But if you want to analyze swarm effects, a group average is highly misleading. First, you should subgroup for different behavioral strategies and then analyze the clusters in isolation.

In FA we trust, PCA, our Lord

It can identify the moment when coin becomes shity...