DUOrc™ — Version Comparison

This page tracks the continuous optimisation of DUOrc™ by TraderLab™ — AutoEdge Systems'™ proprietary AI tuning engine that runs 24 hours a day, 7 days a week. TraderLab™ employs walk-forward optimisation across a rolling dataset, systematically evaluating thousands of parameter combinations across all eleven trading sessions and validating each candidate on out-of-sample periods it has never seen — ensuring results reflect genuine edge rather than curve-fitting. All backtesting and validation is performed on NinjaTrader 8 Playback data, which replays true tick-by-tick Level 1 market data (bid/ask/last) — the highest-fidelity historical simulation environment NinjaTrader offers. It's worth noting that NinjaTrader's built-in optimizer does not support Level 1 Playback data at all — its native optimisation runs only on bar-based historical data, meaning no standard NinjaTrader optimisation tool can match the accuracy of what TraderLab™ produces. This is precisely why we built TraderLab™ from the ground up as a proprietary system sitting alongside the platform. Each version listed in the table below represents a distinct AI-tuned build that has passed this rigorous walk-forward validation process before being logged here.

Every weekend, the latest TraderLab™-tuned release is made available as a public update for all active DUOrc™ subscribers — ensuring your live deployment is always running the most current, AI-optimised configuration. Check back each week to see how the numbers evolve.

TraderTuner142 currently leads this week's series with a Series Score of 12.09 (net P&L of $22,094.86) across 6 TraderLab™-tuned builds. Series Leader is determined by score = edge ÷ (1 + max drawdown ÷ 450) − λ × |H1 score − H2 score| — where edge is each build's average points per trade, discounted for how few trades support it so a hot streak on a small sample scores lower than the same edge sustained over many trades; this rewards efficient, statistically-supported performance per trade, continuously discounts drawdown beyond 450 pts (the $900 risk threshold on 1 MNQ contract) instead of only past a hard cutoff, and further discounts any build whose edge diverges between the first half (H1) and second half (H2) of its walk-forward trades. Builds with fewer than 10 trades in either half are disqualified. Check back as new versions clear walk-forward validation and are added to the table and charts below.

A quick note on how to read these numbers: they show how this strategy performed over the past six months of tuning data, so think of them as a guide to recent behavior rather than a guarantee of what's ahead — average daily P&L, max drawdown, and other figures will naturally shift over time. We chose a six-month window because our walk-forward testing showed it strikes the best balance for tuning this type of strategy, and that's also why TraderLab™ re-tunes it every week, with a fresh build released each weekend.