Research

Research desk

Everything we are currently testing, and the methodology behind every number we publish. Results appear here as testing completes — not as marketing copy.

Methodology

Every backtest we publish states its instrument, timeframe, test period, spread and commission assumption, full parameter set, maximum drawdown, recovery factor, time under water, and trade count. We also recompute results with the single best trade removed, because a strategy whose edge disappears when one outlier is deleted does not have an edge.

We do not publish results we cannot reproduce from the stated inputs. Where a figure comes from a third party rather than our own testing, we link the source and date it.

Currently under test

Grid spacing sensitivity

How the same grid behaves across spacing values of 0.08%–0.25% on XAUUSD, with spread varied independently to separate the two effects.

In progress

Session breakout filters

London and New York open breaks filtered by Asian range, tested with a realistic spread ramp around the session boundary.

In progress

Drawdown-vs-recovery behaviour

Segment-level drawdown analysis across regimes, including the 2020 March dislocation and the 2022 rate-driven trend.

In progress

What we do not publish

Results from a period that does not include a gold drawdown comparable to the strategy's claimed tolerance.
Backtests without the spread assumption stated.
"Win rate" as a headline metric, without the payoff distribution it came from.
Live account statements edited for presentation.