Latency arbitrage
Also called latency trading, stale-quote arbitrage, speed arbitrage
Profiting from speed: trading on a price change before slower quotes or price feeds catch up. An FCA study of FTSE 100 stocks found such races last millionths of a second and make up about a fifth of trading volume.

On this page· 6 sections
When a price changes in one place, it takes a moment to change everywhere else. Some firms race to trade in that moment, and the winner is decided in millionths of a second[1].
Latency arbitrage is trading on a price change before slower quotes catch up. Regulators and economists study it because it adds to the cost of trading for other participants[1][2].
How it works
A resting quote becomes stale when the price it was based on has already moved. Fast traders try to snipe it, meaning to trade against it before its owner can cancel or update it[1]. Several of them often try at once. Some succeed and some fail, and the failures are what made the practice measurable[1].
Budish, Cramton and Shim traced the root cause to market design. Exchanges that treat time as continuous and process orders one at a time turn even public information into arbitrage profits for the fastest[1][2]. In their data, competition did not change the size or frequency of these opportunities. It only raised the speed needed to capture them[2].
The UK Financial Conduct Authority (FCA) published Occasional Paper 50 in January 2020. It used message data from the London Stock Exchange for FTSE 350 stocks over nine weeks in fall 2015[1]. It found[1]:
- about one race per minute for the average FTSE 100 stock, around 537 a day;
- a typical winning margin of 5 to 10 millionths of a second;
- about 20% of trading volume inside races;
- the top 3 firms winning about 55% of races.
Each race was worth a small amount, about half a tick or roughly 2 GBP on average[1]. The paper estimates that removing latency arbitrage would cut the cost of trading by 17%[1].
Budish, Cramton and Shim propose frequent batch auctions, run for example every tenth of a second, so that tiny speed advantages count for less[2].
Example
Say a stock is quoted 50.00 bid and 50.02 ask on one venue. A related index future jumps, and a fair price for the stock becomes 50.05.
- A fast firm buys 1,000 shares at 50.02 before the quote moves.
- The quote then rises to 50.05. The shares are worth 50.05 − 50.02 = 0.03 more each.
- The gain is 1,000 × $0.03 = $30. The seller who was slow to update gave up that $30.
Real races last microseconds, and the FCA found the average one worth about 2 GBP[1]. The mechanism is the same.
Watch out
The FCA paper is about equities on one exchange in 2015. Its figures should not be read as current or as applying to other markets[1].
Latency arbitrage is one use of speed. The model separates it from liquidity provision, which the authors describe as useful[1].
The paper opens with two opposed claims, that the market is “rigged” and that widespread latency arbitrage is a “myth”. Its aim was to measure the size[1].
At TRONEXO
On the simulated Challenge accounts and TRONEXO Accounts, latency arbitrage and price-feed arbitrage are prohibited, as is profiting from latency, network lag or server-side delays. See Funding models & rules and the FAQ.
Sources
- RegulatorFinancial Conduct Authority (FCA)Occasional Paper 50: Quantifying the High-Frequency Trading “Arms Race”: A Simple New Methodology and Estimates (Aquilina, Budish, O'Neill, 2020)Checked November 9, 2026
- ResearchRePEc IDEAS (abstract record)The High-Frequency Trading Arms Race: Frequent Batch Auctions as a Market Design Response (Budish, Cramton, Shim; QJE 130(4), 2015)Checked November 9, 2026
