Michael Covel: News & Info.

An email that came in today (edited down some for readability):

“I worked at a large London futures brokerage firm for much of the ’90s…so had first hand insight into enough of the daily business in terms of trading of Bill Eckhardt and Richard Dennis and many of the ever growing army of “system traders” old and new, such as Dunn, to admire and respect the validity of their strategy.

“They created systems that all, in one way or another, aimed to keep losses linear and profits exponential. All other risk management strategies / systems added on over the years to the initial simple trend following technique, such as time horizon profit takes, fractal time countertrend opportunities etc were essentially bells and whistles adjustments to tweak and enhance the mainframe structure. This mainframe, as you well know, was by definition a long volatility strategy, which relied on the natural kurtosis (”fat tails”) of the products traded to ensure “upside” (profitable) vol. Provided an expected payout profile in excess of the associated expected loss profile suffered during periods of “downside” vol. In essence the system was to make sure you retain enough capital (linear losses) to always be in the market to capture the unforeseeable occasions of exponential gains. Diversification of one’s portfolio across as many uncorrelated contracts as possible was therefore the obvious strategy to counter the truth of unknowable future price trends. So why did the good times stop for trend traders after the 90’s in general? And specifically in the last 3-4 years? I believe it was due to the combined effects of 2 major market changes; the first was the macro economic denouement of the global central banks’ successful anti-inflation strategy. The consequence of winning that war was the advent of an unprecedented era of asset price boom and busts fueled by cheap credit (the yen carry trade)and maintained by seemingly well placed investor confidence in the “Greenspan put,” (LCM bailout, 1% Fed Funds rate following dot com crash to name but two) and pertinent to our system trader heroes, the inevitable resultant long relentless trend to ever lower levels of volatility across all asset classes. That was the 1st major market change. The 2nd was a changed technical environment brought on by the system traders themselves. I know this to be true because of the personal experience of a great friend of mine. A very gifted oxford educated mathematician. He left the main stream in ‘96 to see if he couldn’t replicate and improve upon the successes of the system trader clients I was illustrating to him. After some pioneering personal r & d…he turned £200k into £1 mill in a fully audited personal trading account between ‘98-2000. His systems had at their core the results of the work he had done on the individual characteristics of ,and relationship between, volatility and price action. Volatility is mean reverting over time though produces clear patterns of “cluster formations” in between times, whereas price action is essentially random and prone to the “fat tail” excessive movements I mentioned earlier. The comparatively innovative risk management system he built centered on his bespoke knowledge of volatility and price action allowed him to investigate and harness data from ever shorter maturity time periods. The fact that he could in effect diversify his portfolio not, as traditionally, across products, but across time frames was going to be a neat solution to scaleability problems encountered by others. He focused mainly on a big liquid product such as major f/x pairs, which uniquely traded round the clock. I suspect he was a couple of years ahead of the smarter market quant geeks in his work and some proof of this was his funds results in 2001-2004, which put him in the top percentile of f/x system traded funds. However, his results in absolute terms had already declined. The Deutsche bank index for f/x systems was averaging negative returns, so by comparison his low teen returns, and continuously excellent sortino ratios (which he always targeted primarily) were good, but not exciting enough to attract investors from more main stream assets such as equities. As his competitors caught up with him and more and more systems came into the intraday time frame the acceleration towards the shortest time frame viable became inevitable. At sometime before time zero seconds (personally i think it is nearer time 15-20 minutes even for f/x, when considering funds with AUM of USD 100 mill+) there is a breakdown in the recognizable and therefore codeable relationship between vol. And price. I liken it to a physics experiment i remember from school where we tested the elasticity of different materials. At some point, you reach elastic breaking point and instead of mean reversion the elastic snaps. When it snaps all previously relevant expected reactions are null and void and a state of chaos is gone through. This happens at the short maturity limits of algorithmically codeable vol. Patterns and associated market price action. I believe that the surge in market participants, even with some systems more valid than others, is even as i write this continuing to push system intraday players up against the rocks of this inescapable limit. Worse than this i even suspect that this breakdown point is expanding out along the time curve in correlation to the increasing market participants. So as i see it, these 2 major environment changes: ever decreasing volatility in an already low vol. Environment, across all asset classes and the technically self defeating environment arrived at in short time frame horizons have been the main reason for system traded fund under-performance. Individually they are significant factors but jointly, as seen over the last 3-4 years the effects have compounded. If my memory serves me correctly Dunn chose to follow a reverse system strategy generally, meaning that the system was always either long or short but never flat. In other words his system will naturally risk greater amounts of linear losses but enjoy the associated exponentially greater amounts of profits. Put simply he geared his system to be longer volatility than others. And this has been a losing strategy in the prevailing market environment for most of this decade, and highlights the inherent weakness all systems must work hard to counteract (with their bells and whistles), namely they are a net long vol. Strategy. Whilst you can stay in the game longer by limiting losses to a linear decline, if the losses go on an historically unprecedented run, in theory you will need capital funding resources tending to infinity. Postscript; Dunn’s decline does not negate the validity of system trading in general. The intellectual integrity at the heart of a simple trend following system remains undeniable. Dunn is a cautionary tale of what systems are and are not claiming they can do. They are not claiming to be infallible in all cases, just to be the most consistently efficient method over time of managing risk in an unpredictable price action environment. Personally am bit surprised Dunn seems not to have evolved more checks and balances into his system (eg counteracting short vol. systems) over the years. This is what the smarter system writers have been doing. Of course that means lower upside return expectations as well…But hey, they should know better than anyone else what the golden rule is. If you run out of money, you can’t stay at the casino.