Not too many people remember the second week of August 2007, but for others, that week will be forever etched in their minds. And in many ways, it should be etched in yours as well.
During that infamous week, we witnessed an event that would later come to be known as the 2007 Quant Meltdown. While the events that unfolded took place in a relatively small pocket of the financial market, their effects would have consequences for all investors.
Algorithmic or quantitative investing has been on the rise for many years. One of the most prominent (and horrific) examples was Long-Term Capital Management – which, by the way, is one of the most ironic names possible, considering the fund engaged in high-risk, leveraged arbitrage trading strategies.
LTCM used a quantitative strategy to take advantage of arbitrage opportunities in fixed income markets. In 1998, when a financial crisis in Russia triggered a flight to safety, the firm sustained massive losses and was in danger of defaulting on its debts.
You may also like Populism and the Long-Term Debt Cycle
The amount of leverage LTCM used magnified its problems. Using $100 of borrowed money for every dollar of its own, LTCM held massive positions representing roughly 5% of the entire global fixed-income market. When a “once-in-a-lifetime” event occurred that was not factored into their computer models, it brought the financial world to its knees.
With the help of a bailout by their creditors and the Federal Reserve, a systematic meltdown of the market was prevented. But this event was a just a prelude to what was on the horizon.
Fast forward nine years, and again we saw computer-driven trading leave a nasty scar on the market. This time, it wasn’t just one quant fund, it was many.
Whenever a strategy becomes lucrative on Wall Street, others quickly swoop in to try and capture their fair share. This happened in the run-up to the financial crisis as many quantitative hedge funds began using what’s known as a long/short equity-neutral strategy.
This type of approach involves using computers algorithms to identify and trade both on the long and short side of the market. Undervalued companies with improving fundamentals and momentum are purchased, while overvalued companies with deteriorating fundamentals and momentum are shorted. Employing equal amounts of capital in both directions theoretically leaves these funds “neutral” to market movement – that is, they should be able to make money regardless of whether the overall market itself goes up or down.
These equity neutral strategies have been used for years, and still are today, but once again, they did not account for the impact of low probability or “high sigma” events.
So what happened during that fateful second week of August? This lucrative trading strategy simply stopped working. This time, the dislocation came from a different area of the market where these funds weren’t even investing … subprime mortgages.
As losses mounted for other investors who held positions related to subprime loans (mostly derivatives), these investors had to sell other assets to meet their margin calls. The natural place to take funds from was easily traded equity securities such as Microsoft and GE. The selling here is what triggered the quant meltdown.
Over the course of a couple of weeks, culminating in the second week of August, quantitative funds recorded losses of roughly $500 million. This may not sound like much to big institutional investors, but the carnage was everywhere.
The event nearly took down Goldman’s QIS (Quantitative Investment Strategies) division and caused painful losses at the most prominent quant funds, including Renaissance Technologies, AQR Capital Management, and the famed Medallion fund.
This event would quickly become overshadowed by the financial crisis, but its lesson is one that we need to pay attention to. So what exactly is the lesson? Well, there are a few.
The first is that we need to recognize the changing character and landscape of the financial markets. If you take a look at the chart below, courtesy of the WSJ, you can see the alarming rise in the share of stock trading performed by quant funds.
Notice that in recent years trading by regular hedge funds, traditional asset managers and banks has been on the decline. In its place has been the rise of quant funds, using algorithmic processes to make buy and sell decisions.
Quant funds now account for over a quarter (27.1%) of U.S. stock trades. This is a larger proportion of shares traded than by any other source, except for individual investors, who account for roughly 29% of trading volume.
The rise in the number of trades executed by computer models is staggering and is giving way to a new type of behavior on Wall Street. This is the second lesson: the behavior of the market is changing.
In theory, computers should be able to make much better investment decisions than we humans can. After all, we’re subject to a host of behavioral biases and can’t crunch nearly as much data as even the most unsophisticated computer. But the problem here lies in the fact that just as certain trades become “crowded,” so do quant strategies.
The quant meltdown in 2007 was compounded by the fact that this pocket of the market had become cramped. There were so many players executing the same strategy that when a hiccup finally came, the consequences were magnified. Cliff Asness, of AQR Capital Management, recognized this saying, “We have a new risk factor in our world.”
Now, as the concentration of quant-driven investing continues to grow, we could be in for even greater swings caused by too many computer algorithms executing the same trades, which negatively influence one another. In other words, contagion.
Oddly enough, last Wednesday’s market action reminded me of this type of behavior. It appeared to be riddled with indiscriminate selling at the index level.
Another trend that has arisen in recent years is using ETFs to buy and sell large swaths of the market, rather than individual stocks. Many of the less sophisticated algorithms out there use various inputs (whether it’s news headlines, twitter keyword analysis, or satellite images) to simply buy or sell big index ETFs such as SPY – which tracks the S&P 500.
When index ETFs are bought and sold in large quantities, it affects each of the underlying stocks in the same way. As a result, stocks that really don’t deserve to be sold are, and vice versa. I’ll give you an example.
On Wednesday, while watching the Dow sink 370 points, I noticed that core technology companies such as Amazon and Google were down just as much as the broader market, even though their outlook and prospects are frequently unrelated to the geopolitical concerns that often hold our market captive.
To my mind, this was indicative of index level selling that was simply trouncing all sectors. I was tempted to purchase a few call options on Amazon and Google, thinking they’d rebound in the following days even if the broader market didn’t, but alas, my emotions got the best of me and I never clicked “confirm.”
The big takeaway that I’d like to get across here is that we’re all going to need to prepare ourselves for dealing with more and more market “noise.” I believe that short-term price action may become more erratic as we adjust to a market that is subject to more frequent and larger magnitude dislocations, resulting in large part from quantitative investment strategies.
As machines and computers continue to take over an increasing portion of our lives, the same will happen with our financial markets. And right now, we’re beginning to see some of the consequences of this play out in small pockets here and there. What happens when machines eventually control half of all stock trading? Or three-fourths?
I have no idea. But we need to recognize that this trend is solidly in place and that the behavior of financial markets will continue to evolve at a rapid pace. In this new world, we must be willing to challenge our previous conceptions about market action and adapt quickly to whatever it throws our way.
One last thought here has to do with the primary trend. The vast majority of quantitative-driven trades are short-term oriented, lasting anywhere from a few minutes to a few months. As a result, they tend to influence secondary and tertiary trends more so than the primary trend of the stock market. This means that identifying and investing alongside the primary trend may become even more important (and also more difficult) in the future.
The preceding content was an excerpt from Dow Theory Letters. To receive their daily updates and research, click here to subscribe. Matt is also the Chief Investment Strategist at Model Investing. For more information about algorithmic based portfolio management, click here.