Quarterly methodology notes from the team behind CORE APEX. What changed, what didn't, and what the data is saying about the next regime.
A quiet parameter change after three months of paper-testing. The London open of 2025 became progressively choppier; we widened the entry filter by twelve basis points. Win rate up 3.4 points, average trade unchanged.
Read note →The Yen carry unwind and its second-order effects on our positioning. Four bad days in a row, eleven weeks to recover. What we did, what we considered, and what we didn't change despite the temptation.
Read note →BTCUSD realised vol is now sustained below 45%, down from a 70% average two years ago. We adapted position sizing in November 2025; here's how, and what it implies for risk-adjusted return.
Read note →An experiment we ran in mid-2025. Replacing fixed time-stops with structural trailing stops looked clever in backtest, performed worse live. The unobvious reason: it broke regime-detection's information flow.
Read note →Earnings reactions are dispersed enough that no liquidity-based signal we've designed has positive expectancy net of vol. Three years of data and seven attempts; we're calling it.
Read note →USDJPY went live in September 2025. Why we added it, why we held off for two years before doing so, and the per-instrument constraints in the algorithm that made it feasible to extend the universe.
Read note →Subscribe to get notes early. Subscribers receive each note one week before public posting, with a more detailed version including per-setup attribution. Get started — $30/mo.