High Frequency Trading Competes on Microseconds Nobody Else Notices
Firms invest heavily to be marginally faster than each other. The debate about whether this helps or harms markets turns on which specific strategy is being described.
What the Term Covers
High frequency trading is not one strategy. It describes automated trading operating on very short timescales, and the activities grouped under it differ substantially in their effect.
| Activity | What it does | Effect on other participants |
|---|---|---|
| Electronic market making | Quotes both sides continuously | Narrower spreads, more liquidity |
| Statistical arbitrage | Exploits short lived price relationships | Broadly neutral, improves pricing |
| Latency arbitrage | Trades on stale quotes elsewhere | Extracts from slower participants |
| Order anticipation | Detects large orders and trades ahead | Raises institutional execution costs |
Arguing about whether high frequency trading is good or bad without specifying which activity is being discussed produces an argument nobody can resolve.
The Speed Race
Firms compete on latency: microwave and laser networks between financial centres because radio travels faster through air than light through fibre, colocated servers metres from the matching engine, and custom hardware that processes market data in hardware rather than software.
The economic logic is that the fastest participant captures the opportunity and everyone else gets nothing. This produces a winner takes most structure where marginal speed advantages have disproportionate value.
The social criticism is that enormous resources are consumed producing an advantage measured in microseconds, which delivers no benefit to anyone outside the competition. That criticism is reasonable for latency arbitrage specifically and weaker for market making, where speed genuinely allows tighter quoting.
Why Market Making Improved
The clearest measurable effect of electronic trading is that bid ask spreads narrowed substantially compared to the era of human market makers.
The mechanism is straightforward. A market maker who can update quotes in microseconds carries less risk of being caught by a price move, and lower risk permits tighter quotes.
For an ordinary investor buying a liquid stock, execution costs are considerably lower than they were. That is a real benefit and it is frequently omitted from criticism of the industry.
Where the Criticism Holds
Latency arbitrage is the harder case to defend. When a price changes on one venue, a fast participant can trade against stale quotes on another venue before those quotes update.
The profit comes directly from participants who were slower, and no price discovery is improved by it, since the information was already public. It is a tax on being slow.
Order anticipation is similarly contested. Detecting that a large institutional order is being worked, and trading ahead of the remainder, raises the cost of that institution execution and ultimately of the funds it manages.
The Structural Responses
Several venues introduced speed bumps: a deliberate delay of a few hundred microseconds applied to incoming orders, which removes the value of a marginal speed advantage.
Frequent batch auctions are the more radical proposal: rather than processing orders continuously, collect them over a short interval and match them all at a single price. That converts the competition from being fastest to offering the best price, which is arguably what the market is for.
Adoption has been partial, since venues compete for volume and any structure that discourages the highest volume participants is commercially difficult.
The Fragility Question
Automated market makers withdraw when conditions become uncertain, because quoting into a market you do not understand is how a market maker loses badly.
The result is that liquidity can be plentiful in normal conditions and disappear in seconds when it is most needed. The 2010 flash crash demonstrated this, and the general pattern has recurred in smaller episodes since.
Liquidity provided by participants with no obligation to provide it is genuinely available and it is not dependable, and those are different things.
The Bottom Line
High frequency trading covers activities with opposite effects. Electronic market making narrowed spreads and lowered costs for ordinary investors. Latency arbitrage and order anticipation extract value from slower participants without improving prices. Speed bumps and batch auctions attack the second without damaging the first, and the liquidity provided is real but withdraws exactly when it is most wanted.