HFT: How fast can you really trade?

Over the past two years, public debate has increased on whether to regulate the maximum frequency of trading. The idea of regulation stems from a perception that the so-called high-frequency traders can move in and out of positions in milliseconds, “flipping” capital from one stock to another using market orders before the wider population of traders barely can blink. This perceived high speed of trading has been thought by some to disadvantage traders without access to high-speed technology, and regulators have been called in to level such inequalities.

The “frequency” in high-frequency trading (HFT) refers to the number of times a trader or a computer program places a trading order within a certain period of time. The higher the trading frequency, the shorter the time duration between orders. In the popular press, high-frequency trading often is described as moving money in and out of positions within milli- and micro-seconds. Traders who are registered market makers are allowed to place simultaneous buy and sell orders on both sides of the market price, and may get “hit” within a tiny fraction of a second, regardless of whether the traders are human or automated. As this article shows, however, market participants that are not registered market makers generally are prevented from profitably trading at frequencies faster than 1 second by structure of the markets.

A trader who is not a registered market maker is generally disallowed from placing simultaneous bids and offers on the same security at the same time. Instead, this trader is typically limited to a sequential trading pattern: A buy followed by a sell, followed by a buy, etc. Independent of trading frequency, the key to profitability in such sequential trading is to buy at a low price, on average, and to sell at a higher average price, whether using limit or market orders.

While there has been quite a bit of talk about high-frequency traders trying to detect underlying trading algorithms, such as iceberg orders and stops, and then front-run those orders, the main activity of HFT is speed. While some such algorithms exist, the majority of high-frequency traders are making a bet like everyone else and attempting to gain an edge through speed. These players are closest to the old floor local scalpers who constantly would attempt to buy on the bid and sell on the offer. Doing this, however, is not an easy proposition. If you manage to get an edge on one end, you are not guaranteed a profit. The main challenge at high frequencies is to overcome the bid-ask spread and the transaction costs that can be large relative to market moves. For a trader to profit within a specific time interval in the presence of transaction costs and the absence of liquidity rebates, the maximum best bid achieved during the time interval has to rise above the minimum best offer registered during the same interval, generating a profit potential. If the trader is acting on perceived order flow, it is more likely that he is attempting to buy the offer before or as it goes bid and then take his chances on the ability to sell on the offer. So he is acting on the same information as everyone else — rather than having an information advantage — and trying to beat everyone else with speed.

The case of the best bid exceeding the best offer during a trading interval does not guarantee profitability, but is just a minimum condition for a possible success. Discerning whether the markets are about to move up or down is the key to profitability and a true test of the trader’s analytical abilities. However, regardless of how well the trader can predict future price movements, if the markets do not move within a certain time interval, the non-market-making trader cannot profit.