Escalation: Who You Call And When

Every desk has a real, if often informal, escalation ladder — who a trader tells first when something goes wrong, and at what point a problem stops being theirs alone to fix. Getting this wrong in either direction causes damage: escalating every minor issue makes a trader look like they can't handle their own book, while sitting on a real problem hoping to fix it quietly is one of the most common ways a small issue becomes a large, trust-destroying one.

The rough hierarchy, and the exceptions to it

Most problems start and end with the trader — a position that needs adjusting, a hedge that's drifted, a small error caught and corrected within minutes. The next tier is the desk head: bigger position problems, disagreements about strategy, anything that affects how much risk the desk as a whole is running. Above that sits risk management, who get called for anything that touches limits, unusual correlations across the book, or exposures the trader isn't sure how to size. And separately, compliance and legal sit outside the normal chain entirely, reserved for anything involving potential rule violations, market conduct questions, or conflicts of interest — situations where the right move is never to "handle it internally" first.

The exceptions matter more than the hierarchy itself. A genuine trade error — a fat-finger, a wrong instrument, an unintended short — should be escalated immediately, even if the trader thinks they can unwind it quietly before anyone notices, because errors compound and the earlier the desk knows, the more options exist to fix it cleanly. Anything that smells like a compliance issue, even something the trader believes is probably fine, should go to compliance directly rather than being run past a desk head first — desk heads aren't positioned to make that call, and routing it through them first just adds a delay and a witness to a decision that wasn't theirs to make. The instinct to "make sure it's really a problem" before escalating is exactly backwards for both of these cases: the cost of over-escalating a false alarm is a few minutes of someone's time, while the cost of under-escalating a real one compounds by the hour.

A concrete example: a trader notices, twenty minutes after the fact, that an order executed at ten times the intended size due to a fat-fingered quantity field. The instinct to quietly unwind it before the desk head notices is the wrong one — the right move is an immediate call to the desk head and risk, even though the position may already be back to normal, because the earlier everyone with authority over the book knows, the more choices exist if the market has moved in the meantime.