How-to-run-rank-Strategy-on-chosen-DataSets

You can automate that using the StrategyRunner, saving the results of each run to file, repeating the process on a list of DataSets. If you need help with that, we have the Concierge Service for custom programming jobs and consulting.

You can do that right now. My feeling is that the likelihood of implementing a "DataSet Rankings tool" will remain pretty low.

If we were to implement it I could see it as a new tab in the Strategy Rankings tool, it would fit quite well there, but so far the request has no demand (votes.)

QUOTE:
developed a simple strategy and would like to backtest it across multiple datasets.

This approach is statistically flawed. I would lump all your datasets into one big dataset and then do a Symbol-by-Symbol parameter optimization on that combined dataset.

Afterwards, I would perform a Cluster Analysis with your favorite powerful stat package on the parameters the optimizer found (above) and let that analysis group (cluster) stocks with like preferred-value (PV) parameter settings into subset datasets. Now you can trade on those subset datasets employing pooled parameter settings specific to each of the smaller datasets. IMPORTANT: Let the cluster analysis determine which stocks belong into each of the smaller datasets. I would recommend not trying to cluster more than 6 or 7 parameters.

Any researcher that's publishing research papers in either sociology or psychology can help you setup a cluster analysis. Also, first-year graduate students in statistics take a multi-variant course where one of the methods introduced is cluster analysis, so anyone with a Master's degree in stat can help you set this up. An advance stat book in statistics (like that use for first-year graduate students in statistics) will have a chapter on it as well, so it's a well established methodology.

Happy numerical analysis to you.