High frequency trading (HFT), which is like day-trading on near fatal doses of amphetamines, sometimes seems like an investment fringe activity. In reality, it is central to how we trade now. Estimates vary, but Andy Nybo of Tabb Group estimates that HFT strategies account for 61% of trading activity in U.S. equity markets (see “In the majority”). CME Group reported on their fourth quarter earnings conference call that 43% of the trading volume on CBOT came from proprietary trading firms, primarily algorithmic, which probably means high frequency trading. Over recent years, almost every measure of HFT on just about every exchange shows it going up — fast. Perhaps more disturbing to some is that a very small group of participants are making up more than half the volume (see “Crowding everyone else out”).
Individual trades vary in duration. High frequency trades are those on the very short end of this spectrum. There is no rule as to when, exactly, a trade lasts so long that it can’t be counted as high frequency. Presumably, if you have to ask, you are not quick-witted enough to understand the answer. As a rule of thumb, however, it seems any trading plan where trades last longer than a second is something else — day-trading, perhaps.
One reason why it is difficult to give an upper boundary to high frequency trades is that high frequency traders do not think about how long their trades last. They think, rather, in terms of market structure and market economics. Market economics change as trade duration lengthens. Here we will analyze high frequency trades in terms of the ways they exploit market structures of various types. Roughly speaking, we will also be starting at the shorter end of the spectrum and working our way up.
Perhaps the least defendable form of HFT is something called flash trading, at least pertaining to equities. Under SEC-regulation NMS rules, exchanges must route buy orders to other exchanges when that exchange is offering a better price. In which case, the exchange loses the transaction fee, which it does not want to do. Flash orders are, in part, an attempt to end run this rule. If an exchange allows flash orders, it displays the order for 500 milliseconds before routing it elsewhere, which means that for a twentieth of a second, certain traders have a monopoly on trading information.
That monopoly is pure gold. Assume your research tells you that Hatfield and McCoy Conflict Resolution Consultants, which is currently trading at $50 a share, is a buy at any price up to $50.20. You put in a buy order for 50,000 shares of Hatfield and McCoy with a $50.20 limit. Quick Joey Small’s software reads your order and puts in an IOC (immediate or cancel) order to sell you 100 shares at $50.01 and your software eats it. “Hmmm,” says one of Quick Joey’s subroutines to another, and puts in an order to sell you another 100 shares of Hatfield and McCoy at $50.02 and, again, you get your shares. Quick Joey sells you shares in one hundred lots all the way up to $50.20. When your software balks at $50.21 a share, Quick Joey sells you every share you will buy at $50.20 a share under the assumption that as soon as they take care of you, the price of Hatfield and McCoy will drop to $50 a share. Many traders have noticed that prices seem to shoot through their stops much more frequently now than they used to. This is why.
Incidentally, in the old days, IOC orders used to be called FOK orders, which stood for fill or kill. Reverting back to older and, in many ways more honest, terminology, you just got FOKed.
However, in the options world, eliminating flash orders would create added cost to customers if their orders are redirected to another exchange and that exchange has a maker taker pricing structure. If a customer has 200 lots to sell at the NBBO (National best bid/offer) and gets 100 done at one exchange exhausting the bid, the remainder would be flashed to market makers before they are sent to another exchange. If that option is taken away and the order is sent to another exchange that charges a taker fee and is filled at the NBBO, the customer is not better off.
It is important to understand that Quick Joey Small took your money not because he is smarter or even because he is faster than you are. He took your money because Joey and his cohorts have a monopoly; they have non-public information specifically about customer orders. If your order had been broadcast widely, you would have gotten better fills.
Among other things, the history of the markets is the history of insiders, floor traders, for example, taking advantage of outsiders, which would be you. At one time, for example, market prices would scrawl across your television screen with a 15-minute delay. There was no technical reason for this, but the delay gave insiders a 15-minute advantage and that was reason enough. These days the delay is a fraction of a second, but that is enough to make some people very rich (see “Low latency = high profits”). And more to the point, it is a way to take money out of your pocket.
Should HFT be banned?
The key to such HFT is speed, and to that end exchanges provide on-site computer hosting called co-location. Through vendors, FCMs and eventually end users can route their orders directly to servers located in the same location as the exchange matching engine, eliminating latency. Exchanges point out that anyone can access this service provided that they pay the fee. Still some argue it gives some market participants an unfair advantage.
Many traders argue that high frequency trading, even when it doesn’t involve flash orders, invariably involves front running and, therefore, should be illegal.
However, front running needs to be defined. In a discussion on high frequency trading from Futures Industry magazine’s January 2010 issue, Don Wilson of DRW argues that HFT is not front running. He said front running is “acting on non-public information and, more specifically, on customer orders that have not yet been made public to the marketplace.” He added that because HFT firms don’t have customer orders, they can’t front run.
Given Wilson’s definitions, he’s right, of course. But so what? One chairman of the board of a large hedge fund who asked for anonymity argues this way, “The FCM and the exchange have an obligation to execute my orders in the most efficient manner possible. And what that means is that if my order is entered first, I am expecting it to be executed first. What they are doing is granting someone a right to go in front of me, which I think is fundamentally wrong. Some exchanges, such as the [Intercontinental Exchange], execute on a first in, first out basis. Other exchanges do not,” he says.
Still, even without flash orders, even if all orders on all of the exchanges are executed first in, first, out, it is not clear that high frequency trading is good for anyone other than high frequency traders. Again, the hedge fund chairman says, “People who have this business say, ‘I’ll get you this machine, too.’ All this means is that if I get the machine I’m screwing other traders, too. And if everyone had it, they wouldn’t have a business… They’re gaming the system.”
Rick Redding, CME Group managing director, says that everyone has the ability to co-locate servers to get the fastest execution times. “It actually levels the playing field if people pay for it,” Redding says.
Economists distinguish profits earned by innovation from profits earned just by owning the resources, which they call “rents.” What an economist would call “rent” includes things you and I wouldn’t, such as the cost of labor. In high frequency trading, some of the profits are nothing more than rents. What is being rented here is speed. If the price of gold on the Comex floor goes slightly out of line with the price of gold on the CME Group’s Globex platform, the first trading firm to identify the misalignment and push prices back in line earns 100% of the rent from being first. To you or I, it may mean nothing or less than nothing whether this problem is fixed in five milliseconds or one millisecond or one hundredth of a millisecond, but it means everything to the competing traders.
The reader should think of this as an information technology arms race. Strength, smarts and insight do not matter. Whoever moves fastest wins.
This has led the exchanges to provide co-location. For this kind of trading, location matters a great deal or, rather distance matters. Failing the kind of technical breakthroughs that win Nobel prizes, information cannot travel faster than light, which, in turn, cannot travel faster than 186,282 miles per second. This means that if your hardware is located a mere 100 miles from the exchange’s computers, it can’t respond faster than once a millisecond. Actually, a one-millisecond response time assumes that the message is sent in vacuum and that certain technical problems have gone away. You can’t make certain trading strategies work from more than 100 miles away. This gives the companies that co-locate a considerable advantage.
High frequency traders invariably note that anyone can get their hardware placed next to the exchanges; they just have to pay the fee. This is true but it is not also clear that it is relevant. The Standard co-location package for ICE alone is $4,000 for installation and then $3,500 per month. This gives large traders a serious advantage.
While it can be argued that certain trades have an advantage, in the old floor days customers would complain about being at a disadvantage to locals and invariably the answer that came was that is why they paid the half million or so for the membership. Within the pits, location was still key. Taller brokers had an advantage. Those not blessed with height would buy four-inch heels. Certain booth locations around the pit were a premium. Locals and filling brokerage groups would literally wrestle for position every day.
Infinium Capital Management is one of those firms that capitalizes on trading speed. Infinium is a proprietary trading firm and market maker for futures and options. They make markets in less liquid contracts and in the back months of liquid markets and for that are provided fee discounts and “allowed to disseminate mass quotes in products such as options,” says Infinium CEO Chuck Whitman. “ These mass quotes are only permissioned for market makers in these products. If someone who is not in the program attempts to flood the market with mass quotes, they will be denied by the exchange. This is designed to keep processing (quoting) transactions to a minimum to optimize exchange networks,” he adds.
Whitman notes that they can be kicked out of the program if they are deemed not to be fulfilling their obligation. He also points out that in some cases, as was the case in energy, the program could be ended as the market gains liquidity and there no longer is a need for a market maker. He adds, “In trading these products we assume the risk that we can’t get out of our positions.”
Clarity is needed
What isn’t quite clear and needs to be explained judging by the controversy is what is the advantage, if any, afforded those high frequency traders who are adding to the volume on these exchanges.
There are people, highly sophisticated experienced traders among them, that feel some of the these high frequency traders have an advantage over other market participants that goes beyond simply paying for a service available to all.
The greatest attribute of futures exchanges is transparency. CME Group and all futures exchanges need to make the various programs available to high frequency traders as transparent as possible.
Daniel P. Collins contributed to this article.
Fred Gehm is a fund of funds industry consultant and the author of “Trust is Not an Option: Evaluating and Selecting Investment Managers.” He also owns and edits www.fredgehm.com