High Frequency Hyperbole, Part Deux
In our last episode...
Recently our Wall Street Journal op-ed, High Frequency Hyperbole, took on Michael Lewis's claim that the stock market is "rigged" against Main Street. We are not a high frequency trading (HFT) firm and don't have a dog in this fight, but most of the participants in this high-decibel debate are "talking their book," meaning arguing for their own interests. While some are just trying to sell their latest yarn, too many are intentionally scaring the public, lobbying for special advantages and calling for rule changes that could hurt investors by raising costs for all, undoubtedly declaring a victory for "fairness" in the process.
Were our comments, and those of others making similar points, exhaustive, conclusive and convincing to all? Sadly, no. The debate has not calmed or moved on to more constructive discussions of market microstructure as we suggested. While a number of large asset managers have now made statements supporting HFT, new charges against HFT continue to gain prominence. Some claim HFTs "make too much money with too much consistency (almost every day) not to be cheating." Others claim they engage in "front running" and have "rigged the market against retail investors." Another common accusation is that they provide "fake" liquidity and hence are much less useful than they might appear. It will shock no one to discover that we think these latest accusations are also largely off the mark. While the charges may not always be completely false - we do not, and did not, defend every trading practice by every participant done at high speed - they too often are misrepresented to provide support for self-serving, anti-competitive regulation, or simply leveled to market a new service (or best seller!).
Let's start with the easiest one, HFTs' supposedly excessive profits.
HFTs' profits are often held up as prima facie evidence of their guilt. As we have said before, many HFTs serve as market makers, and we believe it is a near certainty that HFTs make far less money in aggregate than market makers did in the lower-tech market structure of years past (you know, the structure the anti-HFT crowd writes elegies about). You would expect this as HFTs are viciously competitive with each other and don't enjoy monopolies like the specialists in the prior single-exchange market structure did. HFTs also face lower business costs (burled walnut reception desks and classy portraits on the wall cost more than computing power these days unless the computing power is also burled) than did the market makers of the old days. Combine extreme competition with lower costs and basic economics says they will charge less and make less. We think the publicly released financials of some of the major HFTs bear this out. Or consider that in 2000 many billions of dollars were spent by Wall Street firms to buy traditional, human market making firms like Spear, Leeds & Kellogg and Herzog Heine Geduld, whose operations now all have been replaced by lower cost and lower market value HFTs. HFTs have indeed brought creative destruction. As usual with such destruction it's the consumer, in this case the investor, who benefits, while the destroyed and the romantic write books about how the new world is unfair.
A common accusation is that HFTs must be cheating, as some HFT firms make money almost every day. If you are an investor trading against the bid-offer spread (a "liquidity taker" in the parlance) this would indeed be quite a feat. To pick an example close to home, we don't come close to profiting every day, nor do any of the most famous names in investing. In fact, for regular investors profiting a little more than half the days turns out to be pretty darn good. However, for an HFT providing liquidity, earning positive revenue almost every day is not surprising. It is incorrect to think of them as traders or investors making bets each day or to look only at their trading revenue.
Investors come to the market at different times to buy or sell and need market makers to be in the middle to facilitate their trading. While people have disparaged "middle men" since time immemorial, there is a very important reason they exist. Barter is hard to arrange and end buyers and sellers gain confidence from the price information of other transactions. As middle men, market makers provide a service and charge a fee, in this case the bid-offer spread. They can lose what they make from the bid-offer spread, and more, as they may have to unwind the transaction at a less favorable price.
When an HFT firm transacts by buying something at the bid, it would make the entire bid-offer spread if it could turn around and sell exactly what it bought at the concurrent offer price. But it often cannot. Usually it takes time to execute in this other direction, and often the market price moves against the HFT in the interim. In other words HFTs can easily get "picked off." "Picked off" means the market maker bought from someone who knew the price was likely to fall (or sold to someone who knew the price was likely to rise) either due to news or a flood of other orders in the same direction. Given that the HFT is providing liquidity, it is very difficult for it to perfectly protect itself against more informed investors.
On some transactions the HFTs lose, and on some they win. They have the built-in advantage of the bid-offer spread, but the game is biased against them due to being picked off at times. On average they make money, but not the full bid-offer spread, only a fraction. The secret to the consistent, daily profits of HFTs is not that they make money on almost all transactions, it's that they do a lot of transactions each day and rely on the law of large numbers. If they do enough of these transactions such that the average bid-offer spread they charge is higher than the average pick-off costs imposed on them, they can earn positive trading revenue almost on a daily basis, just like specialists and human upstairs market makers always did (people complaining about profit consistency leave out that this is not new with HFT).
Making money almost every day, even if it's by diversifying across an enormous number of slightly better than 50-50 bets, still sounds great, no? Well, it may be, but it's only available after an enormous investment in knowledge and systems, and even then it is not available at unlimited scale (if you are providing liquidity you can't provide more in aggregate than is desired). We'd like to make money every day, but we think there is a better use for our financial and intellectual capital than investing in creating HFT strategies.
Moreover, and more important, trading profits do not equal net profits. Saying HFTs make money every day is like pointing out that a grocery store's gross profits (sales revenue minus cost of goods sold) are positive every day. Now, the grocery store doesn't necessarily earn a net profit every day, or any day, as its gross profits may not be sufficient to cover salaries, rent, utilities, unsold goods and other costs. Likewise HFTs apparently regularly make profits on providing liquidity, what people call their trading profits, but still have to pay for the whole rest of their operation (admittedly no burling but lots and lots of computers and computer programmers, a little real estate, lots of cases of Jolt Cola bought on eBay, etc.) Consistency of gross profits is simply not particularly important in a world where only net really counts.
Legal front running!
Front running of the nefarious type typically relates to the use of information entrusted to you by a client to get ahead of the market. That is illegal and unethical because it betrays the client's interest. This is decidedly not what is going on with HFT.
Still, HFTs are often accused of "front running." Let's be clear, HFTs indeed try to do something like this, but all, or almost all (remember we make no claim to have checked every HFT's trading!), of it is legal, ethical, and only uses public information. It's also what traders of all types have always done since before computers, electronics, and the steam engine: guess at the direction of the market or a particular security and position themselves accordingly.
Common evidence of front running includes the oft-told story of an investor who sees shares offered at a price he likes on multiple venues and tries to buy them all at the current offer price. He finds after he buys the shares available on the first venue that the others are gone in an HFT minute (an HFT minute is a minute co-located to a New York minute). He claims that HFTs see his order on the first exchange, buy ahead of him on the other exchanges and then offer him the shares at a higher price. He gets mad, as he wanted them all at the price shown, and he becomes an HFT hater with a story. He makes accusations that exchanges provide faster access to HFTs that allow them to see his orders before they match at the exchange.
First, regardless of any speed advantage HFTs may have, they simply cannot see the orders that this investor sent. Anyone who understands trading knows that orders sent at marketable prices (a buy order at or above the offer price, a sell order at or below the bid price) are not published by U.S. stock exchanges. (Until 2009, some U.S. exchanges offered an order type called "flash" that allowed exchanges to disseminate information on marketable orders. They ceased the use of this order type in August 2009). Speed advantages can help HFTs get information faster than others, but it cannot help them get information that was never available to them. What HFTs and others do see are trades (matched quantities and prices) that occurred at the first exchange.
Second, if HFTs did buy ahead of this investor on the other exchanges, that would not be considered front running. The HFTs were only making educated guesses. They saw a trade and guessed that the buyer wanted more, or that more buyers would come to market. Feel one rain drop, buy umbrellas. If the HFTs had been wrong, they would have lost money. However, the actual event was likely even less nefarious, if less nefarious than zero is possible. Rather than an HFT buying ahead of this investor, it's more likely that other HFTs making markets on the other exchanges simply canceled their offers, again based on an educated guess.
Taking this a bit further, the HFT canceling offers on the other exchanges is often the same one who just sold the first batch. That is, the HFT had posted an offer to sell this stock on multiple exchanges. It's akin to putting an item you want to sell up on eBay and Craigslist. If you sell it on one, you remove it from the other (albeit not at the speed of light). You don't have two to sell, you never did, and you did nothing wrong in posting it in both places and removing it from one after you sold it on the other. But someone who wanted two of what you were selling might be annoyed. That's understandable but he has no legitimate beef. If he voices his annoyance loudly enough, however, he may convince a politician he's been wronged.
Most of what is being talked about as "front running" is really just the legal, ethical and economically beneficial practice of order anticipation, which is a fancy word for educated guessing. Yes, when HFTs, or market makers since the Stone Age (try to buy just a few flint arrows and you'd find cave-traders guessing you'd buy more and the price of other flint arrows rising) see buying, they try to figure out what that kind of buying means for the future and sometimes act on their guesses.
This accusation follows from the one above. HFTs often cancel their orders when they think that prices may move against them. There is a misconception that specialists in the old days stood out there with their limit orders ready to take losses, while HFTs are merely providing "fake" liquidity.
There are many problems with this accusation. First, being able to cancel orders is, and has always been, key to any market maker strategy, whether HFT or not. For example, specialists in the old days also canceled and lowered their bids when they saw selling pressure building up. The markets were opaque at those times so it was much harder to observe what went on. What was worse in those days is that specialists had "zero" latency by virtue of being right there at the booth, while for others it took tens of seconds to cancel their orders. In other words, the relative speed advantage of market makers in the old days was actually far larger than it is today, even if all the speeds today are faster. If and when large sell orders arrived in the market, specialists could cancel their bids, while the bids of everyone else would end up providing the liquidity to these sellers at adverse prices to the non-specialist buyers (often Main Street!) Similarly, HFT market makers try to cancel their orders today as well, but they are on a more level playing field. They also try to compete on latency with each other, but being faster today is a function of their investment in technology rather than a structural advantage as it was in the past.
Second, the practice of market makers guessing what kinds of orders are going to come to the market and adjusting their bids and offers is actually a good thing for investors. If that sounds counterintuitive, remember that the willingness to stand in the middle of buyers and sellers depends on an ability to not get "picked off" by informed traders, traders who hold valuable short-term price information, or at least not picked off so often as to wipe out their profits from the bid-offer spread. Computers may be super-fast, but they can still be pretty dumb versus a person (or another computer) that knows more.
In over the counter markets, for example many fixed-income markets, market design is different and market makers can protect themselves from getting picked off because they know the identity of the trader and the size of their orders. By knowing whom they are dealing with, they can offer cheap liquidity to small or uninformed traders and widen bid-offer spreads for large or informed traders. In contrast, equity exchanges are anonymous. Electronic market makers cannot tell whether they are providing liquidity to an informed or an uninformed trader, or whether they are dealing with a small piece of a large order.
An obvious choice for market makers in the equity markets would be to widen bid-offer spreads for all investors to get compensated for getting picked off by informed traders. That is exactly what you would expect in a monopolistic setting such as the NYSE with specialists acting as market makers. Unfortunately for HFTs (and fortunately for investors) in the current market structure, competition prevents that tactic. Widening the bid-offer spread will mean that the HFT will lose the order flow to market makers with tighter bid-offer spreads. In order to provide tight bid-offer spreads to investors, HFTs need to figure out how to distinguish (statistically, certainly not every time!) the orders of informed traders from the orders of other investors.
That is why, in the equity markets, HFT market makers turn to order anticipation. They try to figure out (again educated guessing, not knowing!) what kind of trail a typical informed trader leaves while executing his order. They know that since informed traders have valuable short term price information they execute their orders very quickly. For instance, if HFTs see a lot of trading at the offer in a short timespan they may anticipate that there is an informed buyer out there. They cancel their offers and make new offers at higher prices. To the extent that they can correctly anticipate the informed traders, they can compete with other market makers to provide a tighter bid-offer spread to the uninformed investors and win their order flow. In essence, the competition between HFTs occurs on two dimensions: those who can offer tighter bid-offer spreads and those who can anticipate orders better so as to get picked off less harmfully. This is exactly what classical microeconomics suggests.
It is difficult to sympathize with traders who complain that bid-offer spreads widen and prices move against them as they try to execute large orders. If these are informed traders, they can't claim a right to making money at a less informed party's loss (actually that's exactly what they often do, but they don't seem to phrase it quite like we do). If these are uninformed traders (as are most asset managers and individuals at the very short-term trading horizon of a market maker) then they shouldn't be rushing the execution of their orders in the first place as they lose little "alpha" by trading more slowly. By trading fast they may confuse the HFT market makers into thinking they are informed short-term traders, and they end up paying unnecessary transaction costs. It is both those whining about not being able to pick others' pockets as they did in the past, and those confused souls (including some very large, very well regarded confused souls) who are trading too quickly and hurting their clients for no good reason, who do most of the complaining in these cases. Neither has much of a case.
Finally, let's talk about the accusation that the market is rigged against Main Street, the retail investors buying 100 or 500 shares of a stock through a retail broker website such as Ameritrade or Schwab. In our view, this is the least sensible of the charges because retail investors are well protected and arguably have it better than anyone else in the U.S. stock market.
We already explained that the bid-offer spreads are at historic lows on exchanges at least partially because HFT market makers compete. Investors who can trade patiently, because they are executing small size and do not have fast-breaking or non-public information, get even tighter spreads due to HFTs' ability to anticipate informed order flow. Retail investors, who go through the retail brokers, actually pay less in spreads than what HFTs charge other investors at the exchanges! Yes, if the stock is trading at say $10.50 bid and $10.51 offer, a retail investor is likely to pay less than $10.51 when they buy (and get more than $10.50 when they sell) through these broker websites. This information is publicly available in the reports that brokers have to file with regulators.1
The reason why retail investors get such a sweet deal is that U.S. stock market design allows retail brokers to send orders from its customers to "wholesalers." Wholesalers are nothing but market makers, and they existed before HFT came into existence. There are many wholesalers who compete to get the flow privately from retail brokers and make much tighter markets than what HFT market makers are able to make on the exchanges. The wholesalers can make tighter markets because they are not afraid of getting picked off by a retail investor. While it may sound odd, basically making it clear you are not informed (in a very short-term way most people and asset managers aren't) is a big advantage as a market maker can offer you a very tight bid offer spread without fear you are picking them off.
What if a regulation is introduced to prevent retail brokers from sending the order flow privately to these wholesalers? It is pretty straightforward to see what will happen. This flow will make its way to exchanges.2 The addition of more benign flow on exchanges will decrease the spreads even more. Large asset managers like us will benefit, but the retail investor will pay more in spreads than they do today, because their uninformed orders will now be indistinguishably mixed in with our order flow. The cost of retail brokers will go up as well: they get paid by the wholesalers in return for sending them order flow, and instead they will have to pay exchanges. That additional cost will likely get passed on to retail investors in terms of higher commissions. So for both reasons retail investors will pay more.
As it stands, HFTs provide tight spreads on exchanges, and retail investors circumvent them completely by trading directly with wholesalers. On the other hand, retail investors indirectly enjoy these services provided by HFTs because wholesalers are required to provide the same or better spreads than what's available on the exchanges. Ironically, US equity market design is indeed "rigged," but in favor of small retail investors.
Bringing it all together
In summary, we don't believe HFT profits are excessive or excessively consistent. We censure illegal front running as strongly as anyone, but it has near nothing to do with HFT per se. Canceling orders in the process of providing liquidity is key to any sort of market making, whether HFT or not. We support the right of HFTs, or anyone, to try to guess the direction of the market, using order flow or any other public information. We not only support the right, we celebrate the successful exercise of that right as it adds to public welfare by making markets more efficient and lowering the cost of investing. Lastly, we believe markets are "rigged" in favor of, not against, retail investors.
All four authors are with AQR Management LLC where Asness is Founding Principal and Managing Principal, Brown is Managing Director and Chief Risk Officer, Mendelson is a Principal, and Mittal is Managing Director and Head of Trading.
 Brokers are required to disclose certain order execution information under SEC rule 605 and SEC rule 606.
 Retail brokers typically send marketable and market orders to wholesalers and passive limit orders to exchanges.