There are too many one game upsets regardless of which system anyone uses to say “This system isn’t any good.” We’ve seen a #18 Notre Dame beat #3 Florida, an unranked Stanford beat a #1 Notre Dame, and on and on and so on for examples of upsets using human polls. One game outcomes don’t invalidate a predictive model, which F+ is. Now, could Ole Miss replicate this result? Yes, it could. No system is going to predict every outcome of every game as there are hundreds/thousands of factors at play, but when looking at the totality of play of every non-garbage time offensive and defensive snap and taking into account the totality of opponents’ snaps and these opponents’ snaps, the system says team X has performed better than team Y but that does t mean team Y cannot beat team X in a one game scenario and/or that the team matches up better vs. that team than it does with others.
F+ had full strength PSU as an 8 point favorite over full strength Ole Miss—Vegas had PSU as a 5.5 favorite over Ole Miss, too, knowing that Robinson, a top 20 NFL pick, a player playing an elite difference making position, i.e., edge rusher and that a top 5-7 NFL pick in OT Ola Fashanu weren’t going to play. Does Vegas also not know what it’s doing? I have not seen a team with a higher F+ be an underdog vs. a team with a lower F+ on a neutral field unless those F+ numbers are almost exactly the same.
F+ had Texas as a 4 point favorite over Washington—Vegas had Texas as a 3.5 point favorite.
F+ had Michigan as an 8.5 favorite over Washington—Vegas started Michigan as a 3.5 point favorite and then all the money came in on Michigan moving the line to 5.5 at kickoff.