Flickering Quotes And Quote Lifetimes
Some quotes in the order book appear and disappear within milliseconds, and how long a quote actually survives before being canceled tells you a lot about whether it represents genuine trading interest or a fleeting probe.
A "flickering" quote is one that appears in the order book and is canceled again within a very short time, often milliseconds — sometimes before any human trader could even see it on a screen, let alone react to it. Quote lifetime — how long an order actually sits in the book before being canceled or filled — is a direct measure of this, and it varies enormously across participants: a resting limit order from a slower participant might live for seconds or minutes, while a high-frequency market maker's quote can be posted and pulled within a few milliseconds in response to a signal elsewhere.
Very short quote lifetimes are not inherently manipulative — much of it is legitimate market making constantly adjusting to new information, canceling a quote the instant it becomes stale rather than risk being picked off. But flickering can also be used strategically: posting and immediately canceling a quote to test for hidden liquidity, or to create a fleeting appearance of depth that other algorithms react to before it vanishes, both of which regulators watch for as potential quote-stuffing or spoofing patterns.
For a quant analyzing microstructure data, quote lifetime distributions are diagnostic: a market with a large share of sub-10-millisecond quotes signals heavy high-frequency participation and suggests displayed depth is unreliable — an algorithm trying to trade against the visible book should expect a meaningful fraction of that liquidity to vanish before an order can reach it, which is exactly why execution algorithms often discount displayed size by an estimate of how much is likely to flicker away.
Quote lifetime — how long an order survives before cancellation — separates genuine trading interest from fleeting probes; markets with heavy flickering have unreliable displayed depth, which execution algorithms need to discount for rather than take at face value.
Practice in interviews
Further reading
- O'Hara, High Frequency Market Microstructure (2015)