Conditional Predictions on Stock Prices?

I would like to use analysis services to analyze stock prices.

I want to find conditional probabilities:

P (YpriceChg >= 10% s.t. Ydate between A and B| X Price Chg >= 20%)?

… Like given a price change of X percent or greater, predict the probability of a price change of Y percent or greater, within a specified time window (like 2 days, 3 months etc.).

I also want to add a support filter, like:

N > 30 cases (i.e., there have been at least 10 instances of a 10% or greater price change, for the chosen time window)

I have a database of prices, monthly, daily, etc.I also have a number of cols that compute statistics such as pChg1M, pChg-1M, vChg1d.Like price chg 1 month forward, price change 1 month backward, volumeChg1d forward.Ideally, I would like to minimize the column flags necessary for the experiment.Can you offer some hints, as far as setting up appropriate columns/flags and choosing a algorithm (maybe decision trees, association rules, or NB)?

Regarding algorithms, since you are ubdoubtedly going for accuracy, I would recommend neural net or logistic regression.