Projecting Favre's impact

Posted by ESPN.com's Kevin Seifert

Brett Favre's televised interview Monday night brought some on-the-record clarity to his courtship with Minnesota: If his arm heals from surgery as expected, Favre wants to sign with the Vikings later this summer. And based on some of his comments -- including, "We should be pretty good" -- I think it's fair to say he believes it will happen.

So it's not too soon to start asking some competitive questions. How would Favre's return affect the NFC North? To what extent, if any, will he elevate the Vikings? And how would they fare if Favre has a setback and remains retired?

There are many ways to address those questions, from both qualitative and quantitative standpoints. Subjectively, many media observers have chosen Chicago as the trendy offseason pick to leapfrog Minnesota and win the division after its acquisition of quarterback Jay Cutler. Intuitively, the addition of Favre would seem to return the advantage to Minnesota.

The quantitative answer is a bit more complex. ESPN Research recently commissioned a statistical simulation of the NFC North from AccuScore, a company that -- in the interest of full disclosure -- uses computer-generated results to predict NFL games for ESPN.com and also provides advice for gamblers and fantasy football players.

I'll explain more about their process, one that I think provides some interesting insight but ultimately should be regarded for what it is: a computer simulation. First, however, let's look at the results. If Favre remains retired, AccuScore essentially projects the division to end in a three-way tie between Green Bay, Minnesota and Chicago. Below are their projected standings, based on the average results of 10,000 simulated seasons:

With Favre as a 16-game starter, AccuScore projects the Vikings improving by one victory and winning the division with a 10-6 record. A look at the Favre-adjusted projections:

I spoke this week with Stephen Oh, who programmed this simulation for AccuScore. He explained the process in detail, which I'll do my best to boil down in simpler terms.

AccuScore develops about 100 variables to create profiles for every NFL coach and player. Examples of coaching variables include a coach's play-calling tendencies in certain situations: first-and-10, third-and-long and in the red zone. For a player, AccuScore develops a profile based on his historical performance in various situations: grass fields, turf fields, division games, against Cover 2 defenses and the like.

The digital profiles of each player and coach are combined to form a digital team, and a computer program then plays out a season based on the team's actual schedule.

"Kind of like Madden without the graphics," Oh said. The process is repeated 10,000 times for each team to allow variables to play out randomly. The NFC North standings above are the averaged results of the 10,000 "seasons.

This is by no means a perfect system. Using the same method last year, AccuScore correctly predicted five of the eight division winners. They were wrong on the AFC South, where they had Indianapolis as the champion, the NFC East (Dallas) and AFC East (New England). It's worth noting all three teams finished 2008 with winning records.

So replacing Tarvaris Jackson and Sage Rosenfels with Favre, the simulation found the Vikings would improve just enough to win the division. Interestingly, that extra victory would come from outside the division, if you consider that neither Chicago nor Green Bay's record is affected significantly in the Favre simulation.

But for me, the most interesting part of this is trying to pinpoint how Favre would make the Vikings better from a statistical perspective. We've spent some time discussing Favre's struggles in the latter half of recent seasons. So exactly where would his impact lie?

First, let's take a look at how the computer compared a 16-game season with Favre to a full season with Jackson and Rosenfels splitting the snaps:

When you look at the numbers on the surface, Favre's performance isn't substantially better than the Jackson-Rosenfels combination. But upon further inspection, Oh said, you find that Rosenfels is projected to throw 11 of the duo's 18 interceptions. Also, Jackson dragged down their combined completion percentage by connecting on only 56 percent of his passes for 2,650 yards over a full season.

Based on each quarterbacks' history and factoring in the current offense, the simulation finds that Favre would throw fewer interceptions than Rosenfels and elevate the Vikings' downfield passing efficiency relative to Jackson. Would he produce Hall of Fame numbers? Probably not, and I think we could all agree about that even without a computer's help. But he would, as the saying goes, be better than the alternative in some key areas.

Again, I think you should take these simulations for what they are: Projections based on a thorough set of historic markers. I think they verify for us that Favre would not be a cure-all for the Vikings nor a dominant player in the NFC North this season. But in a division that is otherwise hotly competitive, he could be the difference.