Guides / Position Sizing Across Multiple Strategies
GUIDESizing multiple strategies without stacking hidden risk.
Each strategy's backtest can look "safe" in isolation and still combine into a dangerous account when you run several at once. Here's why, and how to size the account instead of each piece separately.
Quick answer: sizing each strategy independently and adding up the risk assumes their losses are unrelated. They usually aren't. Strategies that trade the same pairs, sessions, or setup type tend to lose at the same time — so "3 strategies at 2% each" isn't a 6% risk budget, it's closer to however correlated those three actually are, which is almost always higher than it looks on paper.
Why sizing strategy-by-strategy hides risk
Run one strategy's backtest and it might show an 8% max drawdown at a 2% risk-per-trade setting — comfortably inside most people's limits. Add two more strategies, each individually backtested to a similar 8% max drawdown at 2% risk, and the temptation is to assume the account-level risk is still "fine," since each piece looks fine on its own.
That reasoning only holds if the three strategies lose money at different times. In practice, algo strategies built around similar logic — trend-following, breakout, the same handful of major pairs — tend to draw down together, not separately.
Where correlation actually comes from
- Trading the same pair or a tightly correlated pair (EURUSD and GBPUSD, for instance) under similar logic.
- Running the same session window, so all three strategies are exposed to the same news release or liquidity gap.
- Sharing a strategy family — three variations of the same trend-following idea will tend to agree with each other, including when they're wrong.
- A single account-level event (a broker outage, a slippage spike, a shared stop-loss level) that touches every open strategy at once.
A worked example
Three strategies, each individually backtested to an 8% max drawdown at 2% risk per trade. If they were genuinely uncorrelated, combined drawdown in a bad month would land somewhere well under a simple 24% sum, because losses would partially offset. If they're 60–70% correlated — realistic for strategies sharing pairs or a trend-following bias — a shared bad regime can push combined drawdown toward 15–18%, nearly double what any single strategy's backtest implied, without any one strategy individually breaking its own rules.
Where traders get this wrong
- Equal-weighting every strategy regardless of how correlated they are with the rest of the book.
- Sizing a brand-new strategy the same as one with a long, stable live track record.
- Adding or removing a strategy without re-solving the weights for the strategies that remain.
- Treating each strategy's account-level exposure as independent just because they're logically separate systems, when they share one live account's equity.
What to actually do about it
Treat the account as the unit of risk, not the individual strategy. Weight each strategy by how it actually behaves alongside the others — including how correlated its drawdowns are — rather than giving everything an equal slice. Re-solve the weight split whenever the strategy roster changes, since adding or dropping one strategy changes the correlation picture for all of them.
Allocate solves the correlation-aware weight split for your actual strategies, not a naive equal-weight guess.
Open Allocate