The subject lines generated for your campaigns are designed to maximize the amount of opens, converting recipients into readers. The split tests are run to provide statistical significance on opens so you can be certain the winning subject line will ultimately deliver the best return from your final send.
​
To avoid picking a winner at random when selecting by clicks, the split sizes would have to be so much greater. As a result, there would only be a tiny minority of your audience left for the final send. This means you wouldn't see the meaningful incremental effect as a result of testing in the first place.

Read more about why you should select the winner based on opens here.