International Forecaster Weekly

The Robots That Rule the Markets

Aside from the unfairness of HFTs, there's also the inherent risk that comes when algorithms that are programmed to perform thousands of transactions a second are let loose on the markets... it's almost impossible for a single trader to massively swing the market one way or another in a single stroke, but a glitchy algorithm employed by an HFT can decimate a market in seconds.

James Corbett | September 14, 2013

In the markets, seconds are eternities. If you missed Apple's latest iFlop by a matter of minutes you're already behind the trade. Only the chumps and amateurs wait for a trend to develop before buying or selling; the real traders have already worked out where the market is heading and made their moves within fractions of a second that are imperceivable to the average human. That's because the real traders are robots.

            The computer-ization of the markets in recent years will not have escaped the attention of any but the most casual traders...and even those casual traders are increasingly managing their portfolios themselves online. Just this past month, the automation of the markets hit home when computer problems caused the Nasdaq to halt trading for three hours. Such disruptions are becoming more common, with notable examples in recent years including the May 2012 Nasdaq meltdown over Facebook's first day of trading and the May 2010 “flash crash” that saw the Dow Jones plunge 600 points in 5 minutes. More than just an occasional glitch or nuisance, however, this computerization is transforming the way the markets operate...and tipping the playing field even further in the favor of the institutional giants who can afford the best programmers and the best algorithms. Forget The Smartest Guys in the Room; the Enrons of tomorrow will be built by The Smartest Robot in the Room.

            A recent news story highlights the concern. An employee of Flow Traders, a “high-frequency trading” firm in Manhattan, was charged with stealing the company's proprietary trading algorithms in an apparent attempt to set up his own rival firm. In the new normal of computerized trading, the most valuable inside information is not the lowdown on the next big merger or the inside scoop about Apple's next product announcement, it's the bits and bytes in the computer code that enables the big trading firms to react to any announcement a fraction of a second before its competitors.

            So what is “high-frequency trading” anyway? At one level, it involves the practice of using computer algorithms to automate securities trading. But this isn't your typical buy-and-hold trading that examines fundamentals and does due diligence and tries to invest money in a winning company, or an algorithm that's just programmed to buy or sell at certain levels. High-frequency traders buy and sell stocks thousands of times a second, hoping to gain even a fraction of a penny of profit on any given trade. The secret is in the volume. This process, completed and repeated thousands of times faster than any human could hope to process such a transaction, can be lucrative for whichever firm has the fastest code, or even whichever firm can situate their computers closest to the exchanges' own servers.

            So how fast are the trades involved? The SEC authorized electronic exchanges in 1998. At that time, an average HFT trade took several seconds. Now that time has been whittled down to a matter of microseconds. With each decrease in processing time the HFTs gain a greater advantage over human traders and over whatever competitors are using a slower system.

            So what's the big problem? So some smart computer programmers managed to parlay a technique for arbitraging a fraction of a penny out of a trade into a trading strategy in itself. No big deal, right? Wrong. In fact, the world of HFT creates an unfair advantage for the big boys who can afford the best algorithms and the best trading systems.

            How so? Well, for one thing there is the fact that the high-frequency traders are effectively front-running the markets. To understand this we have to go back to Regulation NMS (RegNMS), a financial regulation promulgated by the SEC that was the end result of a long process of consolidating the fragmented system of regional exchanges for traders so that any stock could be bought from any exchange. Under RegNMS, bids and offers anywhere in the country can be routed to whatever order is coming in at that moment. In theory, this is a wonderful thing for capital markets, since it allows buyers and sellers to always find the best possible bid or price for their transaction. The problem is the speed of light. It takes approximately 100 milliseconds for a quote to leave an exchange and physically travel across the country. To put things in perspective, the blinking of an eye lasts about 300 milliseconds. So under any normal circumstances, this latency period would not effect anything. But the HFTs are smart. They co-locate their servers with the exchanges (at a cost of $14,000 per month in the case of the Nasdaq's co-location facility in New Jersey), meaning the data is pumped into their computers before it gets sent anywhere else in the world. And because they get that fraction of a second head start, they effectively get an advantage over the rest of the market. They can, in effect, see the market's future and trade accordingly before anyone else even has a chance to see the data.

            How significant is this? Well, RegNMS was finalized in February 2007, and from 2007 we see an exponential increase in trading volume across the Dow and S&P. This is not because more investors suddenly decided to start transacting more frequently from this point on, but because the HFTs suddenly had the perfect opportunity to make oodles of money front-running the markets, and they cashed in on it. Estimates are that high-frequency trading now accounts for over 50% of market volume. The HFTs have taken over.

            Of course it's not just this front-running ability that makes HFT so lucrative. There are all sorts of new manipulations that this robotic trading has opened up. Last year, HFT whistleblower Haim Bodek made waves by talking about the ways that high-frequency traders could exploit obscure points in the RegNMS legislation to game the system. One such exploit, “queue-jumping,” allows HFTs to get ahead of other traders on the order book during price moves, giving them an unfair advantage. Other predatory practices involve collusion between HFTs and certain ETF firms, including ETFs courting HFT business by telling them in advance when they are going to buy and sell stocks that make up funds to adjust for market moves (“rebalancing”), allowing the HFTs to front-run them.

            Aside from the unfairness of HFTs, there's also the inherent risk that comes when algorithms that are programmed to perform thousands of transactions a second are let loose on the markets. Other than a “fat finger” transaction that makes its way through, it's almost impossible for a single trader to massively swing the market one way or another in a single stroke, but a glitchy algorithm employed by an HFT can decimate a market in seconds, as in the flash crash of 2010 or the “Knightmare” of 2012, where a malfunctioning Knight Capital HFT began buying and selling $2.6 million of NYSE-listed stocks a second for 45 minutes, costing the company 40% of its value ($440 million) in less time than an average worker's lunch break.

            Having said all of this, there are two reasons for hope when it comes to the David and Goliath battle of HFTs and regular investors. The first is that HFTs are in retreat. Although a few years ago they accounted for upwards of 70% of all transactions on the market, that number is now down closer to 50%. The HFTs have become so efficient at wringing “sub-pennies” out of transactions that even they are finding it harder to make money out of the scam at this point.

            Secondly, it seems that the regulators are finally stepping in to put an end to the wild west of HFT trading. Just this past week the CFTC released a document proposing a series of regulations that could be used to curb some of the excesses of the HFTs. Although a mixture of hair-pullingly obvious statements (“In a trading environment where a single algorithm can submit hundreds of orders per second, risk management systems operating at slower speeds could allow an algorithm that is operating in unexpected ways to disrupt one or more markets”) and toe-curlingly tepid accusations (HFT front-running could “have an adverse impact on market fairness”), it is at least an admission that the system is broke and needs fixing. Earlier this month, Italy became the first country to crack down on HFTs, levying a 0.002% tax on transactions lasting less than half a second, a meaningless charge for average investors but a devestating one for HFTs that depend on the ability to transact thousands of times a second.

            Of course, more government regulation and taxes on transactions are not the real answer to the problem of HFTs. The real answer, as always, would be in allowing greater freedom in the market. Allow individual exchanges to regulate HFTs however they want, and allow investors to decide if they'd rather invest via HFT-free Exchange A or HFT-laden Exchange B. Sadly, however, free market solutions are treated like ideas from another galaxy in the current environment, so the average investor will have to cross their fingers and wait for the SEC or CFTC to save them...or choose to avoid the phoney-baloney rigged markets altogether.