Whether it's applying the basic rules of calculus to map the dynamic topology of "financial space" or by simply cutting down human costs, the inexorable rise of high frequency trading (HFT) towards market-wide dominance is a revolutionary trend that is changing the way we think about investing and how to best deal with such a fast-growing phenomenon.
For a close follower of financial and technological trends, it's been interesting to watch the evolution of HFT from a small ripple propagating through the markets to a vast tsunami crashing upon exchanges around the world. Without a doubt, arising deep within the ocean by tectonic shifts in market structure, the accelerating pace of algorithmic trading has caught everyone by surprise.
Even though life quickly turned back to "normal" after the flash crash on May 6, 2010, HFT has emerged as a lightning rod for criticism as to whether the sudden record-breaking correction was a bad omen of future financial events or a large mass of intelligent algorithms slowly synchronizing with our daily human lives.
Rather than making a prediction as to which scenario above will play out, I believe we should at least understand the basic reasons why HFT is revolutionizing the market since it appears almost certain that it will eventually dominate not just the stock market (as it does now), but also foreign exchange, futures contracts, and the derivatives market. More than likely, without any major changes by regulatory bodies around the globe, HFT will quickly gain asset-wide dominance of the entire global financial marketplace.
Here are the six reasons why I believe this to be true:
1. Scientific or technological advancement and innovation
This one's a no-brainer. The only comment worth mentioning is that, as noted by others, some of the software being utilized for pattern recognition HFT goes back to Cold War weapons technology used for detecting stealth aircraft (see Kevin Slavin's TED talk presentation: How Algorithms Shape Our World). Also, a large number of the PhD scientists, computer programmers, and physicists that developed the supercomputers and software necessary for closely mapping the dynamic topology of "financial space" came straight out of CERN labs attempting to unlock the keys of the universe by smashing atoms together near the speed of light. The only difference now is that this technology is being used instead to unlock the keys of universal profit by bombarding stocks, futures, and various currencies in the Large Hadron Collider that has become our markets. Has it been successful so far? I don't know, perhaps companies paying over a quarter of a billion dollars laying cables across the Atlantic just to shave 6 milliseconds from their trades says something.
2. High correlation of asset classes
Over the past two decades there has been a steady increase in correlation among various asset classes, most especially in the stock market. According to Hedge Funds Review, "The correlation of value and growth stocks to the performance of the S&P 500 has spiked from less than 0.3 in the 1991-95 period to 0.99 and 0.98 respectively over the past five years. The correlation of non-US stocks to the S&P 500 is up from 0.34 in the early 1990s to 0.85, while emerging market stocks have a correlation of 0.79 compared with 0.38 going back 20 years."
Cited in the same article above, Stuart Rosenthal, CEO of Factor Advisors, says that "correlation is the new risk" and that they see "no signs that cross-asset correlations will decline." Instead, he believes that the longer-term trend is up. So how does this relate to HFT you ask? Having a low or negative correlation between various asset classes, stocks, etc. is the fundamental building block of traditional portfolio management and the main way to realize the benefits of diversification. Complex quantitative models fine-tuned over the years by financial analysts determine just the right allocation of stocks, bonds, currencies, and other assets based on their relative correlation to one another so that the volatility of an investor's portfolio can be minimized. With correlations increasing, the traditional model of investing loses its effectiveness. If we put buy-and-hold diversification strategies on one end and HFT on the other, the move towards greater correlation shifts a greater percentage of capital towards research, development, and execution of speculative momentum trading with HFT leading the charge.
3. Opaque financial information / accounting data
Most companies can't afford a team of financial analysts to wade through 50 pages of financial reporting data in order to detect earnings manipulation and accounting irregularities. Furthermore, even if they could, that still doesn't guarantee they'd be able to decipher the complex quantitative pricing of a company's derivative positions. Warren Buffet highlighted this issue in his often quoted 2002 annual report by saying, "derivatives severely curtail the ability of regulators to curb leverage and generally get their arms around the risk profiles of banks, insurers and other financial institutions. Similarly, even experienced investors and analysts encounter major problems in analyzing the financial condition of firms that are heavily involved with derivatives contracts."
As the "major problems in analyzing the financial condition of firms" go up-especially through the pervasive use of derivatives-the process of discovering a company's intrinsic value becomes more costly. Successful traders have long circumvented this problem in connecting fundamentals to market price by simply relying upon technicals. What's most interesting, however, is that this process requires little in the way of human intuition and more upon the strict adherence of a mechanical process. The best will even tell you that emotion simply clouds your judgment-something that machines don't have any problem with.
4. Volatility (HFT as applied calculus)
With market uncertainty posed by fear over a eurozone collapse, war with Iran, political debt-wranglings, foreign exchange fluctuations, excess liquidity, central bank intervention, fat-tail or "black swan" events, and numerous other problems facing us today, it is no wonder the markets are highly volatile. Then again, the markets have always had more or less volatility-one of the major ways we define risk. So how do you minimize this risk to investors? The answer you'll often hear is through diversification and long holding time periods. That is, the less correlated your assets are and the longer you hold them the more likely you are to overcome the temporary ups and downs of the market. The only problem with this, of course, is that sometimes people end up buying at a market top and don't get back to even until ten years later, if at all. Thus, as many of us know, volatility can have drastic effects both in the short and long term and can defeat the most perfectly constructed portfolio.
So where does HFT fit in? At some point, someone must've thought: "Why don't we apply the basic rules of calculus to the stock market? A stock (or asset's) price is really a complex curve exhibiting non-linear behavior so, rather than trying to predict its direction, let's simply break it into a series of extremely small intervals that capture its motion in real-time. The faster our execution speed the more we can profit from the market's every move." When you consider that calculus revolutionized the field of mathematics and helped bring about the Industrial Revolution, applying its basic (and yet completely unorthodox) principles to the market is no trivial matter.
5. Cost of human capital
As mentioned previously, it takes a team of financial analysts to sift through all the companies they invest in to detect earnings manipulation and accounting irregularities. If you're a large investment bank you can afford it. If not, you simply pay for their expensive research. Either way, you're still spending a lot of money. Of course, that still doesn't include the various portfolio managers, brokers, and other high-paid positions that typically accompany a successful investment firm. With high frequency trading firms, machines do most of the work. No more portfolio manager, analyst, or broker. Instead, you have people that watch blips on the screen all day to make sure the machines are humming a profitable tune. For the most part, it's a self-running operation. To a large investment bank, hedge,fun or trading firm the math is pretty simple: Cut half your workforce, find someone to develop a new HFT strategy in futures contracts or the foreign exchange market, and rent out a few rows of servers co-located to the NYSE. Excluding the math geek(s), your costs of human capital have gone down dramatically. All further funds can go directly towards upgrades or expansion into colocation centers around the world.
6. Regulatory changes
The last reason for the HFT revolution (or the "Rise of the HFT Machine" as Nanex and Zerohedge refer to it) is largely accredited to measures enforced by the SEC some years ago in order to enhance and modernize the existing market structure. As told in Deus Ex Machina, "Regulation NMS (National Market System) minimized the required time of trade execution from the normal 30 seconds down to 1, making the dependency upon high speed computer-driven orders almost a necessity." Whether the massive 523 page document was a necessary overhaul of the extremely outdated market structure or a test case for the unintended consequences that result from poor planning, I'm not sure ... perhaps both. Either way, according to Paul Rowady of the TABB Group, the regulatory changes made by the SEC that now favor high frequency trading will probably end up with explosive results. As he writes in Real-Time Market Data: Circus of the Absurd:
Like the lighting of a long fuse, the market structure we have today in U.S. equities results from Regulation NMS. Intended to foster the competing forces of low costs (through competition) and high transparency (through a consolidated quote and last sale data-feed mechanism), the byproduct of these rules is essentially the mad frenzy over tiny slivers of liquidity that we have today. When mixed with increasing levels of automation since the 1970's ... the fuse that was lit with Reg NMS so many years ago just might be leading us to a bomb.