V.3:5 (162-165): Data Smoothing using a Kalman Filter by Vince Banes

Data Smoothing using a Kalman Filter by Vince Banes

"The analysis of continuous pricing information works well with this form of filtering"

The concept of optimum estimation was introduced by Dr. R.E. Kalman in 1960. The first major application of his theories was to reduce the noise found in the data of modern navigation systems. With these theories, system designers were able to use knowledge about the noise in a system to filter out its effects and to improve system performance without additional hardware. This led engineers to apply his principles to many different problems. These same principles can be applied to technical analysis of stock and commodity prices.

Because the first use of Dr. Kalman's filters was limited to certain environments, few people took the time to understand the mathematics. A technical "cult" has been built around these equations, for many a Ph.D. has been earned from studying them. At first glance, the original papers presented by Dr. Kalman are an impenetrable fortress to be understood only by the highest of the most high in the world of mathematics. This paper will present, for the common man, the simple second order case of Kalman filtering and how to apply it to data smoothing.

The equations are quite simple but very effective in filtering out the high-frequency noise found in most measurement systems. The analysis of continuous pricing information works well with this form of filtering.