Ranking NFL QBs by Total QBR tells us more about what the Cowboys should pay Dak Prescott

Last week, we took a look at the number of games in which currently active NFL QBs recorded a passer rating of 100 or more and then looked at their 100+ rating games as a percentage of games started.

We saw that Dak Prescott looked a lot better than many expected, while other QBs looked worse than many expected, prompting questions about whether the stats had been cherry-picked to make Prescott look good, about whether passer rating was a valid stat at all, and about whether any stat showing Prescott in a positive light was automatically a fake stat.

So today, as promised in the original article, we’re going to look at an EPA-based metric to see how today’s NFL QBs hold up under an advanced stats microscope, both in terms of good-game and bad-game percentage.

To do that, we turn to ESPN’s Total QBR.

The trusty NFL passer rating was designed to evaluate the passing game only. It does not account for a quarterback’s running game, his ability generate first downs, the amount of sacks he takes, penalties, garbage time stats, and many other things.

ESPN’s QBR is an attempt to fix some of the weaknesses of the traditional passer rating. ESPN’s Sharon Katz and Brian Burke (formerly of Advanced NFL Stats) explain the concept:

Traditional box score stats distort the performances [...] because they (1) fail to account for all of the ways a quarterback can affect a game, (2) don’t put plays into the proper context (a 5-yard gain on second-and-5 is very different from a 5-yard gain on third-and-10), and 3) don’t acknowledge that a quarterback has teammates who affect each play and should also get credit for everything that happens on the field.

ESPN’s Total QBR includes a lot more than the traditional boxscore stats, and instead tries to include all of a quarterback’s contributions to winning, including how he impacts the game on passes, rushes, turnovers and penalties. Total QBR, much like the Expected Points concept, looks at every single play, adds context (e.g. down-and-distance, score differential, win probability etc.) and then allocates credit to the quarterback and his teammates to produce a clearer measure of quarterback efficiency. The metric is expressed as a number on a 0-to-100 scale to produce a player’s Total QBR.

An average quarterback will have a QBR around 50, and a Pro Bowl-level player will have a QBR around 75 for the season. On a game level, however, a QBR of 75 means that holding all other factors constant (defense, offensive teammates, etc.), a quarterback’s team would be expected to win about 75 percent of time, given that level of QB play.

In the tables further down this post I look at 42 NFL quarterbacks and the games they played between 2016 and 2018. In games in which those 42 QBs had at least 10 pass attempts and posted a QBR of 75 or higher, they are a combined 292-77-1 for a win percentage of .793, which is pretty close to the 75 percent ESPN talks about above.

If we accept that a Total QBR of 75 or more denotes a good game by the QB, it follows that a QB with a lot of 75+ QBR games is a good quarterback.

Similarly, if we look at the games in our sample with a Total QBR below 40, those 42 QBs combined for an 84-266-4 record, a measly .246 win percentage or about 25 percent.

It follows that if a Total QBR above 75 is a good game, a Total QBR below 40 is a bad game.

And with all that out of the way, here’s a look at the 24 NFL QBs with at least 30 starts since 2016 along with their “Good game percentage” (Games with a Total QBR above 75 as a percentage of total games started). We’ll look at the QBs with less than 30 starts a little further down this post.

75+ QBR Games (click on column headers to sort)

If you sort the table by “Good Game Percentage” you’ll see Tom Brady and Drew Brees lead all other QBs over the last three years by quite a big margin, as they should. Brady and Brees are easily the two best QBs over the last three years and the only two QBs truly deserving of the word “elite”.

Behind Brady and Brees there is a second tier of QBs who are not at the same level as Brady and Brees but stand out against the rest of their peers by a significant margin. This tier includes Ben Roethlisberger, Dak Prescott, Matt Ryan, Carson Wentz, and Russell Wilson, and may extend to include Philip Rivers, depending on where you feel the cutoff point should be for that second tier. If you want a quarterback that can win games for your team, with his arms, with his feet, with his late-game heroics, or in any other manner, these are the guys for you.

The third tier includes QBs that have an above average good game percentage (the 42 QB average is 29.3%). It includes Alex Smith, Andrew Luck, Aaron Rodgers (remember this is only about the last three years), Matthew Stafford, and Jameis Winston.

The rest is a list of below-average QBs that have a progressively worse good-game percentages. Some of them are close enough to the average (Jared Goff, Marcus Mariota) that there is hope they can put one or two bad seasons behind them and move up the ranks, others like Eli Manning and Joe Flacco are simply bus drivers who are carried by their team more than they carry their teams.

---

On to the mostly younger QBs with less than 30 starts over the last three years. I felt that the small sample size could possibly distort the overall picture, which is why I’m listing them separately.

Sticking to the logic of the tiers we used above, Patrick Mahomes is a clear standout here, and his second year in the league rivaled and maybe even surpassed anything we’ve seen in the league so far. But can he keep it up?

Deshaun Watson and Josh Allen look like they could potentially be tier two QBs, but they too have a lot more games to play.

Tier three has four above average QBs in Nick Foles, Baker Mayfield, Sam Darnold, and Jimmy Garoppolo, but beware the small sample sizes. Mayfield and Darnold at least have youth on their side, so their arrow may be pointing up, but nobody really knows what to expect from Foles and Garoppolo except for their financial advisors.

Beyond that, we have a bunch of below average QBs, most of whom have already been relegates to backup duty.

---

“So,” some might say, “Dak has had some good games. But my eye test tells me he’s had some bad games. Why are we not looking at those?”

Every QB has bad games. But some QBs stink it up more than others, even if you only have eyes for Dak Prescott.

So here’s the same exercise as above, except with games with a QBR below 40. We’ll start off with the 24 QBs with more than 30 starts again, though this time a low “Bad-game percentage” is what you are looking for.

If you sort the table by “Bad Game Percentage” you’ll see a first tier of QBs (Andrew Luck, Matt Ryan, Aaron Rodgers) that simply don’t have a lot of stinkers on their ledger, which means their performance very rarely is the reason the team loses. That does not automatically make them elite QBs, but the numbers here suggests they minimize their errors and bad plays and consistently keep their teams in the game, losing efforts notwithstanding.

The second tier here consists of Ben Roethlisberger, Dak Prescott, Carson Wentz, Tom Brady, Matthew Stafford, Philip Rivers, and Kirk Cousins.

The third tier are still above average QBs (42-QB average is 27.5%), but the have about twice as many stinkers per season as the top guys in the league. It’s a small list with Drew Brees, Marcus Mariota, Jameis Winston, and Russell Wilson.

The rest is once again a list of below-average QBs that have a progressively worse bad-game percentages.

One thing of note here are the W/L records in these bad games. Guys like Dak Prescott (0-7 in sub 40 QBR games), Matthew Stafford (0-8), Jameis Winston (0-9), or Derek Carr (1-12) have really bad win percentages in these games. Others, like Cam Newton (7-9), Tom Brady (3-4), Drew Brees or Kirk Cousins (both 3-6), have much better win percentages, which suggests their teams are better able to overcome a bad day by the QB, whereas the QBs in the first group seem to play on teams that are unable to bail them out on a bad day.

---

On to the players with less than 30 starts, same procedure as above.

Hard to believe, but Patrick Mahomes has not had a game with a QBR below 40 in his short career.

The rest of the players listed are more or less similarly ranked as they were in the previous table.

---

In principle, you want a QB who doesn’t cost you too many games (and ideally wins a few games for you too). Posting a QBR below 40 is a good way to lose games, even if you have a team that can bail you out on occasion with a good running game, a strong defense, or a big-play special teams unit.

And if you want to be a top QB in this league (and be paid like it), it’s not enough to have a bunch of 75+ QBR games that give your team a good chance to win. It’s at least equally important to minimize the number of bad games in which QBs actively lose games for their teams.

Over 47 games (with min 10 pass attempts), Prescott had 18 75+ QBR games. Only three of his peers have a better good-game percentage. And he’s had just seven sub-40 QBR games. Only four of his peers have a lower bad-game percentage.

That smells like a pretty big contract.