When To Requote

A market maker's quote sits in a queue at a given price, and every exchange rewards being early in that queue with priority for the next fill. Cancelling and replacing a quote — to reflect new information, adjust for inventory, or simply widen out — sends it to the back of a new queue. A quote-update policy is the rule set that decides when that trade-off is worth it: react fast enough to avoid being run over by a moving market, but not so often that you're constantly sacrificing queue position and paying cancellation fees for adjustments that didn't need to happen.

The core trade-off

Requoting too slowly leaves a stale price on the book — exactly the situation a faster competitor exploits by trading against it before the market maker can react, a losing trade known as adverse selection. Requoting too fast burns through queue priority and, on some venues, message fees or order-to-trade ratio penalties, for changes that may not have mattered much. A sensible policy typically only triggers a requote when something crosses a meaningful threshold: the reference price (say, a correlated future or the midpoint) moves more than a set amount, the market maker's own inventory crosses a limit, or a competing quote at a better price appears and needs to be matched or ceded to.

Worked example

A market maker quotes a stock with a threshold rule: requote only if the reference midpoint moves by more than half a tick, or if inventory changes by more than 500 shares from a fill. A midpoint drift of a quarter-tick doesn't trigger anything — the quote stays put, keeping its place in the queue. A half-tick move does trigger a requote, even though it costs queue priority, because leaving the old price up risks a much larger adverse fill against a market that's clearly moved.

Other triggers besides price

Price drift isn't the only reason to requote. A market maker also watches its own inventory: after a string of fills on one side, it typically skews its quotes to encourage trades that bring inventory back toward flat, which requires a fresh quote even if the reference price hasn't moved at all. Scheduled events matter too — many desks widen or pull quotes automatically a few seconds ahead of a known news release or auction, then requote once the immediate volatility has passed, rather than trying to react in real time to a burst of repricing that a threshold rule alone would handle badly.

What this means in practice

Getting the thresholds right is close to the central design problem in automated market making: too tight and the strategy churns its own quotes into oblivion on fees and lost priority; too loose and it eats adverse-selection losses whenever the market moves faster than the policy reacts. Firms tune these thresholds empirically against their own fill and adverse-selection data rather than picking a single universal rule, because the right balance depends heavily on how fast a given instrument actually moves, how competitive the queue is, and how expensive cancellations are on that particular venue.