trading currency correlations - is it an edge or does it just double your exposure

[derekwhite]

i've heard two completely opposite takes on correlation trading. one school says understanding correlations helps you avoid overexposure (don't be long eurusd and long gbpusd at the same time since they move together). the other says you can use correlations actively to identify divergences and trade the mean reversion.

which application of correlation is actually useful for retail traders?

[dreamchaser]

both applications are valid but they're solving different problems. the risk management application (avoiding duplicate exposure) is straightforward and useful for everyone. if eurusd and gbpusd are 85% correlated and you're long both, you're effectively double-sizing one dollar trade. the divergence trade is more sophisticated and requires understanding when the correlation breaks and why - that's harder.

[Dean]

the risk management application should be part of everyone's framework. beyond eurusd/gbpusd: gold and nzdusd are correlated (both risk-on assets), usdcad and crude oil move inversely. being long gold and long nzdusd with a short usdjpy is not three independent trades - it's essentially one leveraged risk-on bet. understanding this clustering prevents catastrophic losses when risk sentiment flips.

[Lucas]

where do you find reliable correlation data? i've seen wildly different numbers depending on the time period used.

[Ryan]

correlations change over time which is exactly the problem. the right frame is rolling correlation rather than static. a 60-day rolling correlation gives you current regime information. myfxbook has a correlation matrix, tradingview has correlation tools. what matters more than any specific number is the direction - are two pairs currently highly positively correlated, negatively, or uncorrelated? the precise number matters less than the regime.

[william_h]

tried active divergence trading for about 4 months. the mean reversion trades work but the timing is everything. a divergence can persist for much longer than seems rational before it resolves. also the correlation can shift permanently during a divergence - what looked like a temporary spread was actually a regime change. only works if you can hold through significant adverse movement with defined risk. harder than it looks.

[Adam]

use correlations for risk management first. active divergence trading is advanced and timing-sensitive. start with not doubling up correlated exposure.