A Mathematical Look at the Raiders’ Current Quarterback Situation

Here we are in mid-March and the Las Vegas Raiders’ number four has managed to live through another hectic free agency. Of course, he could get traded tomorrow and this whole discussion would be for not. However, for the time being, let’s take a look at where the Silver and Black stand at signal caller.

For this article, we are going to do something a little different. I’ve put together a couple of regressions based upon the 2019 season. A regression is a statistical process that estimates the relationships between a dependent variable (the output) and one or more independent variable (the inputs).

Essentially, you can build a mathematical equation to predict the estimated outcome of one statistic based on one or more other. I did this through the computer software R.

The Basic Regressions

The Process

I complied the passing touchdowns, interceptions, completion percentage, passing yards, and wins for the Top 32 quarterbacks last season. Next, I extrapolated all the stats besides completion percentage for quarterbacks that didn’t play all 16 games. For example, Deshaun Watson only played in 15 games, so all his stats, besides completion percentage, were multiplied by (16/15) to get his stats to that of a player who was on the field in 16 games. After this, the data was ready to be run through the regressions.

Expected Passing TDs based on Passing Yards

In this regression, the amount of 2019 passing yards is used to come up with the expected passing touchdowns. This tells us how many touchdowns a signal-caller should throw based on his amount of yards. Based on Derek Carr’s 2019 Passing Yards, he was expected to throw 25.8 touchdowns. He actually threw 21 TDs in 2019. This means two things. First, Carr didn’t throw as many TDs in the red zone as the league average would expect. This makes sense as finishing drives was a problem for the Raiders in 2019. Marcus Mariota should help in some redzone packages (more Mariota stuff later). Alternatively, this means that if Carr has the same type of season in terms of yardage next year, he will almost certainly throw more touchdowns than he did this year.

Expected INTs based on Passing TDs and Passing Yards

In this regression, the amount of Passing touchdowns and passing yards is used to come up with Expected interceptions. For Carr, this number came out to 13.1. He actually threw 10, which is glowing for Carr. This means that he took care of the ball much better than league average. Critics will say this is because he is Captain Checkdown. I’m not here to debate that point, I’m just looking at the numbers.

Expected Wins based on Passing TDS, INTs, Completion %, and Passing Yards

This regression was the granddaddy of them all for my basic statistic regressions. This tells us the Expected Wins based on 2019 Passing touchdowns, interceptions, completion percentage, and passing yards. Before I reveal the results, I want to give you a couple of interesting tidbits about how the equation came out. Passing yards had a tiny negative coefficient. This means that the model believes that the more passing yards a quarterback throws, the less wins he is expected to earn in a season. Signal-callers throw for meaningless yards when behind in games, good for the box score, bad for the scoreboard. Also, touchdowns have a bigger positive coefficient than interceptions, which have a negative coefficient. So, a 2 TD-2 INT game is better than a 1 TD-1 INT one. Interesting stuff as the regression predicted Carr to win 7.9 games in 2019. The Raiders actually won 7.

Related: Maliek Collins Will Elevate Raiders Pass Rush

So Carr played slightly better than the actual number of wins. I believe this to be impressive for him because of the type of wide receiving corps he had, quality of the Raiders defense, and schedule they played. However, the increased number of victories wasn’t insane, so it isn’t like the team was completely dragging him down. A note for the Mariota over Carr people. Using the formerly Titans quarterback’s extrapolated 2018 stats, he was expected to win 6.1 games. Take that for what is is.

The Advanced Regressions

The Process

Truthfully, the results of the above regressions left me wanting a better model. After all, there is more to quarterback stats than just passing yards, ask Lamar Jackson and Josh Allen. Also, fumbles are a factor in signal-caller’s play. Plus, completion percentage isn’t a great stat. The difference between a good completion rate and a bad one isn’t much in the NFL. So, I made a couple of adjustments for this go around. This time I took the top 30 quarterbacks in the league, these are the ones that ESPN gave a QBR for the season. This excluded Mason Rudolph and Joe Flacco, who were included in the previous model, making it more competitive. Frankly, both those guys played horribly last year and it effected the equations. I first compiled the number of interceptions plus the difference between fumbles and fumble recoveries. I didn’t just do fumbles lost because a quarterback losing a ball and an lineman diving on it has nothing to do with him. It is completely luck and the signal-caller should not be rewarded for getting bailed out by another player. I then complied rushing + passing touchdowns, total yards, and QBR for each quarterback. Again, I extrapolated the stats besides QBR.

Expected Wins based on Rush+Pass TDs, Total Yards, and INTs+(F-FR)

I jumped straight to wins, that’s the most fun to look at. Also, victories are certainly a quarterback stat to a certain degree. This regression reveals the Expected Wins based on 2019 rushing + passing touchdowns, total yards, and interceptions + the difference in fumbles and fumble recoveries. For Carr….*DRUMROLL* 8.9 wins. This was unexpectedly high. The main reason was the turnover differential. Carr only fumbled two more times than the ones he recovered. Around the league, that difference was much greater. Again, I plucked in 2018 Mariota, he came out with 6.6 wins.

Expected Wins based on QBR

This regression gives the Expected wins based on 2019 QBR. This was Carr’s best Expected Win total with 9.1 wins. Not too surprising considering, he was 10th in QBR. 2018 Mariota was 8.2 expected wins. Carr critics don’t go crazy yet though. Mariota had a better QBR in 2018 than Carr did the same year. Something to keep in mind.

*NOTE

I couldn’t run QBR with the above regression like I originally intended to. This is because QBR factors in the other stats in some way. Double counting the data is bad. The result was a crazy equation.

Conclusion

Carr was better in 2019 than his stats totally revealed. Given his actual numbers, he should’ve won more games based on his play alone, even with his less than ideal wide receivers, defense, and schedule. However, the results show what even Carr critics will agree with. He isn’t a Top 5 quarterback. I don’t think you’ll find a sane person on Earth that would argue that Carr isn’t a Top 20 QB. He is a good player on a favorable contract.

The discussion really boils down to two things: Is Carr a Top 10 QB and are there better options available? Alex Smith was a good quarterback on a favorable contract, but the Chiefs took a risk on turnover-prone prospect from Texas Tech. The result was a Lombardi Trophy. The Raiders are in a good position to get some wide out help in the draft, hope Mariota pushes Carr to improve, the defense makes a couple stops, and if that doesn’t result in a competitive playoff appearance, then do move on from him, whose contract will be just as tradeable a year from now as it is now. It’s do or die for Carr in 2020 and his 2019 numbers support reasons to be optimistic.

You May Also Like: Second Wave of Free Agency will be slow for Raiders

0 0 vote
Article Rating
Advertisements

Leave a Reply

5 Comments
Inline Feedbacks
View all comments
Joy Washington

We could debate on the do’s and don’ts of DC’S 2019 season. But the real fact of the matter is that if Gruden and company felt MM was their guy DC wouldn’t be there. One thing We do agree on, Derek needs to step it up. Look at what New England did to Brady.

Travis King

First of all. DC is a way better QB than Mariota. Second, you some one that can get open and catch the ball. Third, when Tyrell and Refro were healthy they had a couple of recievers along with their awesome tight ends. So no matter what it is not fare to judge DC when you have other play makers. Period!

Travis King

I meant they need play makers. Not far to judge him with no one who can get open or knows the play book.

Jesper

It would be a lot of fun if you did those regressions for losses instead of wins.
You write “Also, victories are certainly a quarterback stat to a certain degree” I would argue that losses is a better stat for evaluating QB play, simply because it is a lot easier to loose a game single handedly than it is to win one for a QB.
Run them again, and see how DCs stats matches up for number of losses. That’d be really interesting!

Anonymous

D-Fense