Quote from bwolinsky:
Looks good. Started in 1974, their site shows a compounded 18% since then. Excellent.
<i>
Trading Strategies
DUNNâs WMA program utilizes a long-term trend-following strategy. This is a full reversal strategy in which the program is almost always invested, either long or short, in each market it trades. The WMA program seeks to extract profits from up and down trends, resulting in a stream that exhibits a low correlation with traditional asset classes. WMAâs average holding period is approximately 6 months.
Revolutionâs Mosaic program (exclusively licensed and distributed by DUNN) utilizes a short-term pattern recognition strategy. Mosaic is an ensemble of trading models that seek to identify higher order patterns via statistical analysis and exploit these patterns with trading algorithms. Mosaicâs average holding period is approximately 4 days.
</i>
The WMA is a bit volatile and not my first choice of indicators, but that was really all there was back then. The WMA's average holding period is too long to assign any significance to their trades. Any more than a month and no matter your rationale was for entering probably had nothing to do with anything that happens that much later. This probably produced outsized returns since inception, but I know the WMA holding period of 6 months is too long to be considered credible. Much more research than technical trading seems to go into that.
Their mosaic program I believe are more realistic and probably account for the majority of their profits. I don't know if your performance statistics are true, but I have no reason to question them.
Their indicators holding out 6 months is a red flag, and anybody watching continuous commodity futures since that time is just going to see a big up swing for most of the reporting period. Although I have periods in my strategies holding for multiple years, I know that as long as you're with the trend you should hold onto it until it reverses. That sounds like what they're doing here.
The short term pattern recognition most likely doesn't know why their methodologies work. Statistically a lot of spurious correlations can be created even if there is only a slight outperformance.
A 50% drawdown is reasonable, but this probably does put them in the high risk category.
What this means to me is that if I can produce returns at the historical average of 150-180% annually there is no problem attracting a lot of money to my trading practices. I have less data than them, but most likely the in-house development of the WMA identified bottoms more than likely, actually, and these were used as the basis for taking leveraged positions in lognormal futures contracts.
Dunn would be an interesting to company to read about for sure, but I have to say nearly all of their strategies from the 1974 period could probably only handle daily, weekly or monthly data. It doesn't look like if their holding period is 6 months that that classifies as trend following. More than likely the research team researches the best commodity plays, and virtually any program will give signals when taken over a long period of time that might buy on a dip or short a high. Doing so is not actually that simple, but the programs sound like they have built enough track record so that no one questions the soundness of their strategies.
