The Striker Report - gold standard for objective trading research

Paul Priestley

System Developer

Program(s) Developed: Diffusion Trading System

Interviewed by John F. Gallwas, Founder of Striker Securities, Inc.

Priestley Systems developed "The Diffusion Trading System", one of the top performers at Striker in 2013. The owner and system developer is a European trader and wishes to remain anonymous. Messages can be forwarded to him by Striker. After majoring in physics at Cambridge and Oxford in England, he has pursued a career in teaching for the past forty years, as well as using computers to trade his personal account. He built his algorithms from scratch for his own trading and has only recently permitted other traders to participate.

John F. Gallwas: The first question every reader wants to know is why anonymous?

Developer: Well, I just put a lot of value on privacy. A successful system could turn its developer into a minor celebrity and I don't envy celebrities their lifestyle one little bit! Life would be less pleasant if I had people calling me at every hour of the day to ask about my systems. A lot of people miss the days before mobile phones when you could enjoy a meal in a restaurant without getting phone calls. I don't use a mobile and I still enjoy living in that world!

John Gallwas: What was behind your decision to lease your systems to other traders?

Developer: Since I've already put the work into developing them, I may as well enjoy the extra income stream. We only trade markets with good volume and high liquidity and usually at the most liquid time of the day, so I don't need to worry about shifting the market price until my number of clients is quite large.

John Gallwas: Tell us why your systems are based on pattern recognition?

Developer: Trading methods are based on a variety of things. Some use market fundamentals, which means one person's opinion versus the combined opinion of everyone else in the market. There are good reasons for the majority to win most of the time. Some systems are based on a spectrum of beliefs from religious texts to astrology: I can't really comment on those! Some are based on looking at graphs and trying to spot patterns. When a human being is involved in the interpretation of patterns, a subjective component is inevitable and what that person wants to see can have a big impact on what he sees.

The only way I know to research market behavior scientifically is to develop an algorithm which a machine can use to identify some trend or pattern. Even chaotic data can have patterns underneath the random noise and computers can now be used to find subtle patterns which were not obvious back in the day when people were trying to spot chart formations by eye. You won't find an algorithm which wins all the time, but you only need one with good probability.

I'd like to say a word about "curve-fitting", because the expression has a pejorative connotation which belies what it really means. By its very nature, any system which extracts predictions from past data is curve-fitting that data. If you're going to analyze data scientifically, you can't help curve-fitting it. What is important is to recognize that for what is, to understand its dangers and to avoid them. You can fit a formula to any data-set at all if your formula is complicated enough. You can avoid that by using the largest data-set available and seeking the simplest algorithm.

John Gallwas: What are your other key objectives when designing a system?

Developer: Return alone is of little value if you have to take big drawdowns to get it. I look for a high ratio of lifetime return to the biggest peak-to-valley drawdown. The more consistent the slope of the equity line, the better the chance that the algorithm has worked consistently.

The profit-per-trade is important too, because of trading costs. I once developed a system whose lifetime return (before commission and slippage) was over a hundred times the maximum drawdown, but it wasn�t tradable in practice, because the profit per trade was comparable with the bid/ask spread in the market.

Because I developed my algorithms from scratch using my own methods, they have a low correlation with other systems.

I don't reveal the details of my algorithms. I've seen developers of successful systems do that and then have their systems copied.

John Gallwas: Describe the key points of your Diffusion System, which trades natural gas, and has been a frontrunner in Striker's performance tables.

Developer: This system does have a fundamental component in that it trades long-only. I developed it as a reaction to a lot of studying I'd been doing about energy markets and the direction I think they're likely to move over the medium and long term. Discoveries and new technologies have pushed the gas price down so much that industry has a big incentive to use more of it. The time lag to develop the technology (for natural gas powered trucks and planes, for example) should make the price trend upwards over the next few years. But the natural gas price has always been volatile. So I looked for an algorithm which could harvest a price rise over the medium term while avoiding some of that volatility.

John Gallwas: Do you need to update your trading systems to keep pace with changing market conditions?

Developer: It's amusing to see some of the systems advertised on the internet, which are always being updated. The updated systems have an impeccable record in back-testing, but they're not always so eager to show you their performance going forward. I remember watching one developer keep adjusting his system while his archive version always seemed to do better than the new one in real time. One of the great benefits of a site like Striker is that it lets you see how a system actually performs.

Having said all that, it can sometimes be valid to make gradual adjustments in parameters, providing the adjustments are based on your whole (large) data-set and not just on recent data.

John Gallwas: Will you be introducing additional products in the near future?

Developer: I've always one or two ideas on the back burner, but I can't predict when I'll be happy enough with something to release it.

This interview is for informational purposes only and is not intended to be a solicitation of any kind. Trade only with risk capital. The risk of trading can be substantial and each investor and/or trader must consider whether trading systems are a suitable investment.