OldPlayer
Member
I don't believe you dismiss data from any period of time. You can weight the data accordingly. Sure the game has changed in 150 years. But it's changed significantly since 2009.
But if you're trying to predict "the best," you need to evaluate data from the time period that makes a team "the best." How many times has the champion been determined during the bowl games? How many times has the champion lost during the early season? If you establish your dependent variable (winning a championship), you evaluate which independent variables (rushing yards, passing yards, turnovers, non-offensive TDs, sacks, etc) help to predict success. Remember that success in all past years has been based on opinion (polls, CFP rankings), not data.