HFT under the microscope

A high-frequency trading firm can buy and sell in millionths of a second, which is faster than a human blink by a factor of about 300,000. That speed can make markets look deep in calm moments and then vanish when prices jump, which is the core warning in a recent YouTube explainer on the subject. (youtube.com) High-frequency trading usually means computers posting and canceling orders across exchanges so they can capture tiny price gaps measured in fractions of a cent. The firms spend heavily on microwave towers, fiber routes, and exchange colocation so their machines sit physically close to matching engines. (youtube.com) That matters because modern market-making often works like a grocery store constantly rewriting shelf prices while shoppers are still walking the aisle. If the data feed says the shelf price is stale, the machine pulls the quote first and asks questions later. (iex.io) The best-known failure came on May 6, 2010, when United States equity markets suffered the Flash Crash and the Dow Jones Industrial Average briefly plunged about 1,000 points before rebounding. A joint report by the Securities and Exchange Commission and Commodity Futures Trading Commission said a large automated sell program in E-mini S&P 500 futures helped trigger the chain reaction. (sec.gov) That report said high-frequency traders did not start the event, but they rapidly traded in ways that passed risk back and forth instead of absorbing it. In plain English, the fastest firms acted less like shock absorbers and more like a hot potato line. (sec.gov; cftc.gov) Regulators answered with market-access controls after 2010. Securities and Exchange Commission Rule 15c3-5 requires brokers with direct exchange access to maintain pre-trade risk controls designed to limit financial exposure and enforce regulatory compliance before orders hit the market. (sec.gov; ecfr.gov) Exchanges also started building brakes into the road. IEX says its 350-microsecond “speed bump” delays incoming and outgoing messages just enough to reduce trades against stale quotes, which is a direct attempt to blunt pure speed advantages. (iex.io) The argument in the new video is that this arms race can fool investors into mistaking execution engineering for investing skill. If one strategy wins mainly because its server is a few hundred microseconds closer to an exchange, the edge may disappear the moment liquidity thins or another firm buys a faster route. (youtube.com) That is why stress tests for systematic funds cannot stop at backtests of signals like value, momentum, or mean reversion. They also need scenarios where spreads gap wider, order books empty out, and the strategy discovers that its “liquidity” was only there in good weather. (finra.org; youtube.com) High-frequency trading is not automatically predatory, and official studies have long noted that these firms often supply day-to-day liquidity. The harder question is what happens in the 30 seconds when everyone’s model sees the same danger at once and every machine tries to leave through the same narrow door. (cftc.gov; sec.gov)