Estimating Model Lifetime

Joaquin Narro and Monica Caamano

Although they are closely related, we must differentiate between the lifetime of a model and the lifetime of the underlying strategy. We would say that a model is born when it is ready to be traded, regardless of where the underlying investment strategy is in relation to its cycle. In other words, the model is born when it “finds” the strategy. The model cycle then traces the investment cycle of the underlying strategy. Investment strategies undergo profitability cycles in response to factors such as changing business conditions, the number of competitors entering or leaving the industry, the type and magnitude of opportunities available, etc.

The model dies when the performance of the underlying investment strategy follows a pattern consistent with a purely stochastic process – that is, the performance of the underlying strategy is just random noise, which is an indication that the market inefficiency has closed. We presented an example of a randomly generated equity function for a target model strategy in Chapter 5, where we saw how it is possible to have positive performance over a sustained period even though the underlying process is random.

CATEGORICAL LIFETIME