High-Frequency Trading Simplified

Imagine you’re in a grocery store looking at apples. You like the look of them, note the price, and decide to grab 5 apples. But something strange happens. Right after you pick up the first apple, the price goes up. You get the first apple for the original price but have to pay more for the other 4. This is an analogy for the complaints against High-Frequency Traders (HFTs).

In stock market trading there has always been an advantage for those who can react fastest to new information. It may seem that the criticisms of HFTs by Michael Lewis and others is just the usual moaning by losers about their inability to compete with winners. But there is more to it than this.

Let’s get back to the grocery store and the apples. Remembering your first experience with the price of apples rising while you picked them up, you decide to just watch the price. You stand there for a long time and the price doesn’t move. The next day you watch the price of apples for a long time and again it doesn’t move. Finally, you reach out to scoop up 5 apples in one motion. But again the price jumps in the split-second between the first and second apple. What’s going on?

You know prices go up and down once in a while. This is a part of life. But in this case, you are the cause of the price increase. There appear to be a whole basket of apples being offered for a certain price, but for some reason you can only get one at that price.

At any point in time while stock markets are open, a certain number of Apple shares are offered for sale at a given price. Let’s use the example 5000 shares available for $119 each. You’d like to think you could get all 5000 shares at that price. However, many traders found that if they placed an order for all 5000 shares, they would get, say, the first 100 shares for $119, but the price would go up for the rest. It’s certainly possible to just have bad luck. With Apple trading so many shares second by second, it’s possible for someone else’s order to get in ahead of yours.

But let’s look at this at computer speeds. A computer’s clock or “heartbeat” is about a billion times faster than a human heartbeat. One of our seconds is like decades to a computer. To us, a display of all trades of Apple shares would go by as a blur. But to a computer, these trades arrive days or months apart.

At computer speeds, the time it takes for an order to get to a stock exchange is quite long. So, it’s not too unusual to place an order and then someone else’s order arrives first. But this still qualifies as somewhat of a coincidence. If it happens repeatedly, something strange must be going on. It’s not just bad luck that other orders are coming in first. Just placing an order for 5000 shares seems to be the cause of the competing orders. Just as reaching for 5 apples in the grocery store seemed to cause a price increase after the first apple, reaching for 5000 shares of Apple stock seems to cause the price to rise after the first 100 shares.

To understand what is actually going on, we need to understand that there are multiple exchanges where stocks trade. So, the 5000 Apple shares available for $119 is actually a collection of smaller numbers of shares available at different locations. When you order 5000 shares, your order gets split up into multiple orders from different exchanges.

The multiple orders do not all get to the different exchanges at the same time. The time differences are imperceptible to people, but are like weeks to a computer. Critics of HFTs say that when one small order gets to one exchange, HFTs see it and then race to the other exchanges to buy Apple shares ahead of the other pieces of the split-up order so they can resell the shares at a slightly higher price. Some call this “electronic front-running.” Many exchanges actually cooperate with HFTs and take a slice of their profits. HFTs are accused of many things, but electronic front-running is the main criticism.

The net effect is that if you’re an institutional investor who wants all 5000 shares available at $119, you can’t have them. Of course, it’s not that you’re somehow entitled to these shares. Someone else who acts first independently is allowed to get in there first. But it’s a different game when it’s your own order than causes most of the $119 shares to disappear.

It’s normal for traders to have insight into each other’s actions. The question is how fine-grained this knowledge should be. If the market offers 5000 Apple shares at $119, it seems reasonable for traders to learn that you bought all 5000 shares. Then they devise strategies based on the possibility that you may try to buy more. But allowing HFTs to respond 100 shares at a time makes a mockery of the notion that there were 5000 shares available at $119.

Big traders are used to what are known as “market impact costs.” If you’re trying to buy 100,000 shares of Apple, you should expect the price to rise before you’re done buying. But the question is whether market impact should kick in after the first 5000 shares in our example, or whether it should kick in after the first 100 shares.

In his critique of Michael Lewis’s book Flash Boys, Peter Kovac’s book Flash Boys: Not So Fast argues that big traders should expect market impact costs. He says the fact that sub-parts of trades affect the market “isn’t exactly revelatory.”

Applied to our example here, Kovac essentially argues that big traders should expect some market impact if the first 100 shares of a much larger order get traded first at one exchange. This is certainly true given the way markets work now. The question we need to answer is whether we prefer to have a stock market where market impact costs wouldn’t kick in until after buying the first 5000 shares that were advertised as available across all markets.

No matter what speed limits we might try to place on the stock market, there will always be some traders who are faster than others. The fundamental question is whether we want a market where the fracturing into many sub-markets is a critical part of making money from trading or whether we want a market where the current overall quotation for a given stock has some meaning.

Should individual investors be concerned about all of this? You can only lose money to HFTs when you trade. The dollar amounts at stake are very small on each trade. For buy-and-hold investors, this whole story is mostly a yawn. Even do-it-yourself investors who day-trade have more to worry about than losses to HFTs. The main concern is the effect of electronic front-running on institutional investors who trade big blocks of stock. So, if you own actively-managed mutual funds, HFTs may concern you.