How does High Frequency Trading work?

HFT has been in the news more, thanks in part to Michael Lewis’ new book, Flash Boys. This article presents a simple explanation of how and why high frequency trading works. Nearly a third of the market is traded by machines against their human counterparts.

India’s largest stock exchange National Stock Exchange (NSE) says that nearly a third of trades in India are done through algorithmic trading. In the US such trades accounts for over 70 per cent of the volume while in the Europe it is approximately 40 per cent of the volume.

High frequency traders try to profit from the price movements caused by large institutional trades. When a mutual fund sells a million shares of a stock, the price dips—and HFTs buys on the dip, hoping to be able to sell the shares a few minutes later at the normal price. Likewise, when a pension fund buys two million shares, the HFTs short-sell the stock, hoping to close their position at a profit. (Short selling is selling stock you don’t own; you borrow the shares from a stockbroker, sell them, and then later buy the stock to return the borrowed shares.)

Knowing the giant in the market

High frequency traders use computers to process electronic data feeds from numerous data points to make a trading decision. These decisions are then converted into orders which can be executed not in micro-seconds but milli-seconds. Making decisions based on so much information and at such speed is humanly impossible. Thankfully most of the programs compete against each other for their share of the profit.

The process or the logic of collecting and disseminating information from various sources is programmed into the computer and hence the name algorithmic trading or more popularly known as algo trading or program trading. Such style of trading has got some bad press as every time market witnesses a sharp and quick fall, they are blamed for it. There could be some truth in that but it is not deliberate on the part of algorithmic programs. Rather it has more to do with logic behind the program which gets triggered to capitalise on the pricing aberration. To understand it better we will have to understand various logics or the strategies on which algorithms are written.

Arbitrage trading

Machines have replaced arbitragers in most of the broking houses in India who used to run large desks to take advantage of the price mismatch between various exchanges or cash market and futures market. Various computer arbitrage strategies are now being performed by computers leaving little on the table for those who still operate manually. One of the biggest brokers in India with over 700 arbitragers has now reduced his team to less than 10 and one computer.

Market making

Undoubtedly one of the main reasons for the proliferation of HFTs globally is because various exchanges used their services to provide liquidity and they pay them to do so. Higher liquidity attracts other traders and institution players to the exchange as the transaction or impact cost is lowered. Market making is one of the simplest forms of trading strategy which involves placing a buy and sell order in order to capitalise on the bid-ask spread. As soon as one leg of the order gets filled the program covers the position with a small profit. HFTs provide the market depth and reduce the cost of trading for other participants in the market.

Tape trading

One of the biggest advantages that HFTs have is they collect information of volume and quotes and decipher it to predict where stock movements are expected in terms of institutional order flow. Not only do the programs read the actual trades that are taking place but they also peep into the order books displayed by the exchanges to arrive at their decision. Thus as soon as an institution places a big order to buy or sell the algorithm gets triggered and jumps in line ahead of the order of the fund. In the end the institution ends up either buying higher or selling lower from the algo trader. Some exchanges have however curtailed such trades by giving institutions a dedicated trade execution window through which the algo trader cannot peep.

News based trading

Algorithms are now being programmed to interpret news and act smartly much before humans can react. Thus let’s say if the RBI governor does not cut interest rate, an algorithm which is scanning various news feeds will read, interpret and act in milliseconds even before a human can finish reading the headlines.

Low latency strategies

These strategies are straight out of a sci-fi movie. Various algo funds/traders have setup shop close to the stock exchanges in order to cut down on time the signal takes to travel through the wires. High end computers and other equipment’s now minimise the time taken for the price quotes to reach the computer of the algo traders. The intensity and speed of the trades can judged from the fact that Italy has imposed a tax on equity transactions lasting less than 0.5 seconds. Where win and loss can be decided in a matter of milli seconds, no compromises are made for getting the latest gadgets.

Do high-frequency trades affect individual investors?

Computers are clearly at an advantage when compared to an average trader as it can not only gather and analyse more information but also because it can see through more than an average person. However, their use has helped increase volume in exchanges across the world and reduces impact/transaction cost.

For a retail investor this offers relief as his cost of getting in and out is reduced. But for an institutional investor, who is the main target segment of the high frequency trader along with his peers, an algo trader is chipping away profits from entry and exit of these funds.