Right more than wrong. They include all sorts of data points. Every once in a while they wildly over or underestimate teams when things like returning production and recruiting out perform what was expected(FSU and BYU). An example there would be Kingsley leaving being a positive not a negative. Thats not going to be captured. Caleb and Jake being 100x better than they were the previous season, thats not a data point. Maybesome close to the program felt like that could be the case here, but not something a model is gonna forecast.
It doesn t mean that FSU fans were wildly optimistic or BYU fans were wildly negative. It also doesnt mean they should all stop writing or using data to forecast seasons. You can see 85% of the teams are very close to where they forecasted. If it were 100 there really wouldnt be a need to play the games