Andrew Keller’s article Robocops: Regulating High Frequency Trading After the Flash Crash of 2010:
This emphasis on speed is the primary defining aspect to high frequency trading: it is the main difference between traditional investment management and HFT. It also distinguishes HFT from other algorithmic trading strategies. HFT is a subset of algorithmic trading, where both use programmed algorithms to execute automated order submissions and automated order management. However, it is common for a non-HFT algorithmic strategy to hold traded securities for days, weeks or months, whereas HFT traders usually end the trading day flat, with no significant holdings. Furthermore, ultra-fast trading speeds are not necessary in a non-HFT algorithmic strategy; HFT, on the other hand, uses strategies that require speed to gain advantages in the market.
“smoking” is an HFT scheme that exploits slow traders by offering attractive limit orders, then quickly revising these prices to take advantage of an unsuspecting slow trader’s market order.
Presumably, the exchanges know the players and have a better starting point than government regulators in attempting to understand HFT methods and strategies. HFT computers are also tied directly into the exchanges’ computer systems which provide the exchanges an advantage in compiling data. On the other hand, the exchanges have strong incentives to provide free reign to HFT traders: they earn high rents from co-location, and significant fees from large amounts of trading. The exchanges want HFT traders to continue playing a significant role in the markets, and may not be a reliable regulator.