Capital Usage and Risk Constraints
Every desk trades inside limits. They can feel like an obstacle to a trader who is confident, and understanding why they exist changes that impression.
Why limits, rather than judgement
A trading strategy with positive expectation still loses sometimes. Over many trades the law of large numbers delivers the edge, but only if you are still trading when it does.
The mathematics is unforgiving. Lose 50% and you need a 100% gain to recover. Lose 90% and you need 900%. Because losses compound against you asymmetrically, capping the downside raises long-run growth even at the cost of some expected profit. That is Jensen's inequality applied to a career.
Limits are not a tax on good traders. They are what converts positive expectation into realised profit, by guaranteeing you survive the variance that stands between you and the average.
The main constraints
Position limits. The maximum you may hold in an instrument or group. Usually the most binding constraint for a market maker, and typically enforced pre-trade: the order is rejected before it reaches the exchange.
Loss limits. Maximum loss per day or per period. Hitting one usually stops trading for the session. This is deliberately blunt, because a trader having a bad day is exactly the person least well placed to judge whether to keep going.
Value at Risk. A statistical limit: the loss level exceeded only of the time. VaR summarises a whole book in one number, which is its appeal and its weakness: it says nothing about how bad the tail beyond it is, and it depends on a distribution assumption that understates fat tails. See VaR and expected shortfall.
Stress tests. What happens under a specific scenario: the market falls 10%, volatility triples, a correlation goes to 1. These complement VaR precisely because they do not rely on a distributional model.
Margin and capital usage. How much of the firm's capital your positions tie up. Capital committed to one desk cannot be deployed elsewhere, so capital efficiency is itself a performance measure.
Why they are automated
Limits are enforced by systems, not by asking. Pre-trade risk checks reject offending orders, and breaches trigger automatic flattening.
This is a considered design choice. The moment a trader most wants to exceed a limit (deep in a loss, convinced the position will come back) is exactly the moment their judgement is least reliable. Removing the decision from the individual is the point.
It also protects against the technical failure mode: a malfunctioning algorithm can accumulate an enormous position in seconds, far faster than a human could intervene. Several firms have been destroyed this way, and pre-trade checks are the defence.
Living within them
Good traders treat limits as a resource to allocate rather than a wall to press against.
Running consistently at 95% of your position limit leaves no capacity to take advantage of an unusual opportunity, and no room to absorb a fill you did not choose. Traders who stay well inside their limits during normal conditions have capacity available when it is worth something.
"You are at your position limit and a great opportunity appears. What do you do?" The answer is not to ask for a limit increase in the moment. It is to have managed your capacity so this situation is rare, and to reduce elsewhere if you genuinely want the trade.
Capital allocation
At firm level, capital is allocated between desks by risk-adjusted return rather than raw P&L. A desk making $10m with $100m of risk is a worse use of capital than one making $5m with $20m. This is why the Sharpe ratio and similar measures matter more to a trader's career than the absolute number.