Tuesday, February 10, 2009

Signal Analysis as a Model for Stock Price Momentum

Technical analysis is a fairly useful modeling technique to determine the momentum in the market and are very similar to how signal processors filter out components in Electrical engineering. The fractal-like characteristics for different time horizons can give mixed signals when it comes to buy or sell decisions. The key is to pick an appropriate time window which is appropriate for the current market and stock.

A signal being analyzed will have different frequency components which are similar to price movements in the stock. For instance, an active day trader would have a similar view as a high-pass filter. They have short time horizons and focus on the volatility in the stocks prices.

In contrast, A moving average technical trader or buy and hold investor by definition has a longer time horizon and their behaviour is similar to that of a low-pass filter. By ignoring the noise in the market and day to day volatility, they can ride the larger and longer swings in a stock price.

Particularly in this environment, using regression analysis to forecast movement can be dangerous. For different investment styles, one might see the start of a short term reverse head and shoulders formation (indicating a buy signal) versus another who might see momentum heading for decline.

The best advice is to incorporate the appropriate leading indicators into your model which can help paint a clearer picture. The problem is that current volatility across different time horizons may give more false positives so it falls on those who design the models to be prudent when reading tea leaves in the market.

No comments: