Accounting for correlation in multi-pair risk limits
I have a question about how you guys structure your maximum portfolio risk when dealing with highly correlated assets. Currently, my hard limit is 1.5% total risk per trade, and I try not to exceed 5% total open risk across the whole book. This works fine on paper when my setups are spread across independent sectors.
But lately, I've run into issues when multiple setups appear simultaneously on pairs like $AUDUSD, $NZDUSD, and even some cross-rates like $AUDJPY. Individually, each trade fits my criteria and stays within the 1.5% limit. However, during high-impact macro data releases (like yesterday's US CPI), these setups behave almost as a single position. I ended up taking a hit that felt way too large for a 'diversified' set of trades because they all moved in tandem against the USD.
How do you factor correlation into your risk-sizing model? Do you treat highly correlated pairs as a single 'macro block' and split the 1.5% among them, or do you have a specific mathematical discount factor you apply to the position sizes?