How to correlate QB play to winning, for the Gators and their SEC opponents

Embed from Getty Images

Reports are coming out from the Gators fall scrimmages.

Those reports indicate that the areas that the Gators have struggled the past three seasons – quarterback and offensive line – are still works in progress.

In some ways, that is to be expected with the new system being brought in by Dan Mullen. However, these are also the two areas where you can’t afford to have issues. Having them both be weaknesses is how you get the offenses that have frustrated Gators fans over the past two coaches.

A poor offensive line not only limits what a team can do in the running game, but it also forces a QB to deal with pressure. And poor QB play is why Florida has been so poor offensively recently.

There has been a lot of discussion this offseason that Florida’s running backs will need to carry the load. But I worry that if the QB play is substandard, it won’t matter how good Jordan Scarlett or Malik Davis are.

So that begs the question: how much is QB play responsible for winning? Is it 50 percent? 60?

Well, if the metric I’ve developed – yards above replacement (YAR) – is to be believed, it’s more like 90 percent.

Yards above Replacement Primer

What YAR attempts to do is quantify the value that a QB brings to a team compared to an average QB. It is based on one core principle, that a QB’s value is not just tied to his passing ability, but to how much more – on average – he averages per attempt both through the air and on the ground.

The average QB during the 2017 season averaged 7.41 yards per pass attempt and 3.07 yards per rush attempt. YAR says that QBs can generate value both through the air and on the ground. A YAR of zero is exactly average. A YAR between 0.25-0.5 is someone who can win you a game or two but is inconsistent. A YAR between 0.5-1.0 is a QB who can lead a team to 10 wins, maybe more if he has an elite defense. A YAR over 1.5 is a player who has a compelling case for the Heisman Trophy.

Thus, a QB with 200 pass and rush attempts who average 6.41 yards per pass and 4.07 yards per rush would have a YAR of zero. But a much better throwing QB who has 300 pass attempts and 150 rush attempts while averaging 8.45 yards per pass and 1.0 yards per rush also has a YAR of zero.

This gives us a way to compare QBs with significantly different styles.

For example, last year’s Mississippi State team only exceeded the average yards per pass attempt average three times. Florida actually exceeded the average four times. Florida averaged 152.3 yards per game from their running backs while Mississippi State averaged 142.8. But the Bulldogs averaged 32.0 points per game to the Gators 22.1.?

Gators QBs – as hard as this may be to believe – actually outperformed Mississippi State’s QBs through the air, with an average of 6.33 yards per attempt versus 6.16 for State. The difference was that the Bulldogs QBs – Nick Fitzgerald and Keytaon Thompson – rushed for 110 yards per game as well.

This is reflected using YAR. Nick Fitzgerald (6.23 yards/pass, 6.07 yards per rush) had a YAR in 2017 of 0.33 and led Mississippi State to a 9-win season. He was okay, but not great last season.

On the opposite side of the scale, Feleipe Franks (6.28 yards per pass, 0.34 yards per rush) had a YAR in 2017 of -1.45. This was the worst YAR in the SEC last season, and is a big reason why Florida struggled. This isn’t just Franks’ fault, as his rushing yards are no doubt impacted by his offensive line. But it also indicates that he makes poor decisions relative to his peers.

Why is YAR important?

So if you’re reading this, it likely isn’t news to you that Franks did not play well in 2017. It may be news that he was the worst QB in the SEC, but based on the bile being spewed his way this offseason, there are plenty of Gators fans who feel that way.

But this has implications beyond solely just analyzing QBs. I wanted to understand how much QB play correlates to winning, and the results I found were pretty striking.

To examine this, I calculated the YAR for each game for each SEC team in 2017 (called offensive YAR, or oYAR). Then, I calculated the YAR allowed for each game for each SEC team in 2017 (called defensive YAR, or dYAR). I then subtracted dYAR from oYAR (ΔYAR), essentially giving a measure of how good a team’s QB was compared to the opposing team’s QB.

Wins for SEC teams with a positive YAR differential or a negative YAR differential during the 2017 season. (Will Miles/Read and Reaction)

Overall, teams with the higher ΔYAR won 86.1 percent of their games. That rose even higher in conference games, as SEC teams with the higher ΔYAR won 91.4 percent of their games. Much of the drop in the importance of QB play in non-conference games is attributable to the distinct physical advantage teams have in “cupcake” games. It appears that SEC teams don’t unleash their QBs against those teams, preferring to instead pound the ball through the running game.

Regardless of which number you use though, if your QB outperforms the opposing QB, your team will win at a minimum 86 out of 100 times.

This can be further extrapolated to understanding offensive and defensive performances. If we add up offensive YAR (oYAR) for each game and plot versus points per game scored, we get a linear relationship. The same is true for cumulative defensive YAR (dYAR) allowed versus points per game allowed.

Correlation between oYAR and points scored and dYAR and points allowed. (Will Miles/Read and Reaction)

This chart shows clearly why Florida struggled last season. Not only did the Gators have a poor oYAR (-14.0), but Florida also had a poor defensive YAR (6.5). Compare that to Georgia, who had an excellent offensive YAR (24.8) and an excellent defensive YAR (-19.9).

Florida’s poor defensive performance was driven by surrendering an average of 8.0 yards per pass. The good news is that the defense is too talented to post a performance like that again in 2018. The bad news is that returning to 2016 levels (5.9 yards per attempt) is predicated on getting consistent pressure and significant improvement from its defensive backfield.

This looks at individual offensive and defensive units, but subtracting dYAR from oYAR together to get the overall ΔYAR tells us something about the overall quality of a team. And since ΔYAR correlates with winning, that should give us an idea of who overperformed and underperformed in 2017.

Cumulative oYAR minus cumulative dYAR for the 2017 season. (Will Miles/Read and Reaction)

What this chart shows is that Florida’s 4-7 record was perfectly in-line with what we would expect from the QB’s and defense’s play. There were three outliers in 2017: Missouri, South Carolina and Mississippi State.

Missouri had an elite offense (oYAR = 22.8) but a below average defense (dYAR = 2.6). But much of that positive oYAR (13.40) came in 72 and 62-point performances against FBS opponents Missouri State and Idaho, respectfully. Subtract those games from the ledger and Missouri falls right on the line.

South Carolina is an outlier for a different reason. The Gamecocks went 6-1 in one-score games in 2017. That is likely unsustainable, as Will Muschamp-coached teams sported a 47-percent winning percentage in close games from 2011-2016. 9-4 was lucky for the Gamecocks in 2017, and without major improvement from Jake Bentley, they are going to struggle to replicate that record.

Mississippi State is an interesting case. The Bulldogs had an average oYAR of 0.55, didn’t have a game lower than -2.59 (vs. Georgia) or higher than 2.12 (La. Tech). Instead, it was two defensive games that dropped the Bulldogs’ ΔYAR significantly. Against Georgia and Auburn, Mississippi State had dYAR values of 8.39 and 8.21, respectively. Not surprisingly, the Bulldogs lost those games 80-13. But if you make those games just bad defensive performances (dYAR = 2.0), the overall ΔYAR for Mississippi State rises to 16.87, which would be in-line with a 9-win team.

Applying YAR to Florida’s 2018 season

Florida’s offensive YAR was next to last in the SEC (only better than Tennessee). Florida’s defensive YAR allowed was third worst in the SEC (only better than Kentucky and Arkansas).

There’s a lot of work to do.

Emory Jones is a true freshman. Kyle Trask has yet to make a college throw that matters (or a high school one for that matter). Feleipe Franks had a YAR last season of -1.45 and by that metric was the worst QB in the SEC.

It’s going to take significant progress for Mullen to get average play out of his QBs in 2018. According to this analysis, that would take the Gators scoring from 22.1 to 27.1 points per game. That would be pretty impressive and would likely take Florida up to 65-70 in points scored (from 104).

But if the defense doesn’t improve (dYAR = 6.5 in 2017), that is a problem for Florida. The defense actually outperformed its projection last year. Based on that dYAR value, Florida should have given up 32.6 points per game rather than the 27.3 the Gators surrendered in 2017.

Now, we do need to acknowledge that these numbers are somewhat skewed by the extra difficult non-conference game against Michigan last season as well as the cupcake canceled due to the hurricane. But the fact remains, based on these numbers, the Gators could have been worse than its 4-7 record.

But let’s assume that Randy Shannon was just an awful defensive coach and the defense can return to its 2016 levels (16.8 points per game) while obtaining that 27.1 points per game on offense. That point differential would suggest 8.6 wins for Florida this season.

But what if the defense can only improve to 23 points per game this season? That’s a huge jump (likely top-30) considering that based on dYAR, the defense should have given up almost 33 points last season. That point differential would predict a 7-win season for the Gators.

The hope is that Mullen can lead the offense to above average QB play, and that the defense can get back to 2016 levels. But that’s a lot of hope in the quality of Mullen as an in-game coach compared to McElwain.

Mullen outperformed his ΔYAR expectation at Mississippi State in 2017, so perhaps that hope is justified. However, it is worth noting that from 2013-2017 at Mississippi State, ΔYAR predicts that Mullen should have won 40.9 games.

Mississippi State’s record? 40-24.

FEATURED IMAGE USED UNDER CREATIVE COMMONS LICENSE FROm igerard