How to Understand High Frequency Trading

High Frequency Trading: You might not have heard of it now, but get ready, because it's going to be everywhere in the next few weeks. After the New York Times' exposed how advanced Wall Street computers can execute impossibly complicated, and possibly unfair, stock trades, Sen. Chuck Schumer (D-NY) asked the SEC to ban certain types of high frequency trading, or HFT. But what is HFT exactly, how does it work?

There is still a lot of uncertainty about how exactly these trades are executed. As the Times reported in a very clear graphic, computers can "peek" at trade orders 0.3 seconds before they are executed, and actually buy the stock milliseconds before it goes up in value. During this process they "ping" (or listen to) the prices, learning how much people are willing to pay by making a flash bid -- a bid to see the maximum price other buyers are willing to pay.

How does that work? Here's a down-and-dirty analogy in which I'll play the role of the flash trader. Imagine if eBay had a rule where you could cancel your bid within 1 second. I put up some stuff on ebay, and you place a bid for it. Then I place a bid that is higher than the current bid to see if that becomes the new highest bid. If it is, I cancel it within milliseconds. Remember, I don't want to buy the product -- I just want to drive the price higher! This is similar to what critics of HFT think is going on; HFT is able to ping prices with bids that exist for only milliseconds to see how much other buyers are willing to pay to squeeze out the maximum profit.

There's going to be a lot of terminology thrown around when people talk about this, and rightfully so since it is a very technical issue. But the economic ideas are simple, and I want to give you the water cooler guide to thinking through the relevant issues in the debate. When you hear discussion about this going forward, think of how the points made influence these concepts.

Information Asymmetries

As you can see from the quick eBay analogy above, being able to see someone else's information, when they can't see yours, makes for terrible markets. Beyond issues of fairness and equity, which are incredibly relevant here, this also leads to an issue of bad prices. The information reflects only part of what you believed about the stock in question; computers took advantage of your situation, but the information that others see as a result of the front-running doesn't reflect what you actually believed. You hear stories about stock prices jumping all kinds of crazy values, with crazy volume numbers and volatility, because of a few stock purchases, and these price movements reflect the HFT. Now remember that the feedback mechanism of stock prices - if you are a Hayekian - is the whole point of having a market. If trading in markets doesn't aggregate information among many diverse parties but instead turns the price mechanism into a roulette wheel played out by supercomputers ransacking your 401(k) - because believe me, your 401(k) is a great target - what's the point? Does that have any more information than the tyrant social planner?


The strongest claim in favor of HFT is that it is providing liquidity to the financial markets. More liquidity in markets is usually considered to be a good thing, and as such they are being rewarded for providing a service. There's a problem with this though - they have no obligation to provide liquidity. There's no formalized procedure in which they post prices with certain time limits. They, and the HFT practitioners have been very upfront about this, have no obligations when it comes to providing liquidity. And it is very likely that they won't step in when the financial markets need them the most.

So in this sense, liquidity from HFT is like having an airbag in your car that works all the time except when you are in a car accident. It isn't there when you most need it, and it encourges you to drive a bit faster, and take turns a bit sharper, because you think you are protected. So instead of being a risk management tool everyone is aware of, it is instead misinformation.

Power of the Market

Won't markets take care of this by themselves? Remember that there will be brutal competition to get the trades to run faster, hit more stocks and spin more money out of consumers. Even though more players can get into this, that just means there will be more players ripping off the information of other players. The issue isn't that this is concentrated among a few big market players (though that is an issue) but that it is happening to our investments.

In fact, a worrisome trend is traders who are looking to attack HFT with side moves. There is an influential white paper by Themis Trading called Toxic Equity Trading Order Flow On Wall Street. Themis calls them program traders, and their job is to push prices in such a way that it causes a cascade of computers to start going into motion. Now what is happening? There is no actual change in the fundamentals of the stock in question, but prices are moving dramatically. This hurts the ability to learn real information in the feedback mechanism in prices, which gets us back to the question of what we want prices to do in a market.

Momentum Effects

In the UChicago efficient financial markets theory, markets are efficient because, if prices diverge from fundamental values, smart arbitrageurs will always jump in to bring prices back into line. But what if it is more profitable to drive prices further away from fundamental value? This is likely in theory, and for a variety of reasons we believe there are limits to what arbitrage is capable of accomplishing in bringing prices back into line anyway. HFT, instead of bring prices to a fundamental value, brings them to the edge of what a person is willing to pay. This influences other computers, and other traders, who react accordingly. I've spoken to a few quants, and I agree with their assessment that HFT can exacerbate momentum effects. I'm willing to be convinced otherwise, but on the first approximation instead of converging prices to a fundamental value, it seems likely that these computers are actually driving them away, and away in such a way where much of cutting-edge finance theory thinks there is a big weak spot in how financial markets work to correct bad prices.

Hasn't It Always Been This Way?

It is very reasonable to ask "Isn't this same as it ever was, plus computers?" This is very similar to what specialists used to be able to do with the similar "the look" (peeking at someone's information) before the NYSE cracked down on them. In those cases though, "the look" was approximations, not the highly organized front-running of limit prices down to the penny. And the specialists were legally required to provide liquidity in specific ways. They were providing a public good, liquidity in financial markets, and were perhaps getting rewarded too well for it. The HFT in this case don't have the requirements; they can pull their liquidity from the market at any time, and they'll pull it at exactly the moment we need it the most.

So as the debate unfolds, remember to ask yourself, (1) whose information is being exploited by whom and how, (2) does this make financial markets stronger and more efficient - say by providing liquidity - during a downturn when markets need them the most, and (3) what is this doing to the price mechanism - is it helping prices converge to fundamental values or driving them further away? The evidence currently looks like HFT is doing bad things on all three accounts.