NUMBERS: IT’S ALL ABOUT PROFITS
Trying to describe the methodology of Numbers SA’s Global Diversified Futures program is difficult as there are many elements, but they all work together to produce solid risk- adjusted returns. The program is up 54.51% since its February 2008 launch with a worst drawdown of 6.26% and a Sharpe ratio of 1.88.
It combines mean reversion, trend following, statistical and spreading strategies. In addition to the 11 independent sub portfolios, Numbers applies different time horizons to different sectors.
“The mean reverting is one algorithm that applies to all markets. In the portfolio structure we try and diversify the time horizon,” says Numbers CEO Bertrand Savatier. “For example, stock indexes work on a five-day trend duration whereas currencies work on a 15-day trend duration.”
The different time horizons are not based so much on what works in the markets, but on adding diversification. “We want the portfolio to be diversified by time horizon. It is basically a choice of portfolio construction,” Savatier says.
Numbers applies its mean reversion and trend following strategies to interest rates, currencies and stock indexes; it applies a fundamental mean reversion and trend following approach to commodities as well as a strategy to trade calendar spreads, and it applies a statistical model to all the markets.
Everything is modeled in this highly systematic approach. “[For] each instrument, we have 50 or 60 data series we study,” Savatier says.
Applying both the mean reversion and trend following approach to many markets means that the program will, at times, have opposing positions in certain markets. “In mean reverting we look at two trends that are completely independent of the trend following model. It could reinforce positions within the trend. If there is a correction within the middle-term trend, then the mean reverting can reinforce the position in the direction of the middle term trend against the short-term trend,”
While they look at fundamentals, it is not from a discretionary seat of the pants perspective. Numbers will assign values for fundamental inputs and work that into the program in a completely systematic fashion. “We have some models that are purely fundamental. Then we combine those with models that are purely technical,” Savatier says.
As if this combining mean reversion and trend following is not complex enough, part of the program seeks to exploit money flows in the commodity sector. The strategy is long-term and does not simply put on a bear spread in front of the commodity rolls.
“When trading calendar spreads, we try and take into account the evolution of the [contango/backwardation] situation,” Savatier says. They could be trading with or against the commodity funds. “The big advantage of this strategy is its non-correlation, even with the underlying instruments. A calendar spread on corn has zero correlation with the way corn is moving,” he says. Creating numerous non-correlated streams of return is the key to Numbers’ complex systematic approach to trading. It is why they also include a statistical model applied to all markets.
“It is a simple algorithm that is placed on different markets,” Savatier says of their statistical model. “If you trade a pure trend-following system, it is difficult to add diversification because even if the algorithm looks very different, they tend to react the same way in the same situation. If you look at very different sources — including fundamental analysis — then at some point you will be different,” he adds. In backtests, the program has been positive in each of the last 27 years. They test intensively to ensure the program will perform in different market environments. Even so, they are not resting on their laurels. Savatier says they are looking to add a currency overlay based on the carry trade. If they do, expect the new model to be tested in every possible way. Numbers is different and they appear to have covered all of their bases.
For more information on Numbers, visit their Web site, www.numbers-am.com