Miami vs. Ole Miss Fiesta Bowl game prediction by ESPN football computer

An expert prediction for Miami vs. Ole Miss in the College Football Playoff Fiesta Bowl semifinal game from a model that simulates games and picks winners.

Jeff Romance-Imagn Images

And then there were four, as the College Football Playoff is down to the semifinal round with Ole Miss and Miami squaring off in the Fiesta Bowl this week for the right to play for the national championship.

Miami edged out Texas A&M and got past defending national champion Ohio State in the quarterfinal to earn a place in this game behind a tough defensive front and Carson Beck at quarterback.

Ole Miss dominated Tulane in the first round and then slid past SEC champion Georgia in a barn-burner Sugar Bowl quarterfinal to get here, all without Lane Kiffin.

Looking ahead to this week’s matchup, let’s check out the latest college football predictions from the Football Power Index computer prediction model.

The model simulates every NCAA college football game 20,000 times and uses key analytics from both teams and picks winners based on a projected scoring margin per game.

Ole Miss vs. Miami prediction

The model expects the Ole Miss run to come to an end in this game, but forecasts a matchup that could come down to the last play.

Miami emerged as the projected winner in the slight majority 57.1 percent of computer simulations of the matchup.

That leaves Ole Miss as the presumptive winner in the remaining 42.9 percent of sims.

How does that translate to an expected margin of victory in the game?

Miami is projected to be just 1.3 points better than Ole Miss on the same field in both teams' current composition, according to the model's latest forecast.

College Football Power Index

Football Power Index (FPI) college football rankings and computer prediction model are a measure of team strength that predicts a team’s future performance. 

Rankings and scores predictions are based on 20,000 simulations of a team’s season and games, using a combination of key analytics, including scores to date, quality of opponents, team talent, recruiting, and a team’s schedule.

Teams are ranked not in order of talent like in other rankings, but by a projected point margin per game against an average team on a neutral field.

How good is the prediction model?

How accurate was the College Football Power Index computer prediction model last Saturday?

Projecting the games a week ago, the Power Index models correctly predicted 74 percent of all games and hit 48 percent against the spread.

Predicting a total of 799 college football games a year ago, the Power Index computers were correct for 70.964 percent of their final picks, ranking eighth nationally out of 55 other football models.

Over the last decade, the Football Power Index has proven correct on 75 percent of FBS college football game predictions, including in 73 percent of matchups when it favored a team with at least 70 percent likelihood to win.