Research
Empirical findings behind the strategies, published as they happen. A finding here can be corrected or replaced by a later post rather than edited in place -- the same convention the strategies themselves use for versions.
Searching 7,500 configurations for a prop-evaluation base, and what a high win rate actually costs
A staged sweep for a funded-account configuration. Win rate and reward:risk traded off against each other across the whole space, every high-win-rate config failed out of sample, and only a small handful finished net positive in all four periods tested -- all of them the same entry with the reversal engine on.
Is Tokyo Drift's 2024 losing streak a curve fit or a different market? Neither, it turns out
A hand-tuned MGC config was tested against three earlier default sets on identical bars. The tuning is not why 2024 loses -- but the untuned defaults beat the tuned ones with less drawdown, and both make their entire 2.5-year profit in two quarters of 2026.
The Baseline Lost Money — Why Stop-and-Reverse Turned a Losing Open-Range Strategy Around
The v1.5.0 baseline was net negative across 2.5 years of Globex opens. A Stop-and-Reverse mechanism plus session/EMA-stack gating flips that — this is the research behind v2.0.0's design, and the microstructure literature explaining why the raw open is such a noisy signal in the first place.
Eight Experiments on What Actually Works Around VWAP — and What Only Looked Like It Did
A structured empirical lab covering ATR-relative sizing, confluence filter tuning, ATR characterization, VWAP entry mechanics, and — most importantly — what happens to a promising-looking setup once realistic execution friction is applied.
How far does price actually run once it leaves VWAP, and does a fixed-point target survive a re-rating index?
A historical study of MNQ price expansion away from session VWAP, and a quarter-by-quarter check on whether fixed-point TP/SL targets stay meaningful as the index itself re-rates sharply higher over the sample.