Backtesting: pattern finding, any ideas?

 

Hi all, hoping for some input on this from your experience or viewpoint.

After a getting results from a parameter sweep, how do you analyse the data to find 'patches' of good results. I mean, if you have a good set, but change one parameter slightly and suddenly the results are bad, I think it's likely to be a chance occurrence, not really a good set of parameters you can trust. With two parameters its easy to see with a chart, but how can you go about it with, say.. 7 dimensions?

Currently I just have the results listed in Excel (i.e. 1 dimension) and scroll through, noting parameter ranges for good results, but increasing the number of tests will make this process longer. And also, if there are two or more 'good patches', it will confuse the results.

Any ideas?