Trifeature Synergy at Chance Level

Hi, 

first of all, thank you for publishing your code.

I am having trouble trying to run your trifeature experiment notebook, the synergy test reports only exactly 50.00% accuracy every time, instead of the expected 70%, and I am unsure why. 

I would be grateful if you could verify if it's a problem on my side, or provide any hints on what might be going wrong. 

