I've read about a similar scenario (possibly the same one?), and you got the important bits right. You cannot normalize the entire dataset at once; the normali

I've read about a similar scenario (possibly the same one?), and you got the important bits right. You cannot normalize the entire dataset at once; the normalization contains some parity effectively allowing the algorithm to cheat and see the future... but in real life we can't see the future. The simple example that eliminates some of the "black magic" is that normalizing against the entire set lets the algorithm know what the highest and lowest points across the entire data set are - and knowing the lowest and highest lets the algorithm buy low/sell high for all of the known data.