High-frequency trading (HFT) is one of the hottest developments - and most controversial topics - in investing today. It's arrived so fast that many investors have been left scratching their heads, wondering, "What is HFT anyway? Where did it come from, and should I be worried about it?"
The answers have to do, unsurprisingly, with the federal government... but also, oddly enough, with the speed of light.
That's because, although light is the fastest thing in the universe, its velocity is finite. The sun that we see is the sun as it existed about 8.3 minutes ago. That's how long it takes light to cover the 93 million miles to Earth, and it's a measurable amount of time. On the surface of our planet, however, distances are so much shorter that we tend to think of the time required for light to go from any given Point A to Point B as negligible - and for most purposes, that's true enough. It usually doesn't matter that it takes a teensy bit longer for a light beam to travel from New York to San Francisco than it does to Chicago.
With HFT, it does matter.
In times past, analyzing the market's daily actions was sufficient for decision-making. Over the years with electronic trading tools, that period shifted down to hours, and even minutes. Analysts watched stocks continuously on their Bloomberg terminals, and at the first sign of impactful news, they bought or sold.
But these days, everyone has access to the same basic tools. Even the dedicated, stay-at-home day trader can do simple "algorithmic" trading - using a computer to automatically execute trades when preprogrammed conditions are met - if he so desires. Because each invention of the algorithmic trader was so easily copied and cloned, some serious market players have gone in search of sustainable competitive advantages - ways to give themselves an edge that is not so easily eroded.
They started building sector-specific software. They targeted their models at new geographies, watching the entire world's markets together instead of in isolation. They sought to divine actionable patterns from the massive piles of data they were collecting. They applied social science to their models. Arbitrage strategies. Crowd theory. Game theory. All for a better, faster tool to pick winning investments.
As each firm found an advantage in the markets, their competitors aimed to one-up them. Some built smarter systems, hiring away engineers from Microsoft and Google. They focused on faster code, pushing the envelope of parallel processing and simulation technology. They invested in artificial intelligence and flexible pattern recognition. Any kind of cutting-edge science, really.
Others simply put their systems closer to the exchange, to gain a few milliseconds over the competition.
Thus was born HFT. It couldn't even have existed before the advent of superfast computers and the ability of programmers to write some very complex algorithms. Gifted with all this electronic market analysis, intrepid data sleuths began to notice patterns emerging. They saw opportunities to arbitrage inefficiencies in the markets and began to trade those alone.
High-frequency traders have no interest in any company whose stock they're trading. They don't care about its earnings, what sector it's in, who's on the board of directors, how it fares in technical analysis, nor what its long-term prospects are. They may not even know its name. At the end of every day, after trading tens of millions of shares, they really don't want a single share of stock on their books at all.
What attracts them is making a tiny profit on an opportunity that comes and goes in the blink of an eye. And to repeat that over and over, until the tiny profits fill a big bag with dollars.
But it wasn't just the technological arms race that got the ball rolling toward HFT. It also required a little nudge from government regulators who were blissfully unaware of the law of unintended consequences.
Washington, DC's involvement was the result of new trading options that appeared in the late '90s, like electronic communications networks (ECNs) and Alternative Trading Systems (ATSs). In brief, these are trading systems that are not regulated as exchanges, but exist as venues for matching the buy and sell orders solely of their own subscribers.
As Benn Steil, economist and senior fellow at the Council on Foreign Relations, argues: "the historic regulatory model is based on the notion that there are logical distinctions between the roles of exchange, broker and investor. Technological developments have broken that down entirely."
Government took note, of course. Securities regulators have always considered as one of their primary functions the responsibility to see that markets are "fair." When Joe Investor buys or sells a stock, he should be assured of at least an equal opportunity of getting the best possible price, every time. The government fears that if this confidence in the integrity of the system were to be compromised, the whole financial house of cards might come tumbling down (which is ironic, considering that it is creating conditions for such a crash by over-politicizing the economy).
Yet the proliferation of new systems and exchanges meant that there were bound to be price variances among them at any given trading moment. And this created "unfair" arbitrage opportunities for those smart enough and quick enough to take advantage of them.
The SEC's "solution" came in 2007, in the form of Regulation NMS (National Market System). NMS allowed any stock on any exchange to be traded on any other exchange, with the order automatically filled at the best bid or best offer. But for that kind of price discovery to happen, each order must be routed to all potential exchange locations at the same time, before any trade can be executed.
Which is where the speed of light comes in.
If every order is routed to every exchange at the exact same time, then theoretically no trader can gain a leg up on others by being first in line. That's what the SEC intended. Unfortunately, it's not an achievable goal in this universe, where light speed dictates the velocity at which data can travel over a fiber optic network.
It takes about 100 milliseconds for light to travel from New York to San Francisco, but much less than that to make it to a nearby neighborhood. Not a difference that humans can detect... but computers can. And they can fill the gap between, say, 10 and 100 milliseconds - called the "latency period" - with a complex series of instructions.
If you want to exploit the profit potential inherent in this latency, what you must do is site your operations closer to the source of the action than the other guy. You'll choose to be snugged up to the trading floor in New York, the center of most financial transactions, rather than to set up shop in Miami, Dallas, or San Francisco.
That's called "proximity trading," and it's accomplished through the phenomenon known as "colocation," another of today's big buzzwords. New York is the obvious choice for this, and sure enough, there are two huge colocation data centers in the city.
Manhattan, however, has some serious drawbacks - most obviously, the price of buying or leasing real estate. Then there is the voracious power consumption of these massive server farms, also prohibitively expensive downtown. Finally, there is the frightening fear of data loss, companies' need for data protection if something goes badly wrong on Wall Street.
Hence, in the US you'll find most of these facilities in New Jersey. It's a millisecond farther away, but many companies are willing to make the tradeoff, and they've defined a swath of land that in the trade is known as "the donut" - a semicircle around New York City with a width of 30-70 kilometers. Within the donut lie the ideal colocations based on the parameters of land availability, power availability (including backup), construction cost, and likelihood of effective disaster recovery (which goes so far as to take into account the projected blast radius of a small nuclear device).
Enormous outlays of capital are required in order to build out colocation centers around the world's financial centers, and no one would bother if they didn't have clients waiting. Clients they have. High-frequency trading is now estimated to drive at least 50% of the stock market, with some pushing its share as high as 70%. It also takes about a quarter of futures markets. And it is highly lucrative. According to analysts at the Tabb Group, HF traders earned around $13 billion in profit in 2010; the number was probably much higher last year.
HFT has received a full measure of negative press. But there's nothing in and of the practice itself that is bad. At its core, it's no different from business as usual. Market makers have always stood ready to execute both buy and sell orders, profiting off the spread. This is what provides liquidity to the trading floor and keeps the system running smoothly. If anyone from an individual to a mammoth fund wants to trade a stock, they can. The only difference with HFT is that machines can do it faster and more efficiently, making adjustments in fractions of a second if need be, and ensuring that investors do realize the best price on their trades.
At the same time, though, HFT technology permits the extremely rapid placing and withdrawal of orders, up to thousands of times per second. And, it is this speed that has led directly to one of the most controversial of HFT's practices, a ploy called "fluttering."
Using this technique - where thousands of orders are placed and then rapidly canceled before they are acted upon - high frequency traders' computers can nibble at the market, until they find a pattern or an anomaly that exists for only a moment (something that simply may be due to them having a lower latency period than a competitor) and can then exploit it.
To envision how HFT plays out in the real world, here's an excellent illustration from an article by Bryant Urstadt, writing in the MIT Technology Review:
"Imagine that a mutual fund enters a buy order, telling its computer to start by offering the current market price of $20.00 a share but to take any asked price up to $20.03. A high-speed trader ... can use a 'predatory algo' to identify that limit by 'pinging' the market with sell orders that are issued in fractions of a second and canceled just as fast. It might start at $20.05 and work its way down to $20.03, canceling and reordering until the mutual fund bites. The trader then buys closer to the current $20.00 price from another, slower investor, and resells to the fund at $20.03. Because the high-frequency trader has a speed advantage, he is able to do all this before the slower party can catch up and offer shares for $20.01. This speedy player has found the buyer's limit, gathered up and sold an order, and snipped a few pennies off for himself."
Since these anomalies result in differences of only pennies, and since we as retail investors plan to hold our stocks for far longer than a minute, why should we care what HFT traders do?
We shouldn't, defenders of the practice say. In fact, they maintain that by pulling bids and asks closer together, they're providing us with a free service that helps us benefit from proper price discovery. Moreover, they claim that they're well positioned to right a ship that's tipping precipitously, and to steer it back toward fair value.
They say that's precisely what happened during the infamous "Flash Crash" of May 2011, when the Dow plunged nearly 1,000 points in less than 20 minutes. Investigations after the fact have shown that the meltdown was initially triggered by one (human) trader who accidentally tried to sell a massive amount of S&P 500 futures contracts, setting off a string of toppling dominos as other traders' stop limits were breached. HFT didn't cause the crash, say proponents; in fact, it saved the day, bringing the market back before humans could react, by right pricing the assets. Innumerable small trades quickly stairstepped the market back up to nominal value.
Others are rather less sanguine - including the SEC, which called HFT a "contributing factor" in the Flash Crash.
Furthermore, critics point out that HF traders can gouge investors who place market orders. We've probably all experienced buy or sell orders that were filled at some price that seemed out of whack with whatever the stock was doing on that particular day. HFT could be to blame, some say. And the heist takes place in a legal area that is very grey indeed.
Steve Hammer, founder of HFT Alert, explains:
"[Fluttering] is not enough time to get an execution. It's illegal to put in a phony bid or a phony offer but that's what's happening. HFTs create in essence financial spam, which increases the latency in the system and allows them to push prices in one direction or the other. People seeing lots of volatility in a stock who put in market orders are giving the systems license to steal. If they can cross the market and lock the spread for a fraction of a second, they can take out any limit order above or below that price, resulting in a very brief, wide swing in the price of that stock, 5-6% in a single second, even though if we're looking we see no change whatever in price or spread, yet here come all these trades through that are outside the spread at that point in time."
Another claim is that HFT is destroying the futures markets, i.e., those with a legitimate need for hedging are seeing their positions blown up by high-frequency manipulators who cause such volatility that the hedgers are forced into unnecessary margin calls.
Wherever the truth about HFT may lie, the tempest it has caused was bound to generate some new regulations, and it has. A recent SEC proposal would eliminate one controversial tactic of high-frequency traders: the "flash trade," in which exchanges alert designated traders to incoming orders. Critics call it a variation of front-running, an old (and illegal) practice that involves traders buying and selling in advance of large orders.
Nasdaq, meanwhile, announced a new policy in March, under which it will charge its members at least $0.001 per order if their non-marketable order-to-trade ratio exceeds 100:1. (Non-marketable orders are those posted outside the national best bid and offer.) The fee will be limited to those individual market participants that send in at least one million orders per day, although market makers will also be exempted, even though some market-making firms are considered HFT shops.
In the end, it would appear that we'd better get used to HFT. It is likely here to stay, regardless of new regs or what we may think of it. Technological advancement is a genie that, once out, can never be forced back in the bottle. In the investment markets, traders will always try to use new technology to gain an edge, counter-traders will always seek ways to negate it, and government will always invent "fixes" that are a step behind the times.
For those of us who invest in a company for weeks, months, or years, HFT will make little difference. But to be on the safe side, always place limit orders, never market orders.