The San Francisco 49ers are set to build on last year’s run to the Super Bowl, so says a prediction model that simulates the NFL season 50,000 times.
In the simulations run by PredictionMachine.com, the most common result was the 49ers beating the Denver Broncos after last season’s disappointing Super Bowl loss in New Orleans. In the simulation, the 49ers won the Super Bowl 20.1 percent of the time – more than any other team - and they are most likely to play the Denver Broncos, who are the AFC’s representative 14.6 percent of the time. It probably says something that although the Broncos are considered the clear favorite in the AFC, they're still predicted to make the Super Bowl only about one of every seven times in this model.
The prediction model (the “Predictalator") takes into account all 22 players on the field for both teams during an NFL game, the coaching staffs, weather conditions and even home field. To ensure accuracy, every NFL regular season and postseason is simulated play-by-play 50,000 times.
“Inputs are based on strength-of-schedule-adjusted [for each] team and player statistics considering anything that a team and coach has done in his career plus the expectations that come from age, experience and new systems," Paul Bessire, general manager of PredictionMachine.com told Shutdown Corner. "Injuries are both accounted for now for players that are currently hurt and expected to start the season that way and projected for those with significant injury histories. Each player has a probability of playing in a given simulated game based on his injury history”
It does get rather in-depth, as Bessire explains.
“Schemes and playbooks are set up as close as we can expect from data we have on teams and coaches. Some subjectivity enters the equation when setting depth charts and play distribution but that’s where doing this for 10 years really comes into play,” Bessire said. “If and when something changes due to injury or coaching decision different than my assumption, I can take one guy out, put a new guy in and re-simulate to gauge impact. Simulation is the best possible way to make predictions as it can account for every possible interaction on the field.”
It isn’t a perfect science. Last year the simulation had the Philadelphia Eagles beating the New England Patriots 28-26 in the Super Bowl. The Eagles went 4-12. But the regular season predictions are generally more accurate.
There are twists and turns from the simulation results, which mirrors what usually happens in the NFL.
“Over the last decade, less than half of the teams from a previous year’s playoffs have made the postseason the next year. We are essentially saying the same thing,” Bessire said.
“It is hard for me to separate the fan in me and what I know about football, but I would suggest that it constitutes a ‘surprise’ to the average fan that the St. Louis Rams win more games than the Atlanta Falcons, that the Carolina Panthers are ranked in the top 10 in our ‘Power Rankings’ or that the Miami Dolphins have a better chance to make the playoffs than five teams that made the playoffs last year – the Atlanta Falcons, Cincinnati Bengals, Washington Redskins, Minnesota Vikings and Indianapolis Colts– or that New Orleans Saints and Pittsburgh Steelers struggle to get to .500, while the Detroit Lions, San Diego Chargers, Tampa Bay Buccaneers and Kansas City Chiefs all get to 8-8 on average.”
Here’s a look at how PredictionMachine.com is simulating the postseason for the 2013 NFL season:
Wild Card Round
No. 6 Pittsburgh 20 at No. 3 Houston 23
No. 5 Baltimore 23 at No. 4 Cincinnati 24
No. 3 Houston 27 at No. 2 New England 31
No. 4 Cincinnati 21 at No. 1 Denver 26
No. 2 New England 28 at No. 1 Denver 30
Wild Card Round
No. 6 St. Louis 21 at No. 3 New York Giants 23
No. 5 Seattle 27 at No. 4 Atlanta 26
No. 5 Seattle 23 at No. 1 San Francisco 28
No. 3 New York Giants 24 at No. 2 Green Bay 31
No. 2 Green Bay 27 at No. 1 San Francisco 32
San Francisco 31, Denver 27
- - -
Kristian R. Dyer contributes to Yahoo! Sports and covers the Jets for Metro New York and is on Twitter @KristianRDyer