Which One?

A friend of mine who is a Vikings fan is concerned about rumors that the Vikings might replace their current QBs with Kirk Cousins, now that he seems available. His position is that while Cousins is a fine quarterback, the cost to obtain him will be greater than any difference between him and the QBs currently on the roster.

The discussion requires more than what I can do with Twitter, even with 280 characters, so I’m going to do a blog post.

The QBs in question are:
Kirk Cousins (29 years old, 57 games started)
Teddy Bridgewater (25 yo, 28 gs)
Sam Bradford (30 yo, 80 gs)
Case Keenum (30 yo, 38 gs)

Teddy Bridgewater is the youngest, and the rest are all essentially the same age. Bridgewater is also the one with the least number of games started, but we still have about two full years of data to work with.

The single most important stat when looking at QB passing is ANY/A. This is Average Net Yards per Attempt. Here is the formula: (pass yards + 20*(pass TD) – 45*(interceptions thrown) – sack yards)/(passing attempts + sacks). Since it is per attempt, we can get an idea of efficiency per play, which will reward skill as opposed to opportunity.  I’m not going to use this raw stat, but instead use the indexed version called ANY/A+ where 100 is average for a given year, so an ANY/A+ of over 100 is better than average, and less than 100 is worse than average.

I’ll also use the indexed version of some other stats. If you see a “+” in a stat, then again 100 is average for a year. Indexing makes it much more clear how a player is doing in a given year.

I’m also going to refer to AV, which is a stat the Pro Football Reference came up with to get a general approximation of player value. You can find more about its use and limitation here: https://www.pro-football-reference.com/blog/index37a8.html. It’s not a perfect stat, but it helps with general things like games played, durability, all-pro years, and that sort of thing. I’ll divide AV by games started to get an idea of what the game average is.

The above focus on passing only, so when I break things down later, I’m also going to refer to QBR, ESPN’s proprietary stat that includes running ability, and the raw rushing and fumble stats. Quarterback rating and ANY/A do not include their ability to rush, and for some quarterbacks, that’s a big part of their contribution.

So let’s see how they stack up on three  main stats. One note, all of these are their career stats, in order to get the largest possible sample size. I’ll discuss individual years as needed.

Anyway, here are the QBs:
Cousins (ANY/A+ 109, Rating+ 107,  AV/gs 0.75)
Bridgewater (ANY/A+ 91, Rating+ 97,  AV/gs 0.79)
Bradford (ANY/A+ 93, Rating+ 97,  AV/gs 0.60)
Keenum (ANY/A+ 99, Rating+ 97,  AV/gs 0.63)

One note, I’m fudging a bit on Bridgewater and Keenum’s ANY/A+ and Rating+, since they haven’t played enough games for Pro Football Reference to have totaled their career results. I calculated the scores as best I could, but I might be off by a bit. Keenum, in particular, I might be shortchanging. Still, these are close enough to work with.

As you can see, the only place where Cousins doesn’t dominate the others is Bridgewater’s AV/gs. In all the rest of the categories, each of the others is below average, while Cousins is well above average.

Let’s glance at the subsidiary stats to see if we can find why Bridgewater’s AV/gs is so high and see if there’s a way that the other QBs can match Cousins. Here we’ll see yards per carry, yards per game, and fumbles per game. Since there are much fewer of these than pass attempts, I’ll toss out a few years where the QB didn’t play much and these numbers obscure, rather than show, the actual talent in question.

Cousins (3.0 ypc, 6.7 ypg, 0.65 f/g)
Bridgewater (4.4 ypc, 14.3 ypg, 0.39 f/g)
Bradford (2.4 ypc, 4.3 ypg, 0.59 f/g)
Keenum (3.4 ypc, 7.9 ypg, 0.42 f/g)

Here we can see that if we take rushing into account Bridgewater is clearly the best of them all. He rushes for more yards per carry and per game and has fewer fumbles than them all. Bradford is awful at running, while Cousins and Keenum are mediocre. Cousins main problem, though, is his fumbles per game, worst of the four.

However, let’s look at QBR, which attempts to quantify all of the above to see if that problem puts one of the other QBs over the top. We can’t see a true average here, so I’m going to list the QBR for each QB the years where they played more than 10 games.

Cousins (71.7, 66.1, 50.5)
Bridgewater (54.4, 57.5)
Bradford (47.0, 30.5, 51.6, 42.1, 57.3)
Keenum (37.1, 71.3)

As you can see, Cousins comes off well here again. Keenum had one very good year this past year, but that might be his career year. Certainly, it represents a major step up from previous years, where he has had some decent opportunities. Bradford has these mediocre years, but was off to a great start in 2017, having a 72.4 QBR in those first 2 games. However, that’s a really small sample size and I would not lend it much credence.

It’s also clear that last year was the worst one as a starter for Cousins. His completion percentage was down, interceptions up, fumbles up, and his ANY/A was the worst of his starter years by quite a bit. Also, in 2017 his sack percentage almost doubled from 2016, leading me to think he was running for his life, thereby hurting his production. Thus, 2017 seems like it’s an aberration, but even with that he was about where the others were, except for Keenum, in QBR, and his career rating and ANY/A is much better than the rest.

But then, what about Keenum? Is that 71.3 reflective of him finally getting a full shot or just a career year. I would guess it’s a career year, as everything screams fluke year. His completion percentage hovered around 60% in all of his previous years, but he had a 67.6% completion percentage in 2017. His TD rate was around 2.9 previous, but was 4.5 in 2017. He threw far fewer INTs per pass. In short, in every category he was markedly better than his previous level of play.

Given their similar age (Cousins being younger by about half a year), and their prior track record, I will bet good money that Cousins would significantly outperform Keenum, given the same OLine and receiver set.

Interestingly, all of these QBs are unrestricted free agents. The Vikings might have a case that Bridgewater is under contract for another year, but it’s iffy and they don’t plan to contest it, thus Bridgewater will be free.

That means all of these will command big salaries if they are signed to be the starter. To discuss them, I will assume five year contracts to be consistent with Jimmy Garappolo’s recent contract that sets the bar.

Keenum will probably be the least expensive, but if signed to be a starter, I would expect a contract in the upper teens. Same for Bridgewater. Bradford is unlikely to sign with anyone for less than $20 million per year, if signed to be a starter, and he will expect that.

Cousins will be the most effective, and I bet he’ll be something like five years, $145 million, with something like $80 million guaranteed. That’s more money than Garappolo, but less guaranteed, which I think is likely given the age difference. I could be wrong, though, and Cousins might get over $100 million guaranteed.

Now, given this research here: http://socalledfantasyexperts.com/aging-curve-nfl-offensive-players-every-single-position, we can guess how the QBs will play out a five year contract.

Assuming Keenum’s only good year is valid, he would looking something like this, based on a normal AV aging curve: 13 AV, 12, 12, 10, and 9, or 56 AV total over the life of a 5 year contract. That’s his ceiling.

Bradford, based on 2016’s AV (last full season), would look something like 10, 9, 9, 7, 6, which is clearly much worse than Keenum. How Bradford keeps getting paid is beyond me. He’s a bad QB and will continue to be so.

Bridgewater is the only one of the QBs on the upslope of his aging curve. If he is healthy, a big if, and if he returns to his 2016 form at age 23, then his AV curve could look like:  15, 16, 16, 15, 15, or a total of 77. That’s probably an ideal scenario for him, though, given the severity of his injury. We don’t actually know if he’s recovered.

I’m going to also choose Cousins’ 2016 year, which matches the curve better and reflects more of his QBR base than last year where his OLine let him down. His AV curve would then go something like 15, 14, 14, 13, 12, giving a total of 68. This seems a likely scenario for me, and those are very good AVs.

In the end, I would still choose Cousins over Bridgewater because Bridgewater’s primary argument is based upon him returning fully healthy and then immediately going back to where he was at 23, thereby fulfilling his aging curve. Cousins, on the other hand, bases his argument on being a much better passer over a larger sample size.

Basically, I don’t think Bridgewater fulfills that aging curve projection. One, I took the best possible scenario. Two, I don’t think he’s healthy. He certainly didn’t look it last year. Also, the Vikings have to know his medical situation better than anyone and they are not trying to keep him for one more year of his rookie contract.

This, to me, is huge. Rookie contracts in the NFL are great for teams. Like I said, a starting QB in the NFL starts at around $15 million per year, whereas the final year of a rookie contract is under $1 million. If Bridgewater was healthy, the Vikings would work real hard to keep that huge hit off their salary cap.

Now, if you know, absolutely know, that Bridgewater can come back and still improve based on a normal aging curve, then you’ll want to give Bridgewater the 5 year contract. The Vikings, however, have chosen to go away from him, meaning of the rest, Cousins is the clear winner to me.

So, there you go. The answer was closer than I expected, and I didn’t actually expect Bridgewater to be the primary competition for Cousins, but age is huge in the NFL. Still, I feel confident that Cousins would be the best of these four options for the Vikings.

Which means as a Cowboys fan, I’m rooting for Cousins to go to the Broncos.

2 thoughts on “Which One?”

  1. Totally with you on the Vikings have better on Bridgewater. So to put things into perspective what were the ANY/A+ of other playoff QB’s this year (you can avoid the whole Phil QB situation ). If they were all in the 150s+ than the 93 to 109 difference isn’t all that spectacular, but Brady is only 120, its huge! Also is the overall QB rating controlled for wins or losses in any way or adjusted for strength of the opposition?

    1. Last year Tom Brady’s ANY/A+ was 120, as was his Rating+. Some of the other top ones last year were Alex Smith (121 / 123), Drew Brees (122 / 121), Carson Wentz (118 / 119), Jared Goff (122 / 117), Matthew Stafford (113 / 115), Aaron Rodgers (105 / 112), Philip Rivers (120 / 109), and Ben Roethlisberger (112 / 107).

      Now, I used career totals above. To put Cousins in perspective, here are his 2015-2017 yearly numbers: (114 / 118), (119 / 112), and (104 / 108). In other words, that 109 career ANY/A+ is really really good. Brady’s is less, actually.

      Again, I don’t think they have better with Bridgewater. I think on a 5-year contract, given normal aging curves, and if he’s healthy, he’d slightly outperform Cousins. 1.8 AV per year over the life of that contract is not much.

      However, I think we can assume he is not healthy, given that from all I’ve seen they really like him as a teammate and yet are still allowing him to be a free agent. There’s no way they release a QB on a rookie contract that they really like and think he can play.

      As to the other, no, they did not normalize for W/L or strength of schedule. However, the way they compute the baseline is by taking three years (or two if it’s the most recent year). Given the fluctuation of SOS from year to year, that will come close to controlling.

      You can find their process at: https://www.pro-football-reference.com/about/glossary.htm.

Leave a Reply

Your email address will not be published. Required fields are marked *