Insights · Risk Management
Published: 6 October 2026 · by the QuantGovernor team
A portfolio of several strategies is only safer than one strategy if its components lose at different times. Correlation — the tendency of returns to move together — is the number that decides whether combining strategies reduces risk or merely multiplies the same bet. It is the least visible risk on a track record, and the one that most often turns "diversified" portfolios into single points of failure.
When two profitable components are genuinely decorrelated, their combined equity curve is smoother than either alone: one's bad week lands on the other's normal week, drawdowns overlap only by coincidence, and the portfolio's worst moment is milder than the sum of the components' worst moments. That smoothing is the entire economic point of a multi-strategy portfolio. But it is bought with decorrelation, not with the number of components: ten strategies that all lose in the same conditions are one strategy with extra paperwork.
Obvious correlation is instrument-level: two gold strategies share every gold shock. The subtler kind is behavioral: two systems on different instruments that both, say, sell volatility spikes or buy pullbacks in trends will lose in the same market regime, whatever they trade. This is why we describe strategies by their logic — scalping, breakout, trend-following, mean-reversion — and not just by their market: the logic tells you when it loses. Olympus is built on exactly this principle: four components chosen so that their losing conditions differ — breakout logic suffers in ranges where mean-reversion thrives, trend-following on FX answers different weather than gold scalping — under one shared −30% equity stop for whatever slips through.
The honest caveat, and the reason this article exists: correlation measured in calm markets flatters everyone. In stress events — a violent dollar move, a liquidity vacuum, a geopolitical shock — many strategies that looked independent reach for the exits together, and measured correlations jump. Diversification is real, but it is partial insurance whose payout shrinks exactly in disasters. A portfolio's risk should therefore be judged on its behavior in its worst weeks, not on average statistics — and capped by a mechanism that does not depend on diversification working, which is the job of the portfolio-level equity stop.
Decorrelation is not always the goal — sometimes concentration is a choice. Our gold systems concentrate on XAUUSD deliberately, because that is where the edge is; the honest consequence, stated rather than hidden, is that they share gold's shocks. The difference between concentration as a strategy and concentration as an accident is whether it is declared, sized for, and capped. Gold's specific behavior is priced into how those systems size and filter — and an investor combining several gold strategies should know they are adding exposure, not diversification.
Related: Maximum drawdown is not enough · What is an equity stop?. Past performance is not indicative of future results.
Risk warning
Trading leveraged financial instruments involves a high risk of losing capital and is not suitable for all investors. Past performance, even when verified, is not indicative of future results. The content of this site is informational and does not constitute financial advice; we never hold client funds. Full details in the Risk Disclosure.