You run five Expert Advisors. Each one has a decent profit factor, a tolerable drawdown, a steady equity curve. Five independent systems, five sources of return - a diversified portfolio.
Or is it? If three of them buy gold on the same news, lose on the same days and recover on the same days, you are not running five systems. You are running one system three times, at three times the size. Every one of them looks fine on its own. Together they carry a risk none of their reports shows.
This post explains how to check that with correlation analysis - in plain terms, with its limits - and how to act on the result.
What correlation measures
Correlation asks one question: when one system has a good day, does the other one too? The answer is a number between -1 and +1:
- +1 - they move together, day for day.
- 0 - no relation: knowing one tells you nothing about the other.
- -1 - they move opposite: one gains when the other loses.
The standard measure is the Pearson correlation coefficient. As a rough guide: above 0.7 the two behave largely as one; between 0.3 and 0.7 they share a meaningful part of their risk; below 0.3 they are mostly independent. Negative values mean one tends to offset the other.
How to measure it properly
Use daily results. Individual trades rarely line up in time, but days do. Sum each system's closed results per day, then compare the two series.
Compare only the days both traded. A day when only one system traded says nothing about how they move together, so it is left out.
Demand enough shared days. Ten shared days can produce a correlation of 0.8 by pure chance. A pair should only be judged once it has a reasonable number of trading days in common.
Look at opposite pairs too. Two systems with a strong negative correlation are worth knowing about: together they smooth each other's equity - or they are cancelling each other out, which is worth knowing too.
What correlation cannot tell you
- It sees closed results, not open risk. Two systems can hold large floating losses at the same moment and still close on different days.
- It changes over time. Systems that were independent in a quiet market can become twins in a crisis - exactly when it matters most. Check it again after big market moves.
- It shows what happens, not why. Two gold systems will often correlate simply because they trade the same instrument. Whether that is a problem depends on how much of your account rides on gold.
- Few shared days means little. Treat any figure from a short overlap as a hint, not a verdict.
What to do with the result
- Count highly correlated systems as one bet. If two EAs correlate at 0.8, size them together as if they were one, or keep only the better one.
- Look for real diversifiers. A system with low or negative correlation to the rest often improves the whole portfolio even if its own results are modest.
- Respect prop-firm limits. A daily loss limit is breached by the sum of everything that loses on the same day. Correlated systems are how that happens.
- Check before adding. Before you put a new EA on the account, see whether it duplicates what you already run.
How Strategy Ledger Pro does it
Strategy Ledger Pro builds the correlation list from your own account history. Click CORRELATION in the panel and choose what to compare: Expert Advisors, magic numbers, individual strategies or symbols. It sums each one's closed results per day, compares them only on the days both traded, shows a pair only when they share enough trading days, and lists opposite-moving pairs separately so they stay visible. The twenty most active rows are compared, and a note inside the list states the method, the period used and how many trades went in. It runs locally, on your data, and never touches your trades.
