Dr. Ewan Kirk is the CIO and Founder of Cantab Capital, a systematic hedge fund based in Cambridge. According to its website, Cantab “employs a multi-strategy, multi-asset approach within the systematic space.”
Before founding Cantab in 2006, Dr. Kirk was in charge of Goldman Sachs’ Strategies Group, overseeing all of the bank's quant technology.
In episode 292 of Trend Following Radio, from November 2014, host Michael Covel talks to Dr Kirk about his background, how someone with a PhD in mathematical physics ends up at Goldman, on trend following in Europe vs. America, and on computer programming being today's literacy...
Read below for some extracts from the interview, or listen to the full episode here...
Systematic trading is a little bit like being pregnant. You either are or you aren’t. You can’t just be a little bit systematic.Ewan Kirk
On his Goldman beginnings...
They were looking for somebody who had some sort of mathematical background and computer programming was also important. I say this at almost every interview; it’s sort of useless being a mathematician without being able to computer programme. Because it’s like being a novelist who can’t write, you express your views or your ideas through computer programming. So I was a reasonably good computer programmer, I had a background in quant and Goldman very kindly took a punt on me. It’s today's literacy (programming), that strand runs through my entire career but also runs through what we do at Cantab. I myself just spent the last two or three weeks programming up a piece of infrastructure. There’s no rest for the wicked as they say.
On London becoming the trend following hot spot...
If you think about the very early trend following, the history of the industry, the Turtles in the mid-80’s, the kick off for the London - or better call it the European scene - probably happened in the mid to late 90’s, with a platter of great firms across Europe; HL, Winton, Bluetrend, Aspect and ourselves. I think partly the reason why we maybe became more successful was more to the scientific approach to investment and statistics. I look back at these and look at the old tattered books about average true range and breakouts and it seems like the dark ages.
We are thinking at a much more statistical sense, more scientific. I clearly believe that a more scientific approach to investing, a more rigorous statistical approach to it, I believe that’s better than just a rule that happens to have worked in the past. Even though we started in 2006/2007, I remember people asking me, particularly in the States, why do you weight your positions by risk? Why don’t you take a constant lot quantity? That’s just madness, in a world where we have a contract, a wheat contract, which is maybe $20,000 and the nickel contract is $250,000, it’s just an insane way of doing it, but it was the tradition.
And our industry is quite conservative, there’s a small subset of investors who want the new thing, believe maybe that there’s some fabulous technique that’s just going to predict what the S&P is going to do tomorrow. But there are also people who want it to be the way it was. And anything that steps outside there is maybe a little bit different.
On losses being part of the game...
People are desperate to invest in something that never loses money. And that is of course why Bernard Madoff existed. I try to invest in things that don’t lose money; I try to come up with a strategy that never loses money. But of course we all want that. But the reality of almost all investing is that if you’re really good or really lucky and you’ve got a very long track record, maybe 20-30 years and you’ve never changed your strategy over that period - which of course none of these things are true for anybody - maybe the best you can hope for is a Sharpe Ratio of 0.8/0.9, maybe 1. Broadly a good investment strategy is something over a long period of time has a Sharpe Ratio of 1. Investors should really want that.
So a 20% volatility with an average gross 20% annual return, that would be great. But that does mean that every two years, it’s going to have a drawdown of 15%, statistically, every four years it's going to have a drawdown of 20%. This is just what happens. Even if the system truly has that return profile, it’s going to experience those kind of drawdowns and it’s going to experience losses. I have a little spreadsheet that I sometimes show to clients when discussing this, which simulates five years of daily returns from something which is a 20% return, 20% volatility process, effectively a Monte Carlo simulation.
And every time you press F9 on this spreadsheet it draws another graph on another realization of this random but positive process. You don’t have to press F9 very often before you get a history that loses money in a straight line for five years which has a 40% drawdown. Remember this is something that's guaranteed to make 20% per annum over a long enough period. So the expectation of losses is something that everyone should build into their investment process at all times.
The majority of our investors are institutions, they’re pension funds, insurance companies, sovereign wealth funds etc. to be very fair these investors they are really quite sophisticated and they understand that. Sometimes when you’re speaking to high net worth individuals and smaller family offices, the desire to protect capital is much stronger; you have to be extremely clear about the fact that there will be things that will happen in the future that will be unpleasant. The other thing you have to explain to investors is that when you’re making money, it probably won’t last.
On discretionary trading...
Obviously by the time somebody is sitting on the opposite side of the table from me they’re probably not going to be asking me that question and I am of course very clear about the fact, I cannot see into the future, I’m not psychic, so therefore I don’t have any skill in that. My skills lie elsewhere. There is an argument, that at least in terms of positioning longer-term, maybe nobody has skills like that, and maybe that’s just impossible.
I think the demand for discretionary trading probably comes from reactive demand, it’s impossible to rate a model that can react to every last move in the market, or react to a piece of information that has come in. September 11th is probably a very good example of a piece of information that arrived, the world changed at two minutes past nine and systematic models knew nothing about it, whereas discretionary traders clearly did. So in those kind of events, it’s quite possible that discretionary traders will outperform and I can understand why people have that demand for it. But it is an extremely difficult job to be a good macro trader and we all know of those people who have great reputations in doing that and it's maybe only a handful of people.
Of course the interesting thing about macro trading is that everyone wants to be a macro trader, it’s really interesting when you hear stock pickers talk, discretionary, equity long / short, very often what they’re talking about is macro things. We think the market's going up, we think it’s going down, and then therefore we are going to buy these stocks. Quite often macro decisions are wrapped up in security analysis, which it probably shouldn’t be. I really think trading macro on a discretionary basis is one of the world’s hardest jobs. And it’s possible that the people who are successful at it, may just be the lucky pennies.
On systematic trading...
There are people who say ‘we have a model but we occasionally intervene’. My view of that is that is not systematic trading. Systematic trading, I’ve used this phrase before, it’s a little bit like being pregnant, you either are or you aren’t. You can’t just be a little bit systematic. I often say to people - maybe pitching me with systems which they then put a discretionary overlay onto - I say 'why don’t you just run two books, run the model on one book and run your discretionary overlay on another book and just see which one makes money?' Once you think about it that way, the decisions are often a way for people to hide really quite complex decisions.
Remember the point of what we’re all trying to do is we’re trying to say ‘what’s more likely to go up tomorrow and what’s more likely to go down tomorrow?’ and that’s the decision. What we have, and many people like us, is we have lots of complex or sometimes simple models which have been tested statistically over many, many years and they're weighted using very sophisticated weighting algorithms and cost control measures and all of these other things, literally millions of lines of code running.
And all we’re trying to do is forecast whether or not something is going to go up or down tomorrow. Very often people intervene in their models by saying things like "I think vol’s going to come off tomorrow" or "I don’t think this model is going to work as well tomorrow". Now that’s a very complex statement, the amount of analysis you would have to do, for example, say a trend following model isn’t working very well and is unlikely to work tomorrow. Since it’s so difficult to do that on something as simple as an asset, the idea you can do it for something as complex as a model which is running on over 120 different assets with various risk weighting and cost control measures on top of it, seems to me to be unlikely.
On proving a strategy is broken...
I don’t think I’m the first person to come up with that, effectively that is the scientifically method that has been working pretty well for humans since the Greeks came up with it. Science is about proving things wrong, it’s not about proving things right. What you’re trying to do is break your strategy, you can never really prove that a strategy works or doesn’t work.
Finance - despite pundits, newspapers and people on the television and maybe even people on the podcasts, from what they say with certainty: "this is going to happen, this is the best way of doing things, our systems are better than somebody else’s systems etc", finance is full of all that. But in fact finance is dominated by randomness, randomness is everything. And so because randomness is everything, you need to be uncertain, you need to have a lack of conviction in certain things because you want to be able to prove that things are wrong.
So when we come up with a new strategy or new idea or new trading system, what we are trying to do is find out what’s wrong with it. And then after months of testing and re-testing and re-thinking about it, if it gets through months of that, then maybe it’s going to work in the future. These are all very weak uncertain statements, but to a certain extent our philosophy of what we do is around that uncertainty and conviction. There are certain things I have conviction about, obviously, I have conviction on the importance of technology, conviction on the importance of risk weighting or coherent portfolio design…but I don’t have conviction about any particular technique. You just do what you think is best and what you think will be persistent.
On doing something that is socially useful...
That’s a hard question to answer, once you start going down that route, what is useful to the economy or useful to the society? The world only needs one car manufacturer; it doesn’t need hundreds of manufacturers or hundreds of models. Nobody really needs expensive meals and restaurants, so expensive meals and restaurants aren’t really doing anything for the economy. Nobody needs podcasts, we can live without them. Once you get into that, you’re getting into some dangerous water.
We are, I believe over time, providing something that is valuable to our investors, I think we are doing that as well as we can, we’re constantly trying to add incremental improvements that make the returns better. We won’t make money every month, we won’t make money every year…we’re doing the best we can to produce positive returns for investors. And they are a wide variety of people, they’re pension funds and where I feel - although this is not quite the same as finding a cure for cancer and certainly not the same as landing a probe on a comet - they are fantastic things to do. This is what we do and we make the returns of our investors’ portfolios better in some way.
Where I think there’s a problem of where I think people are doing something which isn’t socially useful is when relatively simple strategies are dressed up as something complex. People effectively being an index tracker, but charging 2-20 for it. The other thing to remember of course, is trend following systematic trading, the money comes from somewhere. If we make money for investors somebody loses it.
I do believe we do a relatively good thing, if nothing else I provide employment for people in Cambridge, it’s important to remember that this little community is an employer and pays people money.