My grasp of statistics is weak, but I am missing the connection to overfitting. I always thought overfitting to mean something like interpreting a series of coin flips producing HHTHT to mean the coin was fixed to produce HHTHT. Two characteristics of overfitting, then, are that you can only overfit the same theory to one piece of data, and that getting more data will ruin the overfitting.

But here, you can apply the "sun god" concept to practically any myth you want (if you're a good enough arguer). And I imagine that if someone discovered new information about Hercules, Max Muller would be able to think up a reason why it supported the sun god theory.