Reading exposure while the position is still open

End-of-day risk reporting answers a question that stopped mattering hours earlier.

Most brokerage risk reporting is retrospective by construction. Positions are extracted after the session, exposure is aggregated, a report is produced, and somebody reads it the following morning. The report is accurate. It is also a description of a situation that has already resolved itself, one way or the other.

This is fine for the questions retrospective reporting is good at — how did the book perform, where is concentration building over weeks, which segments behave how. It is useless for the question that actually costs money, which is whether something is going wrong right now.

Why it stays retrospective

Rarely because anyone decided it should be. Usually because live state lives on the trading server, the definition of the limit lives in a policy document, and the person who would act on it is looking at a third screen. The gap between those three is filled by attention, and attention is not available at three in the morning on a Sunday.

A limit that only works when somebody is watching is not a limit. It is a hope with a threshold attached.

What has to be true instead

• —Account state is read continuously, not extracted on a schedule

• —The limit is configuration in the same system that reads the state, not a paragraph elsewhere

• —What happens on a breach is decided in advance: notify, restrict, act, or a sequence of those

• —The action reaches the platform through the same layer that detected the breach

• —Every evaluation is recorded with the values that triggered it

That last point is the one that gets left out, and it is the one that matters when a client disputes an outcome. A breach that cannot be explained — which rule, which values, at what time, and what was done — is worse than no rule, because it looks arbitrary from the outside.

The evaluation programme case

Funded trader programmes make the point unavoidable. A drawdown rule that is enforced from yesterday's report is not a rule, because the trader kept trading after the breach and everyone now has to argue about which trades counted. The same rule evaluated against live state has one answer, recorded once, at the moment it happened.

It is the same mechanism either way. Read live, define the limit where the reading happens, act through the same connection, and keep the record. Everything else is a report about the past.