The Gateway to Algorithmic and Automated Trading

The Automated Trader Interview

Published in Automated Trader Magazine Issue 14 Q3 2009

We vary the theme this month, away from this feature’s customary focus on a purely “machine-driven” fund. John Howard, Publisher of Automated Trader, went to meet the team behind the Danix Master Fund, where a system-based approach co-exists happily with, and complements, a much more hands-on style of trading.

The Danix Master Fund has $93million under management. Its core trading styles are an interesting mix of discretionary macro (60%, utilising short-end yield curve trading and options, with an additional 10% 'best ideas' provision) and systematic (30%). The fund's objective is to "balance performance with risk management, targeting consistent risk-adjusted returns", and the role of the systematic component in the portfolio is to provide both performance and anti-correlation on the way to delivering those consistent overall returns.

Markets traded are: STIRS, forex, government bonds, equity indices, gold and crude. The overall fund returned 9.23%, 16.75% and 31.28% in three different share classes in 2007, and 15.37%, 24.44% and 49.6% respectively in those classes in 2008. Margin-to-equity targets of 7.5%, 12.5% and 25% are applied.

John Howard spoke to three partners in the firm: David Dacosta; Nicholas Wood; and Exton Male.

Danix Master Fund

John Howard: Let's start by talking about your recent performance, first for the overall fund, and then specifically for your systematic trading element. Would you say you're satisfied? Were there any surprises?

Nicholas Wood: Overall, we're satisfied with the performance. We've got a pretty even attribution between the discretionary macro and the systematic side, in proportion to their representation in the portfolio. 70% of the profits come from discretionary macro, which is split roughly half between the interest-rate spreads and the options, and around 30% of the performance was from systematic.We like the way those two parts of the portfolio anti-correlate. The systematic and the discretionary have a low negative correlation with each other. That obviously helps the volatility of the overall fund. And we are particularly pleased with how our systematic side held up over the last eighteen months, during the shift from low volatility to very high volatility.

John Howard: Thinking about that, what did you get right?

Nicholas Wood: Strong money management and incorporating volatility as part of the system. The systematic side delivered alpha throughout the big shift in market behaviour. In that sense, recent conditions were a good test of the robustness of our systematic approach. It came through, and we're pleased about that.

John Howard: What prompted the adoption of systematic trading?

David Dacosta: Much of the technical analysis we had been using to support our macro views had a proprietary twist to the extent that our approach merely needed formalising.

Most of our discretionary trading activity is either yield curve or options based. The products traded in the systematic portfolio are invariably different in that they are stock indices, commodities, bonds and foreign exchange but they are also traded on an outright basis. Also, most of the products traded on the systematic side have multiple profiles and the ability to get back into trades which have been stopped out. And lastly, the P&L profile is totally different in that the systematic portfolio has a much lower percentage of winners but much tighter stops and much greater targets mean that the winners can lead to risk/reward ratios as high as 30:1.

John Howard: How do you come up with your ideas, and how do you test them?

David Dacosta: Reading of preferred economists, talking to other proprietary traders and analyzing economic statistics. We do historical backtesting with continual assessment . We source our historical data from CQG and Bloomberg.

Nicholas Wood

Nicholas Wood

"The volatility of our returns has remained the same even with the massive increase in market volatility. That in itself tells a story."
John Howard: On what software?

Exton Male: Custom-designed. We are continuously evolving the way we trade our system, how we bring on new instruments, how we execute, how we monitor execution, how it performs in changing markets. This hunger for information can only be managed by having our own in-house software with our own in-house IT professionals to make the changes necessary to achieve our goals. We use Java (1.6) and C# on .Net (3.5).

John Howard: What are the key metrics you're looking for?

David Dacosta: Consistent returns, non curve fit, minimal draw downs. To avoid curve fitting, our iterative refinement of parameters has to have limited impact on performance.

John Howard: How have the models performed since volatility has subsided?

Nicholas Wood: The system got us long various stock indices, and Crude recently. May this year was our best systematic month since April '07.

John Howard: You trade across multiple instruments, to what extent do you 'tweak' a model's parameters for different markets?

Nicholas Wood: We find that models parameters can differ markedly, reflecting the way they are traded. We don't feel the need to use the same or very similar parameters for each instrument. As long as it is robust.

John Howard: How far does a model have to fall outside the window of expectation before it gets put on the bench?

Nicholas Wood: It doesn't but a consistently poor performance will shrink its participation. One of the reasons for shrinking is to stem the losses in periods of drawdown. At no stage do we want to be out of the game. Over time, we 'top-up' the risk for each instrument. Doing this means that should a trend develop after a highly choppy market, we can still participate.

John Howard: In our conversations leading up to this interview, you've talked about consistency. That must have been a tough objective to maintain over the recent past.

Nicholas Wood: We've seen very consistent returns since August '07 and the start of the credit crisis. The volatility of our returns has remained the same even with the massive increase in market volatility. That in itself tells a story. All three areas - interest-rate spreads, options and systematic trading - have worked for us since August 07.

David Dacosta: It doesn't really matter whether interest rates move up or down; you just want them to move. If you go back a few months, and look at a period running back about eighteen months from then, rates were moving pretty quickly and there were large distortions, anomalies, on the curve that we could exploit. Over the last few months, it's calmed down. Rates are very low, and short-term expectations aren't great. Most people expect, with all the money being pumped in, that rates are going to start moving back up again.

Nicholas Wood: The other part of discretionary macro is more directional; it's the expression of our top-down macro view via options, and that performed very well last year. We were able fortunately to predict that interest rates were going to be cut extremely hard, and we were also able to take advantage of the mispricing of the tightness at the very front end, which was representative of the banks not lending to each other. This year it's calmed down quite a lot but we remain ready to take advantage when opportunities present themselves.

Exton Male

Exton Male

" times of lower volatility, when our discretionary trading may have fewer opportunities, good medium term trends can develop, and our system smooths out our performance by providing alpha."
John Howard: I could ask one of two questions here. If discretionary macro works for you, why systematic? If systematic works for you…

David Dacosta: It's active systematic; we haven't always got positions on.

Nicholas Wood: We've run systems for as long as I can remember. We've run them on stock markets, on interest rates, and more recently we've run them on overall trend following. We like systems because they provide an unemotional contribution towards alpha which dovetails very nicely with our own discretionary capabilities.

John Howard: Unemotional …

David Dacosta: All the emotion goes into creating the system.

John Howard: It's as if you have a strategy of balancing one side off against the other, systematic against discretionary? As if the system is a form of diversification for the discretionary trading, and vice-versa.

Nicholas Wood: We'll bring anything into the portfolio that we feel adds value. To add value it has to add diversification. It just so happens that the systematic profile anti-correlates with the discretionary profile, and in that sense, the two together produce higher-quality returns. Lower volatility, lower drawdowns. In that sense, it adds value to run them together rather than have them stand alone in separate funds. That's a bit unusual, we think, but we like it.

John Howard: Is it going to stay unusual? I mean, are you going to carry on the same way in the future?

Nicholas Wood: On the systematic side, we'll just keep on adding more products to it. We'll keep on adding new programmes if they add diversification.

John Howard: You've talked about how the overall systematic element has a negative correlation to your discretionary activity, and thus works as an effective equity smoothing overlay for the overall fund. So, do you correlate a model's returns against your historical discretionary performance - and if so, how much positive correlation would you permit for a model to enter the production set?

Exton Male: To us this would be a form of curve fitting. Our aim is to build a medium term trend-following system that produces alpha on a stand-alone basis. The positive side effect is that in times of lower volatility, when our discretionary trading may have fewer opportunities, good medium term trends can develop, and our system smooths out our performance by providing alpha.

John Howard: What is the typical lifecycle of a systematic trade?

Nicholas Wood: Five days, and we generate seven orders each day.

John Howard: How far do you allow discretion to enter the systematic environment? Do you ever reject signals on the basis of a personal view?

David Dacosta: Never. There have been so many occasions in the past where every instinct in your body is telling you not to take a signal, or to take off a winning trade just before it reaches its target. Once the system was fully in production we agreed to run it robotically. The risk rewards on the trades in our system are high. By discretionarily ignoring or interfering with trades we could decimate expected performance.

John Howard: Describe the execution process once a signal is generated.

Exton Male: Software tees up the order which is manually checked and executed. The operation is still at a level where we have the benefit of being able to add a human sanity check before going to market. Having said that, we are working on a fully automatic process which will only come into play after stringent testing.

John Howard: A human sanity check?

David Dacosta: Yes. Very important.

Exton Male: At our stage of development, although we do trust our own technology of course, we sleep a little better knowing that we've always got a human casting an eye over orders before they hit the market.

John Howard: It's an objective of yours to eliminate the human element from your software development. What is it that the human does now, that the machine will be doing it a year?

Exton Male: We're putting in a series of failsafes. We could write them down on paper now, but we haven't wanted to prioritise them just yet, because that would take resources away from developing the system.

John Howard: And a human is a good enough failsafe for now. I get it. What are you prioritising now?

Exton Male: At the moment we're focusing on the software that analyses what happens when we add new programs to the system, when we change the parameters slightly. That's our present focus. That will enable us to bring new programs on more efficiently.

David Dacosta: There's another point here. When bond-market volume evaporates prior to an important statistic like US non-farm payroll, for example, if you have systems that trigger the nearest bids and offers, you can find that you trade as much as two points outside the true market value, which can knock a very big hole in your P&L. Seconds later, the volume returns to the market and you're back where you were before. We'll pull the orders out when the volume evaporates and put them back in when it comes back. That might sound over-protective, but speak to anybody who's lost his bank ten million dollars because he was selling at two points down, and he'll tell you that it's worth it.

John Howard: To save ten million dollars, I suppose we can devote a paragraph to defending human intervention in systematic trading, even in Automated Trader. But let's move on before anybody hears us. Tell me about the team. How do you work together, how long have you worked worked together, what's the philosophy, if you use that word, that keeps you together?

David Dacosta: We have a morning meeting at which we'll discuss any fundamental ideas. Then Nick and I will discuss trades together. On the systematic side, apart from adding in new programs, that runs consistently and requires no discussion. Apart from the intererence we just mentioned, with volume before important numbers, we don't interfere in any other way.

John Howard: You obviously enjoy working together. How long have you known each other?

Nicholas Wood: I've known Exton since 1992 and we worked together from '92 to '98. I've known David since 1995 and went to work with him at a proprietary trading house in 1998. Exton joined us in 2003. Danix started in 2005. We've spent half our professional lives working together. We have over fifty years' combined experience working in the industry, through both the good times and the bad.

David Dacosta: We had worked for a proprietary trading shop, and we'd worked there for many years. We felt that we had all the necessary attributes to launch our own fund. The company we were working for, supported us. They said, if you want to start your own company, that's fine as long as we can be the seed investor and we can back it. Their speciality is in metals, and we ran their financial side. They're still with us.
John Howard: And it's a happy relationship?

David Dacosta: Yes. When we're making money out of the market, our previous employer likes us to invest those returns in an aggressive way. As long as generally you provide a good P&L at the end of the year, they're happy. Some of our other investors prefer a smoother profile. We have different share classes so that some investors are traded more aggressively than others.

David Dacosta

David Dacosta

"Our strong money management means we protect the downside aggressively but that doesn't impact our ability to make healthy returns when it's our kind of market."
John Howard: You've worked together as a team for a long time, and you have a long-term relationship with your former employer. Going back to how you work together and the philosophy, if I was talking to your former employers about you, what do you think they might say?

David Dacosta: I think they'd say returns are consistent. When we're going through a period when our style of trading, or global macro, is quiet, our money management will mean that we're unlikely to go down very hard. We'll be either side of zero, generally slightly up, but that doesn't mean we haven't got the ability to accelerate when the opportunities present themselves. Our strong money management means we protect the downside aggressively but that doesn't impact our ability to make healthy returns when it's our kind of market.

Nicholas Wood: If you want to talk about our philosophy, the easiest way to sum it up is just to say: we hate losing money. We wouldn't be happy to be down five when the index is down ten; we want to be flat or up one at a time like that. Right now, for example, we're up 1% on the year, while the global macro index is down 5.5 and the systematic diversified index is down 9.3, both according to HFR.

John Howard: Tell me a bit about risk management.

Nicholas Wood: We run VaR calculations daily. We're not leveraged so in that sense we have margin-to-equity alert limits and our various managed accounts run their own VaRs daily, so we've got additional checks outside our own checks. We have limits on what we can allocate to individual trades, and we have a policy of always being net long options of any one expiration that we're involved in. If you look at our daily P&L results, you'll see that our best-ever day is twice as big as our worst-ever day, and our winning months are twice as strong as our losing months. That probably reflects the long-gamma, long-optionality approach of the portfolio overall.

David Dacosta: Any trade we get involved in, be it systematic or otherwise, we always have hard stops. It's not so much a reversal situation; as the order goes in, we know exactly where the hard stop is. On the systematic side, eventually a trading stop may take over, but you know your maximum kick on the trade as you put the order in. We always have hard stops. We're completely transparent, we trade on recognised investment exchanges, except forex, and everything we trade is liquid, and we don't have any capacity constraints.

John Howard: Okay. Last question. If you were to give one piece of advice to somebody starting out to do what you're doing now, what would it be?

Nicholas Wood: Make sure you've worked in an environment that is like working in a hedge fund before you try and set one up.

David Dacosta: Even if you have proprietary traded before, on a true proprietary basis, no customer flow, no commissions, there is a difference between having new investors coming in every month, and investing for one investor. In the classic case, quite often, when you start making money, your investors are going to want you to be more aggressive with the money you've made from the market, and that's when you take a hit. That's a more difficult profile for a hedge fund; for somebody who's just come in, you're more aggressive than they expect. Also, running your own fund, you have to be sufficiently well organised to be able to spend most of your time trading. If you're not careful, you can end up involved in meetings with lawyers, accountants, administrators, apart from new investors when you make presentations. And with systematic trading, make sure you haven't fooled yourself by having something that's a complete curve-fit.