Latency and high-frequency trading
High-frequency trading firms trade huge volumes, hold positions for seconds or less, and compete on speed measured in microseconds. They now account for a large share of equity and futures volume.
What HFT firms actually do
- Market making: quoting both sides across thousands of instruments, updating prices continuously (lesson 3.1).
- Arbitrage: keeping related prices in line, like an ETF and its holdings, or a future and its index.
- Latency arbitrage: reacting to a price change on one venue before slower participants can update quotes on another.
- Short-term statistical signals: patterns in order flow that predict the next few ticks.
The race for speed
If your signal decays in milliseconds, being slower than a competitor means trading at stale prices. Firms invest heavily to be faster:
- Co-location: placing servers in the same data centre as the exchange's matching engine.
- Faster routes: in 2010 a firm laid a new fibre cable between Chicago and New Jersey to shave milliseconds. Microwave and laser links soon beat it, because signals travel faster through air than glass.
- Hardware: custom chips (FPGAs) that process market data and send orders faster than software can.
Is it good or bad?
For: spreads and trading costs for ordinary investors have fallen dramatically, and liquidity in normal conditions is deep.
Against: liquidity can vanish in stress, as HFT firms widen quotes or withdraw. Critics argue latency arbitrage taxes slower institutional investors.
When it goes wrong
- The Flash Crash (6 May 2010): US stocks dropped nearly 10% in minutes and largely recovered. A large automated sell order met thinning liquidity as algorithms pulled back. Exchanges introduced circuit breakers afterwards.
- Knight Capital (1 August 2012): a faulty software deployment sent millions of unintended orders. Knight lost about $440 million in 45 minutes and had to be rescued.
Key takeaway
HFT turned speed into an edge and made trading cheaper for most investors, while concentrating risk in automated systems. For anyone else, the lesson is simple: don't compete on speed. Compete on time horizon and analysis.
Educational content only, not financial advice. Investing involves risk and you can lose money.