Quote Originally Posted by Flying Mullet View Post
Go for it.
Think I found a way around it. Before I was recalculating an entire correlation table for an entire period, bit by bit (so I'd do it with 120 points, then 121 etc etc), but I realised I only actually need the last set of data for the model to select itself. As that loop was within another loop, that's where it was slowing down, so removing it will speed everything up.

On the other note, do you have experience in Java or C# or one of the more modern languages? Think it wouldn't harm me to learn, but obviously it needs to know what the reward is like compared to the effort.