Some of the variants are more useful but just the kalman itself is only as good as using a weighted moving average because there are no continuosly, stable coefficients. Unscented Kalman is the route you'd want to go down, but if you're trying to keep up to date you're more likely looking for an usncented particle filter. The caveats to any of those methods are if you can produce signal lines that turn arround very quickly... choppy in certain areas and highly profitable in between the choppier trades... Hmmm... Although, even then, while you're doing that someone else is doing this http://www.google.com/search?q=part...=&aql=f&oq=&gs_rfai=&pbx=1&fp=1c030b17bae363c So it might be more important to realize what the simple definiton of a quant is...Someone who applies sciences used in other fields to analyze financial market data for alpha discovery. In other words forget the idea of asking if it would be useful and start understanding why it would be useful to what you're doing.

The Kalman filtering technique rapidly developed in recent decades. It's widely used in many areas such as Aerospace, Earthquake monitoring, Economic trending Control and Inertial navigation. Not like other filters, the Kalman filtering is suitable for multi-input and multi-output system. It provides the most optimal filtering and estimation based on mathematical model describing the system. A classic example of Kalman filtering application is to predict the position and velocity of the object from a set of limited observation sequences containing the noise. It can be found in many engineering applications such as Radar, Computer vision. Meanwhile, it is an important topic in control theory and control system engineering. For example, people are interested in tracking targets in radar, but the measurements of the target position, velocity and acceleration contain noise at all times. The Kalman filter removes the noise and gets a good estimate of the target location by the dynamic target informations. This estimates maybe the current target position estimates(filtering), as well as the estimates of the future(projections). It can also be estimated location of the past (interpolation or smoothing). People often confused with the complex formulas of Kalman filtering. They are eager to find a simple way to achieve the operation. Visual Kalman Filter is a nice tool for training and simulation. It meets the needs of many beginners. Visual Kalman Filter is developed for science researchers based on visual windows interface. It helps people to deal with the dynamic data, and draw predictions and graphics. Users need not to write any code(Such as Matlab, C++, etc.). Users should build the system model first, and get the matrice of the model. That is, you'd better input the matrice A,B,P0,Z,Q,R,etc. in the base equations as follow: X(k) = AX(k-1) + BU(k) + W(k) Z(k) = HX(k) + V(k) In the two formulas, X(k) is the system state, U(k) is the amount of system control. Z(k) is the measured value, and W(k) and V(k) represent process and measurement noise, which are assumed as Gaussian white noise. After inputting the system matrix parameters, click 'step 3', users will get the results such as X(k|k-1),P(k|k-1),X(k|k) and P(k|k). When users click the strings in the listbox, the results and curve will appear. Then just save the results. Visual Kalman Filter is for training and learning as well as analysis of the data. It's not a real-time tool to track the state. There is a dll file for trial, please contact the developer. So, it is easy to operate for the users. Only three steps you need to do. Have a try, maybe it's helpful for you.