How do I know if my gold backtest is a false positive?

Remove the three largest winners, split the sample into thirds and add realistic costs. If any of those three checks turns the result negative, the edge is not established.

Is it wrong to rely on news-day outliers?

Not if the strategy is explicitly a news strategy tested across 30 or more releases. It is wrong when a supposedly all-conditions strategy only earns on three days a quarter.

How many variations is too many?

There is no hard limit, but every additional variant raises the bar the winner must clear. Log them all and be suspicious when the best result is only marginally ahead of the pack.

FALSE POSITIVES IN GOLD TESTING

False positives in gold strategy testing arise from outlier dependence, single-regime samples, look-ahead bias, survivorship across tested variations and omitted trading costs.

Removing the three largest winning trades is the fastest check for whether an XAUUSD edge is real or event-driven.