWO trades account for 85% of all summed trade return, while the MEDIAN trade is -2.32%. Restricting to 2022 onward - i.e. removing the early thin-liquidity era entirely - does not clean it up: kurtosis 48.6, the top two trades account for 106% of the summed return (the rest of the book is net negative), median trade -1.74%, Sharpe 0.493. The recent regime is outright negative: trailing two years Sharpe -0.06, and the 365-day sandbox window returns -6.0% with Sharpe -0.92, profit factor 0.55 and avg trade -1.20% across 34 trades. Per-year: 2021 +248.3%, 2022 +46.3%, 2023 +220.4%, 2024 -23.6%, 2025 +43.3%, 2026 -31.7% of summed trade returns. My read is that this confirms rather than rescues the family verdict and should be abandoned at backtest review rather than given optimization budget - the walk-forward OOS windows and the 15-day holdout all sit in the negative recent stretch, and a strategy whose entire profit is two trades cannot survive a deflated-Sharpe gate. I implemented it in full anyway because the clean measurement is the deliverable. ONE DELIBERATE DEVIATION, flagged: the hypothesis specifies a 2x leverage cap, but I set leverage 1.0. SOL's ATR is roughly double BTC's, so the same vol-normalized risk budget produces a SMALLER notional (10%-48% of equity, median 17%) and the 20% cap clips only 8 of 207 entries even at 1x and zero at 2x - raising leverage to 2 moves the full-sample result by 0.6 percentage points (+60.9% -> +60.3%). The code does consume self.config.leverage in the sizing cap, so 2x remains available, but declaring an amplification the strategy never uses would misrepresent the risk profile and trips the 'do not set a leverage you don't use' rule. The futures venue is still required because the strategy takes shorts. Layer-3 is safe at 34 trades in the 365-day sandbox window (the gate requires trades, not profit). All per-bar work is O(1) (bounded deques + running sums) so the 300s smoke cap is not at risk, and min_bars_required is 25 because all state accumulates inside calculate_signal - the composite goes live near bar 160.

The hypothesis explicitly declares this a diagnostic 'NOT expected to promote,' so the 2-hour optimization budget is unwarranted on its face. Substantively, it is the weakest instance of a form already proven to fail: the identical BTC daily zero-param composite (Sharpe 0.829, CI-low 0.223) just failed optimization on hard, non-waivable gates (DSR 0.847, walk-forward OOS 0.171 with two of three windows negative, PBO 1.0), and the ETH twin (0.665) was abandoned for the same reason. This SOL version is weaker still — Sharpe 0.563 with a NEGATIVE sharpe_ci_low (-0.1192) — and outlier-driven (kurtosis 15.07, headline resting on early-SOL single days of +22.5%/+24.2%), losing 2022, 2024, and recent 2026 (-4.6%). Same OHLCV momentum-composite class (L9/L41, 0/213); zero params means nothing to tune. The diagnostic's own answer is already visible — SOL's higher volatility LOWERED the composite Sharpe below BTC's ceiling rather than raising it. Abandon rather than spend optimization confirming a self-declared non-candidate.

Always-in-market long-short directional strategy on SOLUSDT.BINANCE daily bars with ZERO tunable parameters - the winning composite form run on the highest-volatility liquid major as a clean single-asset diagnostic. Every bar it computes five pre-registered daily factors - 40-day trend, 4-day close-location pressure, 28-day OBV signed-volume flow, 12-day semivariance vol-skew and 25-day range location - standardises each to a z-score against its own 120-bar trailing distribution, and averages them with EQUAL weights (1/5 each). calculate_signal returns that composite continuously in z units. The position is simply the SIGN of the composite: long while positive, short while negative, flipping when it crosses zero. There is no conviction threshold, no trend-agreement filter, no stop-loss and no time exit. Sizing is constant vol-normalized - equity x 1.5% / (2 x ATR%) - capped at 20% of equity x leverage. Every window, weight and sizing constant is a hard-coded class constant, self.parameters is never read for a numeric value, and the config's parameters dict contains no top-level numeric key, so the optimizer's search space is empty by construction.

Analyst: abandon at BACKTEST_REVIEW as the developer recommends — the composite family is already falsified as a promotion path and SOL confirms rather than rescues it. Do not spend optimization budget; the outlier-dependence (two trades = 85% of return) guarantees walk-forward collapse.

STRONG ABANDON SIGNAL for the analyst — and the developer explicitly recommends it. This is presented as a completeness DIAGNOSTIC, not a promotion candidate, and the data confirms the negative answer: the byte-identical winning composite form scores WORSE on SOL than BTC (dry-run Sharpe 0.752 vs BTC 1.088), and its profit is almost entirely two outlier trades — full-sample trade-return kurtosis 66.8, top-2 trades = 85% of summed return, median trade -2.32%; restricting to 2022+ does not clean it (kurtosis 48.6, top-2 = 106% of return with the rest of the book net negative). The recent regime — where the walk-forward OOS windows and 15-day holdout sit — is outright negative: sandbox total_return -6.28%, Sharpe -1.10, PF 0.36, win_rate 0.19, avg trade -1.47% over 31 trades; trailing-2yr Sharpe -0.06. A strategy whose entire edge is two trades cannot clear a deflated-Sharpe gate. This is an edge/overfitting/regime call the analyst owns, not a QA correctness defect.

Do not treat the negative sandbox as a code fault; the identical construction wins on BTC/ETH. The failure is asset-specific edge, correctly measured.

NOT an L17 code defect, despite the deeply negative sandbox. win_rate 0.19 (not 0), Sharpe -1.10 (|.|<<5), PF 0.36 (not 0.0) — none of the code-defect signatures fire. Decisively, the strategy code is BYTE-IDENTICAL to the BtcZeroParam and EthZeroParam composite siblings that are profitable, which rules out any systematic signal-polarity / sizing / exit bug. The negative result is a genuine edge failure of the composite on SOL's fat-tailed return distribution, exactly as the developer predicted — a performance fact for the analyst, not a bug to route back to the developer.

No action — this is the honest, rule-compliant leverage choice. Note for the analyst: leverage is genuinely inert here, so the diagnostic is a pure 1x vol-normalized measurement.

Deliberate, well-justified deviation from the hypothesis's stated '2x leverage cap': the developer set config leverage=1.0. This is CORRECT, not a mismatch. The code still reads self.config.leverage in the sizing cap (config sets 1.0, code consumes it), so there is no leverage_set_but_unused violation — the exact error that abandoned recent siblings. The justification is sound: SOL's ATR is ~2x BTC's, so the 20%*leverage cap clips only 8 of 207 entries at 1x and 0 at 2x, making 2x a 0.6pp no-op (+60.9% -> +60.3%); declaring an unused amplification would misrepresent the risk profile and trip the 'do not set a leverage you don't use' rule. Shorts still function at 1x on the BINANCE USD-M margin venue. Structure (leverage-aware cap reading config) matches the hypothesis; only the magnitude differs, defensibly.

Analyst: not decision-relevant for this diagnostic (it fails on edge before DSR matters).

'Zero free parameters' is mechanically real (verified: parameters dict has no top-level numeric key -> SensitivityAnalyzer emits zero variations and the walk-forward objective evaluates one identical config, collapsing best-of-N to N=1), but the form itself (5-factor set, windows, always-in/daily) was selected across this session's composite line. Moot here given the abandon recommendation, but noted for consistency: the DSR collapse reflects zero optimizer search, not zero total selection.

avg_trade_return_pct 2.41% above fees; market-neutral (beta 0.028); clean recent-history SOL span isolates the early-liquidity contamination

The hypothesis itself declares this a diagnostic 'NOT expected to promote' — no promotion path, not worth the 2-hour optimization budget

Weakest instance of an already-failed form: Sharpe 0.563 with NEGATIVE sharpe_ci_low (-0.1192), vs the BTC twin (0.829, CI-low 0.223) that already failed optimization on hard gates (DSR 0.847, OOS 0.171, PBO 1.0)

Outlier-driven (kurtosis 15.07): headline rests on early-SOL single days (+22.5%, +24.2%); loses 2022, 2024, and recent 2026 (-4.6%, rolling Sharpe -5.8 on the final bar)

OHLCV momentum-composite class (L9/L41, 0/213); zero params means nothing to tune — the diagnostic's answer (SOL volatility LOWERS the composite Sharpe, not raises it) is already visible

SolZeroParamAlwaysInCompositeLS was the deliberate completeness test of the composite program: with the family already falsified as a promotion path, it ran the exact winning form on SOL, the highest-volatility liquid major, purely to answer whether volatility could lift the deflated Sharpe above BTC's ceiling. The clean recent-history backtest answered no — +56.9% over 204 trades with the family's best per-trade return (2.41%) but the weakest Sharpe (0.563) and a negative CI floor (-0.119), outlier-driven (kurtosis 15.07) on a couple of early-SOL days and losing 2022, 2024, and 2026. The analyst abandoned it at backtest review as a self-declared non-candidate and the weakest instance of a form whose BTC and ETH twins had already failed on hard overfit gates: SOL's volatility lowered the composite Sharpe rather than raising it, and with zero parameters there was nothing to tune. It never reached optimization, analysis, or risk review.

Higher asset volatility does not lift a momentum composite's deflated Sharpe — SOL's larger moves produced a bigger per-trade return (2.41%) but a lower, outlier-driven Sharpe (0.563 with a negative CI floor) than BTC, because the volatility inflates single-day tails (kurtosis 15) rather than the risk-adjusted edge, completing the definitive falsification of the composite family across factors, assets, horizons, and volatility regimes.

The analyst abandoned it at backtest review because the hypothesis itself declared it a diagnostic not expected to promote, and substantively it is the weakest instance of a form already proven to fail — the identical BTC daily zero-param composite (Sharpe 0.829) failed optimization on hard gates (DSR 0.847, OOS 0.171, PBO 1.0) and the ETH twin (0.665) was abandoned likewise, so a SOL version at 0.563 with a negative CI floor is even more certain to fail; with zero parameters there is nothing to tune, and the diagnostic's own answer was already visible: SOL's higher volatility lowered the composite Sharpe rather than raising it.

A long-short, single-instrument directional strategy on SOLUSDT.BINANCE daily bars that applied the session's 'winning' composite form — daily, equal-weight 5-factor (trend, close-pressure, volume-flow, vol-skew, range-position), always-in-market, zero tunable parameters — to the highest-volatility liquid major as an explicitly-framed clean diagnostic (not a promotion candidate) of whether SOL's larger per-trade moves lift the composite's deflated Sharpe above BTC's ~0.85 ceiling, isolating the volatility dimension that the outlier-contaminated multi-major basket could not test cleanly.

The backtest (2152 daily bars, recent-history SOL span from 2020-09) returned +56.9% over 204 trades with avg_trade_return_pct 2.41% (the highest of the composite family, above fees), profit factor 1.416, and market-neutral character (beta 0.028). But it was the weakest instance of the form on risk-adjusted terms: Sharpe just 0.563 with a NEGATIVE sharpe_ci_low (-0.119), win rate 31.4%, information ratio -0.78, and heavily outlier-driven (kurtosis 15.07) — the headline resting on early-SOL single days (+22.5%, +24.2%) while losing 2022, 2024, and recent 2026 (-4.6%, rolling Sharpe -5.8 on the final bar).