Risk Management Rules At The Portfolio Level

Correlation rises exactly when you need it not to

The awkward property of correlation is that it is not a constant. Relationships measured over a calm six months routinely tighten during a risk event, and pairs that had been usefully independent converge in an afternoon. Whatever diversification your book had is often gone at precisely the moment it was supposed to help.

The practical consequence is that you should not size as though a measured correlation of 0.4 will hold. Treat closely related instruments as one risk bucket and accept that you are over-reserving most of the time, because the cost of that is a slightly smaller average position and the cost of being wrong is a coordinated loss across the whole book. The general framework is set out in the risk management guide.

Margin usage is a second constraint, measured differently

Risk and margin are not the same number and a plan needs both. A 0.042 lot GBP/USD position with a 120 pip stop carries a full 1% of risk while reserving about $180 of margin at 30:1. A 0.20 lot EUR/USD position with a 25 pip stop carries the same 1% of risk and reserves roughly $720, because margin is charged on the notional value and that position is a $21,700 one.

Hold five of the latter on a $5,000 account and roughly three quarters of your equity is tied up as margin, which does not increase your loss on any trade but does decide whether the sixth ticket is accepted at all, and how much adverse movement the account can absorb before the broker starts closing positions for you. The margin calculator and the leverage calculator show where those thresholds sit for your own account size.

Enforcing the ceilings before the order, not after

• Keep a running exposure line next to your platform: total open risk, and the net figure for each currency you are exposed to. Update it when you open and when you close, not at the end of the day.
• Check the ceiling as a disqualifier in your entry checklist, alongside the chart conditions. A setup that would breach the per-currency cap is a skip, and it should be recorded as a skip so you can see later how often the cap actually bound.
• Re-derive correlation from your own results every quarter rather than from a published table, because it is the correlation of your trades that matters, and your entries cluster in particular sessions and conditions.

These limits are a cap on how much can go wrong simultaneously. They do not remove the possibility of a loss, and no portfolio rule does. What they change is whether a bad day is bounded by something you decided in advance or by how long you can stand watching it.