“By using math, science, and computers to detect slight imperfections in the market, people made fortunes, but in many ways this type of arbitrage was fair. Those who studied the coin in great detail had a better chance of predicting the outcome of the coin flip. But has the pendulum swung too far?”
So is there a market structure problem in today’s high-speed electronic trading venues? More importantly, how does it affect our trading and what do we do about it?
In our view, the problem facing a bona-fide professional trader today isn’t the practice of high-frequency trading itself. Nor can the mad quest for speed be blamed on the underlying maker-taker model for encouraging liquidity in today’s electronic market. It is an indisputable fact that faster automated trading with greater liquidity helps build today’s modern financial markets. But speed in and of itself is not the real issue here. We will need to look deeper into the market structure to tease apart the problem.
We know that liquidity has different time horizons for different market participants; they are not all the same. We recognize that speed is only a technology enabler, not a root cause of the problem. When considering proposals to put in “speed bumps” for high frequency traders, one could not help but wonder if the cart is now in front of the horse. Who is to decide what is fast for whom and what is not? And how fast is fast anyway? Perhaps 350 microseconds delay for all may be a good start? Henry Ford nailed it on the head with his apocryphal quote:
Machine Learning 101: "You've got to learn to slow down. Nobody knows you're a Bot until you start running in front of humans." (Image Credit: Samory Kpotufe).
“No one said they wanted faster horses, they wanted less horseshit.”
Recall that this was the 1900’s, more than a hundred years ago when motoring down the street at 15 miles per hour was considered breakneck speed. But horse manure won hoofs down anyway. So the horse-and-buggy industry soon gave birth to America’s automotive industry. The interstate highway followed, changing the way we get around this country. "Trading is just plumbing," as Andy Kessler explains in his insightful Wall Street Journal op-ed piece, "the risk is not that the markets are unfair, but that markets don't function and things start to back up." And nobody wants that stuff to hit the ceiling fan.
It is understandable that everyone on Wall Street needs to get paid for their privileged participation, from the deal makers to the market makers. One does not begrudge the players their rightful share for as long as they continue to serve a useful role. However, we are staunch believers that technology will ultimately democratize access to the financial markets for new entrants with better ideas or better services. Well-intentioned regulations sometimes have unintended consequences. The SEC’s Reg NMS created loopholes in the U.S. stock markets for some to exploit, for others to work around, but always for entrepreneurs to fix. In our view, quant finance is an eminently “hackable” industry, alongside education, healthcare, and other industries that can be improved with technology; and Silicon Valley is our Ground Zero. Don’t think it’ll happen under the clear sunny skies here? Tell us why. We really like to hear your thoughts.
The problem, as we see it from a trader’s point of view, is with high-frequency tactics that are deceptive, invasive and manipulative. They interfere with fair and honest trading in the market by misrepresenting liquidity, distorting the true picture of supply and demand, and skimming the market. Such ill-intentioned practices are said to frustrate the market (i.e., a nuisance), as opposed to well-intentioned practices that facilitate the market. This distinction is important. We call market participants who employ HFT tactics for deceptive, invasive, and manipulative purposes the fakers, stalkers, and predators, respectively. They are the rotten apples in the bottom of the HFT barrel.
It’s All History Now: Reminiscences of Stock Market Operators on Wall Street. (Image Credit: Forbes).
Technically speaking, the stock market is what is called a "continuous double-auction" trading system that matches buy and sell orders using an "order book" model that is based on "price-time priority". But what really makes things complicated is that there are altogether 13 public exchanges in the U.S. that an order can route to, plus 45 "dark pools" currently in operation. We know that deep within today’s electronic market structure lurk the following types of HFT characters:
A high-speed chase game. (Image Credit: Dreamworks).
- Free Rider (aka Front Runner): Capturing information about what an investor wants from one place, then racing ahead to the next place and trading there at advantageous prices.
- Quote Stuffer: Entering numerous layered, market-moving orders to create a false appearance of buy- or sell-side pressure so as to trade at advantageous prices.
- Feed Blaster: Flooding the market with phantom orders at a high enough rate to slow down a direct exchange feed at any chosen time.
- Stop Hunter: Sending a flurry of market orders ahead of news release to move market one way, triggering stop-loss orders that swing the market in the opposite direction.
- Fire Starter: Sending a flurry of market orders to ignite momentum that moves price in the right direction and in the right way to entice others to jump on the bandwagon.
- Invisible Jumper: Jumping ahead in the limit order queue via a special order type called “Hide Not Slide” that makes them invisible to others.
- Zero-Plus Scalper: Queuing on both sides of bid and ask lines to pick up rebates with little risks, and fleeing to save own skin upon the first sign of trouble.
Like other professional traders, these denizens of the modern electronic markets all seek to learn about latent supply and demand in the market, at minimal costs and with minimal risks, as a precursor to trading. This is what all traders do anyway and is perfectly alright. However, fakers, stalkers, and predators employ HFT tactics that are highly controversial or questionable by the standards of professional conduct that govern human traders. This is where things get dicey. For example, even though HFTs supply the bulk of the orders to the exchanges, many are immediately canceled, and only a select few actually execute. This is a characteristic behavioral profile common among HFTs; it is as though they are trying to play a never-ending game of “catch me if you can” with the many hapless human traders and investors.
Absent any clear written “rules of the road”, how is anyone to judge one way or the other? The rule of law requires that the law be first spelled out, before they can be interpreted, argued, debated, discussed, and finally understood, respected, and obeyed. That’s how we do things in the human world, e.g., when it comes to such matters as: who gets to drive in the carpool lane, or who goes first at an intersection with stop signs. But in trying to comprehend the realm of HFT Bots, absent any guidelines, we inadvertently fall into the trap of trying to see things through an anthropomorphic lens of human stereotypes (e.g., as we have amply illustrated thus far). That’s when emotion overwhelmed reason, with predictable outcome of confusion followed by occasional bursts of “road rage”. The issue here is not speed; it is fairness in the context of unambiguous “right of way” for all market participants.
From the point of view of geography, however, speed is its raison d'être. Let’s take a good look at the map of north New Jersey to see how geography determines high frequency trading speed. Mahwah, where NYSE is located, is just 40 miles north of Carteret, where NASDAQ is located. Weehawken is right across the Hudson, smack in the middle of these two. A trading signal that originates from Lower Manhattan travels up the West Side Highway and out the Lincoln Tunnel. Immediately outside the tunnel, in Weehawken, sits the BATS exchange. So BATS is always the first stock market to receive orders coming from Lower Manhattan. As Eric Hunsader of Nanex explains:
“When you want to buy a lot of stock, you’ve got to go to multiple exchanges to get it. That’s one of the things Reg NMS did. It took all of the liquidity in one spot and split it up. The order, if it comes into BATS first, a high-frequency trader there will see that trade execution. Immediately, on the blue line, up to Mahwah, they will signal up to their other machine to say, “Buy everything that’s available.” And they can do that faster than the order gets to the NYSE. It’s the fastest network there is. They’re able to remove that order on the other side. The only reason they made that purchase in the first place is because they saw this other order coming in.”
A tale of three cities. (Courtesy of Eric Scott Hunsader and Nanex).
For professional traders and investors, the solution to this problem of “free riding” (aka “front running”, but this label is highly controversial) is a program called Thor, created by Brad Katsuyama (of “Flash Boys” fame). Thor was designed to time the trades by slowing down the faster orders to match the speed of the slowest order in the group so everyone would hit, say, BATS, the NYSE, and NASDAQ all at the same time. A high-frequency trader at each one of those facilities couldn’t see it fast enough to react and get to the other exchange. “But Thor only works when the networks are clear and uncongested,” according to Eric Hunsader, “It doesn’t always work, but it works often enough.”
From the point of view of ordinary investors, it is fair to say that HFTs offer greater liquidity and reduced spreads, thus improving trade execution. In any case, retail investors need someone to take the opposite side of their trades and HFTs are happy to do just that. Problem is: retail investors are considered uninformed traders. That means they trade for reasons other than correcting for asset mis-pricings. Statistically speaking, retail investors are very good at making very bad decisions all on their own. That’s why HFTs love to give retail investors what they want. In fact, HFTs are willing to pay good money for the privilege of doing so; and retail brokerage houses are only too happy to oblige. This practice, called “payment for order flow”, is officially sanctioned by the SEC because it allows smaller venues, such as “dark pools”, to compete more effectively with the NYSE. In short, high frequency traders make their profits the old-fashioned way, just by being on the right side of the trade. For HFTs to internally cross trades that are routed their way by the retail brokerage houses, i.e., internalization, speed is not at all an important consideration. Unless, of course, when they decide to pass along orders that they don’t want onto the exchange for trading. Considering that it is statistics, not speed, that really matters here, these order flow internalizers (numbering 200+ at last count) are beginning to look like Type 2 Bots (à la Shannon’s Outguessing Machine) that we have seen earlier.
We are all in this together: You make (market), I take (risk), and she gets paid (commissions). Hmmm... (Image Credit: Wall Street Journal).
We can learn a lot from studying the structure of the equity market. After all, technology and market innovations from equity trading eventually spill over to the foreign exchange market at some later time. A typical HFT derives its unique advantage from a combination of four factors: colocation, data feeds, order types, and rebates (collectively the “four horsemen of high frequency trading”). For new entrants, however, the U.S. equity market represents high infrastructure costs, complex market access rules, and uncertain regulatory risks. In short, the U.S. stock market appears to be a less-than-friendly first market for us to start a trading operation. It’s not for beginners.
In stark contrast, the foreign exchange market is global in nature and subject to little regulation. Venues are free to differentiate themselves; no two venues have precisely the same structure. What’s more, market access is easy and affordable. Our intuition tells us that such a rich market ecology offers the best opportunity for finding a defensible trading niche, and gives us the best chance of survival in the long term. Let’s hope we are right about this.
So what lessons have we learned? First, stay away from markets populated by fakers, stalkers, and predators (or learn quickly to avoid them). Second, slow down (e.g., like Thor) to get there sooner. Rest assured that all our Type 2 Bots take these lessons to heart before they graduate from our lab and enter the real markets.
Slow down ...
... and get there sooner.
References:
- Patterson, Scott (2013). Dark Pools: The Rise of the Machine Traders and the Rigging of the U.S. Stock Market (First ed.). Crown Business.
Lewis, Michael (2014). Flash Boys (First ed.). W. W. Norton & Company.
