QCL Native Pack 001

What it is

QCL Native Pack 001 is a completed research/evaluation package for QCL/Lean-compatible workflows. It is designed for teams evaluating engine-level research improvements in systematic trading environments.

The package is intended for evaluation by quantitative researchers, systematic trading teams, platform operators, and potential licensing or revenue partners.

How the mutations were discovered

The mutations evaluated here were not created by training on market data, optimizing historical returns, or searching financial price series for profitable parameters or patterns.

They were generated by a separate engine built to explore cross-domain mathematics: mathematical structures and relationships drawn from scientific and technical domains outside conventional quantitative-finance optimization, and then transferred into QCL / QuantConnect Lean.

The discovery engine is entirely deterministic. It does not use AI, machine-learning models, or language models to determine the QCL mutations. The mutations are generated by proprietary deterministic algorithms developed for cross-domain mathematical discovery.

Market data entered only at the subsequent validation stage, after the mathematical content of the mutations had already been generated and implemented.

This distinction is central to the result: the mathematical content of the mutations was generated independently of the market-price series used for subsequent performance measurement.

Discovery and validation pipeline

The empirical question is whether mathematical structure originating outside market-price optimization can produce useful executable changes when transferred into a quantitative trading engine.

Primary validation result

The audited NEW15 package produced a +51.6365% mean native price-level cumulative-return delta over baseline QCL across the curated 336-symbol Yahoo validation universe.

This is the measured incremental improvement over baseline QCL under the validated component-identity-aware mutated-minus-baseline methodology. It is not an absolute portfolio return to be compared directly with baseline QCL's own absolute cumulative return.

Exact validated delta: 0.5163650522944162909362596673.

Validation period: 2020-01-03 through 2026-06-04; 1,613 aligned return dates; 336 symbols.

Method: native event-return identity-aware mean-position-sign equity curve.

Risk-adjusted validation

The validated NEW15 differential signal also showed substantially lower measured risk than unmodified baseline QCL.

Relative to baseline QCL, NEW15 showed approximately 51.6% lower maximum drawdown, 39.6% lower annualized volatility, and 65.4% lower ulcer index, while also producing higher Sharpe, Sortino, and Calmar ratios.

Robustness and trading behavior

• Lower maximum drawdown than baseline QCL in 7 of 7 calendar-year segments.
• Lower rolling-window drawdown in 96.1% of tested rolling windows.
• High-volatility-regime maximum drawdown: 11.66% NEW15 vs. 28.99% baseline.
• During baseline drawdowns greater than 20%: 13.58% NEW15 drawdown vs. 30.23% baseline.
• NEW15 was flat on approximately 60.89% of symbol-days.
• Position-change frequency was only approximately 1.337% of possible consecutive symbol transitions.
• 196 of 336 symbols, or 58.33%, finished profitable under the differential signal.
• The largest symbol represented only approximately 6.73% of aggregate absolute terminal contribution magnitude.
• NEW15 Sharpe remained approximately equal to or above the no-cost baseline-QCL Sharpe through the explicitly tested 5 bps per turnover-unit cost level.

Validation integrity

The differential signal was independently reconstructed from raw baseline and mutated component events. In the five-symbol causal bridge audit, 120,960 baseline and 120,960 mutated price-level rows were paired with zero unmatched identities.

The reconstruction produced 8,064 symbol-day rows, exactly matching the authoritative comparison set, with zero semantic mismatches. All audited return dates occurred strictly after their corresponding signal dates.

Bridge validation status: PASS.

Component-level calculation: mutated.Position - baseline.Position. Identity-matched component deltas are aggregated into the NEW15 signal before subsequent returns are used for scoring.

Package boundaries

• Licensed C# / Lean-compatible native engine research/evaluation package.
• Historical validation result, not a performance guarantee.
• No source ownership by default.
• No production trading rights by default.
• No redistribution, resale, sublicensing, or exclusivity by default.
• Execution artifacts or deeper implementation materials are only provided under separate agreement.

Evaluation access

A refreshed sanitized buyer-delivery package is staged separately. Access is contact-gated.

Restricted evaluation license and broader research license terms are available by request.

Request evaluation access through the repository issue link or contact Andrew Cadotte through GitHub.