FanPost

Is [Insert Available QB] Really Better Than Blake Bortles? And Is It Worth It?

The biggest storyline this offseason is going to be the quarterback position, and the Jaguars are going to be big spenders at the position.

QB Stats and Wins

Sports fans love stats, and love to hate stats, so it is important we start here. Two open-source stats readily available for fans are the often complained about Passer Rating (https://en.wikipedia.org/wiki/Passer_rating) and the often less talked about Adjusted Net Yards per Pass Attempt (ANY/A) (https://www.pro-football-reference.com/about/glossary.htm).
These stats aren't perfect, and you should be wary of anyone claiming they are, but you should be much more reluctant to believe people who are dismissive of these stats entirely.

Lets look at the 56 players who had at least 280 regular season pass attempts from 2014 through 2017, and see how their quarterback stats correlated to wins.

dwPZy8l.0.png

This is a correlation coefficient of 0.658 for passer rating and 0.718 for ANY/A.
Now depending on which textbook you grab for reference, or which link you choose when googling "Strong Correlation", these numbers are either well within the bounds for strong correlation or are very close to meeting that criteria.

In layman's terms, there is a pretty big correlation between teams whose QB performs well in these stats, and teams that win more games. While we can all agree that QB is a very important position, it might be more singularly important to get QB right for a team sport that has over 22 players making contributions towards a win.

And that strong correlation stays present when we look at the average percent difference between the QB's 2014-2017 stats and the rough average over that same time span (a true average is higher as teams generally tend to move on from under-performing QBs, which skews taking the lazy average of those 56 players using www.pro-football-reference.com play index using combined seasons)

iCQ6yO1.0.png

This correlation is important, because it is entirely possible to have an above average game in one stat, but not the other.

A First Look at "Cost"

Currently, there are 16 QB contracts on www.OverTheCap.com averaging 19 million dollars or more per season.

Let's look at the top 20 QBs by passer rating from 2014-2017.
7Zgemib.0.png

These might be a good place to start to find QBs who have been worth the money we put ourselves on the hook for 2018. Bortles is 41st over this time span, so you won't find him anywhere near this list.

And ANY/A...

O1VFObr.0.png

Blake Bortles is 38th over this time span.

"Consistency" and Wins

I then decided to look at the 20 QBs likely to be available in free agency according to Overthecap.com, and see the games where they hit that season's league mean in both passer rating and ANY/A as well as the games where they would fall into the lowest bucket on that season's histogram for both passer rating and ANY/A (min 17 pass attempts in a game).

QI0JqkK.0.png

82+ % of games a QB did very poorly in both passer rating and ANY/A, resulted in a loss for the QB's team.
65+% of the games a QB managed to hit the league mean, resulted in a win for the QB's team (73% if we exclude QBs who produced stats in games that resulted in loses of over 10 points, to account for the possibility of"garbage time" passing stats).

How a QB does, can have a very large effect on your team's ability to win games.

Playing Not to Lose

When you have a good defense, and you go overdrafting runningbacks fourth overall, a common theme is to let the other team make the mistakes and to try your best to capitalize on those mistakes while minimizing your own.
Here are how QBs fared in "worst" game percentages along with the estimated games lost due to QB play given the rate at which teams lose the games in which their QB has a very very bad day as a passer.


QMeNdBs.0.png

The only way to increase your odds of losing games by making a change at QB in free agency, is to pay Austin Davis or Ryan Mallett 19 million instead of Bortles.

"You Play to Win the Game"

But there is more to winning games than just limiting mistakes, so let's look at the expected games won due to the rate at which these QBs have average or above games and the rate at which team's win the games their QBs do well enough in.

SgQ0uNo.0.png

Wins, Loses, and the Unknown

You might have noticed that the numbers for expected loses and the number of expected wins, aren't adding up to a full 16 game season. For these games in between the QB seemingly working against the team's win, and not seemingly winning the game by meeting league mean expectations, we will assume the outcomes of these games in limbo are 50/50 to see what an average team led by each QB might be able to accomplish.


YAs7F3M.0.png

How Much Does a Win Cost?

In 2017 the salary cap was $167,000,000, and the average number of wins per season is 8. Lets just say that teams are prepared to spend about 20.9 M per regular season win. Let's assume you are "all in" on offense, and your top 11 cap hits are your 11 offensive starters. The average percentage of pay for the top 22 QB contracts (excludes the players on cost controlled CBA rookie contracts) is about 20% of the money spent for a team's top 11 contracts.

The QB earning 20%, of that 20.9M per regular season win that might be due to the QB's measured statistical performance is approximately 4.2 M per added win.

This would make the expected free agent market at QB look something like this.

q6H64kG.0.png

How Much Was Bortles Really Worth?

Obviously, paying EJ Manuel, Brock Osweiler, and Blaine Gabbert 19 million isn't anything anyone of us are wont to do, even though they have been on par with Bortles in terms of effective QB consistency wins since 2014. So how much was the production we got from Bortles between 2014 through 2017 really worth in terms of attempting to pay for added wins?

To estimate this, I found the conversion rate from between what the estimated worth of QB play compared to a 19M dollar Bortles was to the actual contract salary averages for 3 players signed to be their teams' starters. These players were Drew Brees, Sam Bradford, and Josh McCown. Graphing the relationship between those players' QB play estimated wins and their conversion rates between actual vs estimated contract sizes, and plotting a polynomial line of best fit gave the following results.

M54GGmk.0.png

Which would estimate the contract worth of Bortles play from between 2014 through 2017 at a hair over 1.7 million average salary per season. Paying Bortles for his peak season play moving forward would be closer to a 11.5M average per season estimation, if you feel the need to ignore three years of sunk costs, and the availability of better QBs and more consistent QBs on the open market.

Closing

I'm sure I'm going to field comments from people that still want to estimate Bortles play far above what it really has been, but you'll have to forgive me (as one of those "stats guys") for sticking to the numbers that show that Bortles (by and large) hasn't been who we expected when we drafted him, has mostly provided play worth far less than we were paying him on a CBA cost controlled rookie contract, and that he falls behind his positional peers on the available QB market.

If there is another QB you would like to see this same analysis applied to who wasn't part of OverTheCap's list of free agents, feel free to let me know down in the comments and I'll do my best to get those numbers to you.

Go Jags.

FanPosts do not necessarily reflect the views of the authors of Big Cat Country or SB Nation.