Brute Force Classifier in Action

As an update to my recent post, here is a short video of the brute force similarity search classifier in action.

Non-embedded view here.
The coloured coded candlestick bars are coloured thus: purple = a cyclic market classification; green = up with retracement; blue = up with no retracement; yellow = down with retracement; red = down with no retracement. The upper price series is the classification as per the brute force algorithm and the lower is the classification as per my Naive Bayesian classifier, shown for comparative purposes.  The cyan trend line is my implementation of a Kalman filter, and where this trend line extends out at the hard right hand edge of the chart it changes to become the prediction of the Kalman filter for the next 10 bars, this prediction based on extending the pattern that was selected during the run of the brute force algorithm.

I will leave it up to readers to judge for themselves the efficacy of this new indicator, but I think it shows some promise, and I have some ideas about how it can be improved. This, however, is work for the future. For now I intend to crack on with working on my neural net classification algorithm.