Finding the True MVP with Points Added

Carson Wentz put on a show on Monday Night Football against the Washington Redskins and the talk has turned to whether he is a legitimate MVP candidates in only his second season.

It’s fair to say that Wentz has made marked improvement this season and that he’s been very valuable to his team, but how to we compare him to other quarterbacks and even the rest of the offensive skill positions?

Earlier this season I introduced points added for quarterbacks. My points added calculation is based on the expected points added work done by Brian Burke at Advanced Football Analytics and then with ESPN’s Total Quarterback Rating (QBR).

The quick summary of points added is that it attempts to allocate credit to offensive skill positions for the amount of points their teams have scored versus what you would expect from an average offense in the same game situations, i.e. down, distance, score differential and time remaining.

The methodology

The first component to the points added calculation is the play-by-play expected points calculations. This means you determine how many points a team should expect to score, or give up, at their current game situation and then determine the points added by taking the different between that number and the same calculation before the next play. Whatever happened on that play can now be measured in the value of points added, and then we can parcel out credit to the various players involved.

To determine how to parcel out that credit, I used principals from Pro Football Reference’s Approximate Value calculation, using NFL salary cap allocations to determine how to allocate EPA among different positions.

Unlike ESPN’s QBR calculation, I don’t have the detailed play information to adjust credit based on pressures, drops, etc. I’m only using average assumptions and I’m not adjusting for strength of defense, which has only had a minor effect on QBR.


Quarterbacks account for roughly 30% of all offensive cap spending in 2017. If you assume quarterback value is roughly 90% in the passing game and 10% rushing, these are the credit allocations for different types of offensive plays:

Air yards: 50%

Yards after catch (YAC): 20%

Interceptions: 50%

Pass Interference: 35%

Sacks: 30%

QB Runs: 60%

Fumbles: 30%

The reason these numbers are so low is that the credit for offensive passing plays must also be split with offensive lineman (30% of offensive cap spending) and the other offensive skill positions (40%).

The credit for air yards is relatively high, assuming that the receiver is more responsible for creating yards after the catch. The credit for points added through defensive pass interference is lower than regular air yards, assuming the receiver is more important in drawing that penalty than on regular throws. Interception credit is only 50% to account for the influence of poor blocking and dropped passes on interceptions. Lastly, sacks and fumbles, which are most on sack for quarterbacks, are at 30% to account for the influence of poor blocking, but also to recognize that quarterbacks and their time to throw have a big influence on sacks taken.

And here are the quarterback results for 2017 through Week 7. You can scroll left to see the individual components of the calculation. I also included adjusted net yards per attempt (ANY/A), the pure passing stat most correlated with winning, for comparison.

[table id=88 /]

Non-QB skill positions

In a similar way as with quarterbacks, I used the salary cap allocations for various position as a baseline for calculating these numbers.

Air yards: 20%

Yards after catch (YAC): 50%

Interceptions: 20%

Pass Interference: 35%

Runs: 40%

Fumbles: 50%

The allocations for non-quarterbacks in the passing game is the flip-side of quarterbacks: less credit for air yards and more for yards after the catch. The interception credit is much lower, as interceptions when targeted can be blamed on the receiver for drops or not getting open, but it’s still mostly on the quarterbacks. The pass interference credit is the same, and the allocation for runs is lower, mostly because the low salaries of running backs reflects the relative lack of importance of the runner versus his blockers in the offensive line who are much more highly compensated. Fumbles is also higher since it’s more on the ball carrier when they fumble versus a quarterback sacked in the pocket.

And here are the non-quarterback results for 2017 through Week 7. You can scroll left to see the individual components of the calculation.

[table id=89 /]

The true MVP candidates

This is where we bring it all together. Below are the top-25 players by points added per game, which includes all offensive skill positions.

The “Wins Added” columns is calculated using the regression estimate for wins based on point differential. The regression found that roughly 31 points equals one win. The best way to think of wins added is how many more wins than a .500 record an average team would have with this player.

[table id=90 /]

We all know that quarterback is the most important position in football, and the points added numbers back that up with the top four spots occupied by signal-callers. Surprisingly, Chris Thompson is has the highest points added per game of any non-quarterback, reflecting his uncanny, and potentially unsustainable ability to break long catches. All of Thompson’s points added were from receiving, and running backs typically gather relatively more yards after than catch, which is weighted more heavily in the formula.

After Thompson, there is only one other running back in the top-25, the popular MVP candidate Kareem Hunt. Like Thompson, receiving drives most of Hunt’s value. But you’ll see above in the detailed table Hunt leads the NFL is points added on the ground. Although you’ll see how much less efficient running the ball is by the fact that it took Hunt 123 rushing attempts to generate 5.4 points, but had 8.6 points added on only 29 targets.

Consistency for the top candidates

All the top-4 quarterbacks have been quite consistent this year, with some variation in the results for Deshaun Watson and a lot of steady points added for Tom Brady.

As of this moment, it looks like Brady is the leading MVP candidate, especially considering his long track record and what it says about his ability to maintain this level of performance. It’s possible that Carson WentzAlex Smith or Watson could continue to play lights-out and overtake the old man, but I wouldn’t bet on it.


Leave a Reply