Hi Pablo!
I think jchapman has a good take when comparing results. To me, wrestling is much more binary win/loss, and that is probably the best indicator.
Some wrestlers can win a LOT of matches 4-1, whereas others have much more offence. When trying to apply an algorithm based on score, you would think that those who shoot/score a lot would be better (wins by Major/TF more), but in reality/head to head, that one takedown wrestler who is almost impossible to score on may be favored. When trying to predict a win/loss head to head, scoring margins like that aren't going to be the case. If I had to guess, the most predictive model will weigh the win/loss more heavily in wrestling than in volleyball.
I would be very interested to see what is more accurate in the tournament, this Pablo model, an ELO model, or rankings like Intermat, and that shouldn't bee TOO difficult of a comparison to make, who is ranked higher on what model, and was the prediction correct. The only problem there will likely be a small sample size (models will likely favor the same wrestler in most cases).
Good luck! and finally, you probably have, but have you looked at wrestlestat? they've got a very similar model going: https://www.wrestlestat.com/faq#:~:text=The algorithm factors in the,that have fewer total matches.