Correlation Risk: When “Diversified” Trades Are Actually the Same Bet Twice

Five open positions look like five separate decisions. Long EUR/USD, long GBP/USD, long AUD/USD, short USD/JPY, short USD/CHF — different pairs, different charts, different setup notes. It reads as diversification. It is often one trade: short the dollar, sized five times over, with five separate stop losses that can all get hit inside the same ten-minute move.

Why position count is the wrong number to look at

“How many trades am I in” answers a question nobody actually needs answered. The number that matters is how many independent bets those trades represent, and that number is almost always smaller than the position count. Two positions that move together are not two units of risk — they are one unit of risk, taken twice, with two commissions and two spreads instead of one.

This is easy to miss because correlation doesn't show up anywhere on a standard dashboard. Win rate, average R, and open P&L are all computed per trade. None of them ask whether the five trades currently open would all move against you for the same reason at the same time.

Where correlated exposure actually hides

• Same base or quote currency — any basket of dollar pairs, whichever side you took them on, is exposed to one thing: what the dollar does next. Four dollar pairs is not four ideas about the market.
• Same macro driver— a long on gold, a short on the dollar index, and a long on the Australian dollar can all be the same “risk-off” or “dollar-weak” view expressed three ways. They will tend to be right together and wrong together.
• Same session, same catalyst— three trades opened in the minutes around a single news release aren't three independent reads of the market, they're one reaction to one event, sized as if it were three separate opinions.
• Same instrument class— index CFDs on the S&P, the Nasdaq, and the Dow move together often enough that holding all three isn't spreading equity risk, it's tripling one view on U.S. stocks.

None of these are visible from the position list alone. They only show up when you look at what each trade is actually a bet on, not which symbol it happens to be labeled with.

What it does to your real risk per trade

A 1% risk rule assumes each trade's stop is an independent event. Stack five correlated positions at 1% each and the “5% total risk” on the sizing spreadsheet is a fiction the moment the dollar makes one large move — the five stops don't get hit as five separate 1% events spread over time, they get hit together, and the account takes something closer to the full 5% at once. The position sizing was correct on paper and wrong in practice, because it priced each trade as if the others didn't exist.

The same math cuts the other way with correlated winners: a run of trades that all worked can look like a strategy proving itself across five setups, when it was really one correct read on the dollar getting counted five times. That's a much thinner sample than it appears, and it's worth knowing before the next drawdown, not after it.

Checking for it without a statistics background

A full correlation matrix is overkill for most trading journals. A rougher check catches most of the real cases: for every open position, write down what it's a bet on in one phrase — “dollar weak,” “risk-off,” “oil supply,” “tech earnings.” Positions that land on the same phrase are the same bet, regardless of what symbols they're wearing. Count the distinct phrases, not the distinct tickers — that count is closer to your real number of independent positions.

It's also worth doing after the fact, not just before entry. Pull the trades that lost together during your worst week and check whether they shared a driver. If four of your five worst days were all dollar moves, the position sizing rule that assumes independence is the thing to fix, not the strategy that picked the trades.

getALPHA logs every synced trade with its instrument and direction, so a stretch of open positions can be checked for overlap instead of taken at face value, and the AI coachflags when new entries cluster around the same currency or driver as trades already open — the pattern that turns a “diversified” account into one oversized bet without anyone deciding it that way.