FX

01The study

Six strategies we already tested on ES (RSI(2) dip buying, Supertrend, Donchian breakout, Bollinger reversal, RSI divergence and MACD crossover) are run on the four major pairs with the same rules and the same default parameters. Nothing is tuned for FX. A second, unrelated market is stronger evidence than a second look at the same one: an edge that shows up on several pairs is more convincing, though a real edge can also be specific to one market.

• Variants: 24 per pair, 96 in total, all declared in advance.
• Data: spot FX, 1-minute bid history from an IC Markets MT4 demo server, 2004 to September 2026. Periods: development 2004 to 2012, validation 2013 to 2019, holdout 2020 to 2026 (already used in an earlier test, so flagged).
• Costs: IC Markets Raw Spread terms ($7 commission per round turn on a 100,000 lot) plus an assumed 0.3 pip of spread, about 1.0 pip per round trip (1.2 on USDJPY), higher than the published averages. The MT4 history records no spreads, so this is an assumption.
• Gates: the six protocol gates, with a development t-statistic of 2.5 and a random control matched on each signal's own entry state.
• The cross-market rule: a variant counts as supported only if it passes every gate on at least 3 of the 4 pairs. One or two pairs is a single-pair result, not support.

The full dated specification, written before any FX result was seen: fx_spec.md.

02Results

0 of 96 passed every gate, and 0 met the bar for support (a pass on at least 3 of 4 pairs). 65 were positive before costs and 30 after. The full write-up, every result and the trade logs: Six Strategies on Four Currency Pairs.

03The other FX tests

Each test has its own post with its specification, code and trade logs.