High frequency trading is the placing of very large numbers of orders in a short time: sometimes thousands of orders a second. This covers a wide range of automated proprietary trading strategies that rely on fast execution to be profitable. Some of these are controversial and have even lead to allegations of market abuse.
Most high frequency traders treat the workings of their systems as trade secrets, and different firms have developed different approaches, so it is difficult to comprehensively define or describe high frequency trading. In general if a strategy requires very high speed to succeed, then it is high frequency trading.
High frequency trading has become increasingly important in recent years, and, in the US at least, accounts for most equity turnover in times of high volatility.
Because speed is the key to profitability, high frequency trading systems are designed to react as fast as possible to new information. These systems are sometimes “co-located” in exchanges own data centers to achieve the lowest possible latency (roughly speaking, the time it takes for messages to travel to and from the exchange's systems). This provides additional revenues for stock exchanges. It is a controversial practice because it deliberately seeks to give some traders an advantage over others.
High frequency trading systems also need to get information from sources other than the exchanges as fast as possible. This has an impact on both the design of these systems, and on the systems of information providers who cater to this market.
High frequency trading also requires investment in fast computers, and high frequency traders are major buyers of new technology. Unlike most purchaser's they are willing to pay a high premium to have the latest and fastest, and are less concerned by drawbacks such as compatibility with existing software that deter many others. The problem for high frequency traders is that they are already faster than everyone except the other high frequency traders, so they are largely engaged in an arms race that is a zero-sum game.
Many of the tactics used by high frequency traders to gain tiny (usually sub-millisecond) advantages in speed have been controversial because they compromise the principals that all investors should have equal access to information and that all trades should be treated equally.
One allegedly unfair advantage has already been mentioned: the co-location of high frequency traders computers in an exchange's data centres. In the US, where securities are traded on multiple exchanges, high frequency traders would briefly be shown “flash trades“ before they were routed to other exchanges, but this practice is now being banned.
Another practice is pinging (like co-location, terminology drawn from computer networking). This is the issuing of lots of small orders (typically immediate or cancel orders) at different prices to find out price limits on orders that are not publicly visible (e.g. orders in dark pools). Once price limits have been identified, the high frequency trader trades ahead of of the orders in the dark pool: for example, buying on the markets to immediately sell to the buyer in the dark pool.
In general, the objection to high frequency traders is that they seek to exploit short term trends, and thereby profit at the expense of longer term investors.
High frequency traders argue that they are liquidity providers, and therefore play a similar role to market makers: their profit is a reward for providing liquidity, and that this liquidity benefits all investors.