[sireat]

Those findings while interesting need more work:

The clusters were from back-tested stock data. Ie we knew ahead of the time that there were profits to be made. Was this akin to p-hacking? That is where there more stocks selected with some sort of correlation without the profit potential.

The article alludes that there were more stocks without the raise in stock price after Celgene acquisition.

So the real 'show me the money', 'skin in the game' for this kind of research would be to make predictions on a real time event(merger, acquisition etc).