I prefer hard evidence to go with stats
For example this past season
Princeton went into the tournament with one win and in the semifinal a 2 win Tiger team who had struggled in most weeks went up against a 8 win Team who had blown teams out
Princeton won 35-6
I was in the stands for the exact opposite when a 8-0 Tiger team hosted a below .500 team and lost by 2 scores
here my final answer
Stats when used alone can be faulty
1) You don't get a 100% R squared anything, it's a value between -1 and 1 and anything fully -1 or 1 is going to be messed up data. It's being used here as an indicator of relative correlation of metrics and winning.
2) It's not about picking a winner at all.
you cant have 100 percent r squared in football based on historical stats, just too many factors involved in winning, but you seem to enjoy trying to pick a winner this way and more power to you if it helps you enjoy the game. go colts!!