| Sandy Rattray, Man Group |
As CEO of $13.4 billion Man AHL since 2013, Sandy Rattray has had to steer the firm’s managed-futures strategies, which use computer-generated algorithms to follow market trends, out of a virtually trendless period that began in 2009 and didn’t start to reverse until 2014. Most managed-futures funds, also known as commodity trading advisers, lost money during those years. The flagship Man AHL Diversified fund was no exception, losing 2.63 percent in 2013 and 1.61 percent in 2012. Its worst year, however, was 2009, when it lost 16.8 percent. But the fund bounced back and then some last year under Rattray, gaining 31.83 percent.
Rattray earned a double masters degree in physics and economics from the University of Cambridge and says he began finding applications for algorithmic, or quantitative, investment techniques while working on the trading floor at Goldman Sachs & Co., where he spent 14 years. While at Goldman, he was a co-inventor of the VIX Index, which measures the volatility of S&P 500 index options. He has also found that as a quant investor, he can apply many of the scientific research techniques he learned studying physics to detect patterns in the financial markets. He used that kind of scientific research to build another strategy, the Man AHL Evolution program, which deploys CTA-style trading in a wider range of markets, including interest rate swaps, OTC derivatives, cash equities, options and credit default swap indexes. The Evolution strategy earned 16.9 percent in 2013, when most CTAs were in a slump, and 20.31 percent in 2014. Man AHL is part of London-based Man Group, which manages a total of $72.3 billion in alternative assets. Rattray spoke with Alpha about lessons learned from the downturn and how he merges science with markets.
Q. Do you think, as most CTA managers do, that the five-year slump was a result of widespread policy interventions that interfered with market trends?
A. We think there were two sets of conditions that caused managed-futures funds to deliver unsatisfying returns during 2009 to 2013. No. 1, there were fewer trends and more reversals caused by interventions. And No. 2, the fact that markets became more correlated with each other. It feels, as far as I can tell, as if the very large interventions causing very large reversals seem to be behind us. We were pleased with how the market turned in favor of systematic strategies in 2014.
Q. While some CTA managers saw the downturn as a call to diversify into other strategies, you dived more deeply into trend following through the Evolution strategy. What made you decide to do that?
A. The slight complexity of trend following is that it’s never the case that all the markets are trending at the same time. The other aspect of this is that traditionally, managed-futures funds invest in trends through exchange-traded futures and FX forwards. We think many of the macro markets where futures don’t exist actually behaved fairly normally over the 2009 to 2013 period, with cyclical trends. Brazil had a series of fairly normal policy cycles. In Sweden the economy was growing so fast that in 2010 they put up rates to try to slow growth. Then last year they had two emergency rate cuts because they’d tightened policy too rapidly and had to respond the other way. While crude oil prices were behaving similarly to the S&P 500 from 2009 until late 2014, coal and electricity were moving in cycles of their own. Those movements were great for us, but we were getting this normal market behavior only in over-the-counter markets and not in futures markets.
Q. If it’s so easy to make money with those trades, why aren’t all managed-futures traders doing it?
A. People think the OTC markets aren’t liquid, but many of them are actually more liquid than listed markets. However there is one key challenge. If I buy an S&P 500 futures contract through Merrill Lynch, I can sell it to whomever I please. If I enter an OTC contract with Merrill Lynch but then sell it through Morgan Stanley, then I’ve got two apparently offsetting trades but the sale doesn’t cancel the previous trade. So you need to set up a process for all trades to face a central counterparty no matter whom you trade with. This is something we think we’ve cracked for the Evolution markets.
Q. As a proponent of quantitative trading in which computers make the decisions, what would you say gives it an edge over discretionary trading, which relies on human judgment?
A. If you take the key thing that underlies most discretionary investing, it’s value. You’re trying to find cheap stocks, bonds and such. The key signal that underpins much of our style of quant investing, on the other hand, is price momentum. We’re looking at what the price has done over a certain period of time. There’s been a sort of fashion recently in doing very long periods of back-testing momentum as far back as 100 years; I’ve even heard of a 700-year back test, which is something that cannot be done effectively for other measures.
If you look at academic literature, there have been thousands of papers about value and very few about momentum. I think one reason is that people get a bit sniffy about technical things — how can it be right to make an investment decision simply by looking at the price of that thing and what it’s done in the past? Surely you need to know what it’s worth. I can see the logic of that argument, but the reality is that momentum has been an incredibly powerful signal. You have to ask why this works.
One reason is that pricing is behavioral. All investors look at price charts, and not many rush in when the price is down. Another driver is the information transmission effect. A lot of trends develop slowly. Take shale gas. I’m sure I read about it as an alternative energy source years ago, but it’s just now giving the U.S. what appears to be a tremendous advantage over other energy economies. It didn’t take a week to drill those holes in the ground, it took years. People often think that the market reacts instantly to new information like earnings announcements. That’s true but there are other more gradual sources of macro change that cause sustained price momentum.
Q. Do you see more widespread use of quantitative investment strategies in the future?
A. There is hardly any area of life where algorithms aren’t used — everything from an airplane that can land itself to Google translation. Computers can play chess and drive cars. So I don’t think it should be surprising that investing can be done by computer algorithms. One of our research areas uses the same mathematical techniques that are used behind the self-driving car. I think it’s inevitable that quant techniques will become bigger over time in investing.
The quant industry is relatively young, and there are many more things we’ll learn over time. To give you a simple example, companies publish thousands of items in their financial statements every quarter. Most equity analysts use a small handful of those to publish their reports. There is so much data available there, it can be too much for people to read and digest, but it’s not too much for computers. As you deal with more and more difficult data and tease things out, you can do things that people can’t do.
Q. Do you think there will be a day when computers can produce enough algorithmic data to accurately predict prices?
A. No. There’s always a huge amount of randomness in markets. You hope to be right more than 50 percent of the time. That’s a concept built into what we do; we need to be right just somewhat more than 50 percent of the time to be successful.
