Quantitative Candlestick Pattern Recognition (HMM, Baum Welch, and all that)

Candlestick: measure of high, low, open and close rates in a commodity market.

Uses k-means clustering (probably a vector codebook) make six prototypes of individual candlestick shapes.

Assigns an index to each prototype candlestick.

Compute a matrix of how often one state follows another state.

Use the state indices for Baum-Welch predictors etc