Everyone who participates in the capital markets uses some kind of model to guide decisions to buy and sell. Sometimes the model is strictly fundamental driven, such as "Prices go up following good news." (Actually they go up beforehand.) Other times, the model is technical in nature and contains assumptions such as "Prices revert to the mean." (True, except when dangerously untrue.) Most of the time, the model is based on some hybrid of assumptions about the interactions between fundamentals and price. In every case, long term profitability is a function of how much or how little the model has in common with the way prices actually behave.

At iSigma, our market timing model is based largely on fractal geometry. As Benoit Mandelbrot has argued, the movements of market prices have more in common with the edges of a shoreline or with mathematical phenomena called Levy processes than with the random effects of coin tossing. In practical terms, this means principally two things. First, price series are very similar at any given scale. A one day chart won't be too different from a one year chart. Also, extreme events are far more common than classical statistics might anticipate. Long trends which ought to happen once every million years in a random process manage to happen all the time.

In addition to the large body of growing empirical data to support the case for the fractal model, there are a large number of theoretical reasons to accept such a view. While random walk models predict longer timeframes to be characterized by more smoothing and reversion to mean, fractal models result in different expectations. Market prices are determined by the actions of market participants operating in multiple timeframes. Market participants respond to political events and fundamental developments which occur in multiple timeframes. These political events and fundamental developments often cause extreme reactions among market participants. It only seems sensible to conclude that market behavior will be similar across timeframes and filled with extreme events.

Ever since Mandelbrot started talking about fractal geometry in markets, analysts have tried to put fractals to use ina variety of ways, some better and some worse. There are countless expensive plugins for technical analysis software claiming to contain the fractal indicator or a toolbox of fractal indicators. Some use rescaled range (R/S) analysis and Hurst exponents while others utilize statistical measures such as kurtosis and skew. These are fine for confirming that markets are fractal in nature, but they don't provide any useful information related to trading.

Unlike the majority of so called "fractal systems" for trading, our trading methods are based on fractal geometry, but they are not a matter of combining conventional indicators such as moving averages with some fancy sounding mathematical functions. Rather, the strategies we use are totally unique and entirely quantitative, built from the ground up take advantage of the fractal nature of market action.