Teddy Bridgewater and Quarterback Help

No offense has had it easier this year than the Denver Broncos. What do I mean by that? Denver ranks 4th in points allowed, at 276, but that’s a little misleading. The Broncos have thrown three pick sixes, all from Peyton Manning, and those have put 20 points on the scoreboard (one pick six was followed by a failed two-point attempt). In addition, Denver’s defense/special teams has scored six touchdowns. Those obviously go in the “Points Scored” column for Denver, but in terms of the offense, they didn’t earn those points. So instead, let’s subtract all non-offensive touchdowns scored by the Broncos by the points allowed by Denver. Do that, and the Broncos defense has allowed 214 net points, after excluding pick sixes and crediting the defense for non-offensive touchdowns.

That’s the fewest in the NFL. Last offseason, I wrote an article about Andrew Luck and quarterback help.  It was pretty basic, but I found it interesting enough to recreate today.  Here is the methodology:

1) Calculate how many standard deviations from average each team is in net points allowed (negative means fewer points allowed).1  For example, the standard deviation in net points allowed is 51.6, and the average is 310.4; that means the Broncos have been 1.87 standard deviations better than average in net points.

The Saints, meanwhile, have allowed 425 net points, which is 2.22 standard deviations worse than average.

2) Calculate the percentage of offensive plays that involve the quarterback.2 In other words, this is quarterback pass attempts + sacks + quarterback runs, divided by total offensive plays. Jacksonville leads the way here, as Jaguars quarterbacks (which is entirely made up of Blake Bortles) have been involved on 69.9% of all Jaguars offensive plays.

The average team’s quarterback is involved on 64.1% of plays, and the standard deviation is 3.74%. This means the Jaguars quarterbacks are involved very frequently — in fact, 1.56 standard deviations above average. Conversely, Rams quarterbacks have been involved on only 56.5% of plays, which is 2.03 standard deviations below average.

3) Add the two standard deviations to see how much each team relied on each quarterback the most. For these purposes, a bad defense gets a positive number — for allowing more points than average — which a heavy reliance on the quarterback also gets a positive number. So the larger the total number, the more that is asked of that quarterback.

Let’s use three teams as an example: as it turns out, there’s a three-way tie at the top. We already noted that the Jaguars offense leans heavily on Bortles (+1.56); well, Jacksonville’s defense isn’t very good, as it is 1.16 standard deviations below average. Add those numbers, and you get +2.72.

The Saints defense is terrible, checking in at 2.22 standard deviations below average. And New Orleans relies on the quarterback — generally Drew Brees — more than the average team, 0.49 standard deviations above average. Together, those numbers also add up to +2.72. Finally, let’s look at the Cleveland Browns, who have similar numbers to Jacksonville. The Browns have involved their quarterbacks (mostly Josh McCown and Johnny Manziel) in 69.4% of plays (+1.43) and play alongside a defense that has allowed 377 net points (+1.29).

The table below shows this information for all 32 teams:

RkTeamNet PAQB PercZ-Score NPAZ-Score QBPTotal
1CLE37769.4%1.291.432.72
2NOR42565.9%2.220.492.72
3JAX37069.9%1.161.562.72
4DET35969.6%0.941.482.42
5IND35668.3%0.881.132.02
6MIA33669.2%0.501.381.88
7TEN36566.8%1.060.741.80
8SFO35167.8%0.790.991.78
9OAK35466.2%0.850.561.41
10SDG33066.4%0.380.621.00
11BAL32766.4%0.320.610.94
12NYG35364.3%0.830.060.89
13PHI34462.7%0.65-0.360.29
14NWE26768%-0.841.040.20
15ATL29964.7%-0.220.16-0.07
16PIT28665.4%-0.470.35-0.12
17BUF31562.3%0.09-0.47-0.38
18GNB28264.4%-0.550.10-0.45
19NYJ29263.6%-0.36-0.13-0.49
20TAM34359.6%0.63-1.19-0.56
21CHI35458.7%0.85-1.44-0.60
22WAS31561.2%0.09-0.77-0.68
23CAR27063.8%-0.78-0.07-0.85
24HOU28661.6%-0.47-0.65-1.13
25DAL29859.5%-0.24-1.23-1.47
26KAN22764%-1.62-0.01-1.62
27SEA23763%-1.42-0.29-1.71
28DEN21464.4%-1.870.08-1.79
29CIN24960.7%-1.19-0.90-2.09
30STL26956.5%-0.80-2.03-2.83
31ARI22859.2%-1.60-1.29-2.89
32MIN25556.7%-1.07-1.98-3.06

You may have noticed that the titular character in today’s post has yet to make an appearance. Well, scroll to the bottom of the table above. Minnesota quarterbacks (almost entirely Teddy Bridgewater) has been involved on only 56.7% of all Vikings plays. That’s 1.98 standard deviations below average. In addition, Minnesota’s got a very good defense that ranks 1.07 standard deviations better than average in net points allowed. Add those two numbers together, and you get -3.06. That number doesn’t mean much in the abstract, of course, but it is a pretty significant outlier, as only three teams are within one standard deviation of that number.

What does this mean? Well, I think the simplest takeaway is that Teddy Bridgewater has it pretty good. He has a lot of help, in the form of Adrian Peterson. So, too, does whoever is playing quarterback for the Rams, courtesy of Todd Gurley and a very good defense.3 I don’t think it’s “surprising” to see Minnesota and St. Louis at the bottom of this list. Again, the goal here is not to reinvent the wheel, but to quantify what we know: and we know that the quarterbacks on those teams have a lot of help.

More peculiar is the presence of Arizona, in between those two teams. Carson Palmer is having an MVP-caliber season for the Cardinals, and he’s been the most valuable passer in the NFL this year. So you wouldn’t think of him as someone who “needs” a lot of help. He’s not a “game manager” in any sense of the word. But he happens to be on a great team: Arizona’s defense is phenomenal — and as a result of the six defensive touchdowns and one kickoff return touchdown, the Cardinals rank third in net points allowed, and just one point behind the second-place Chiefs.

In addition, his “usage” rate is pretty low, for a couple of reasons. One, he’s really, really efficient: his high ANY/A average means he will take fewer plays than an average quarterback to build a lead. And that brings us to the next reason: because the Cardinals lead the NFL in points differential, Arizona winds up running a lot. The Cardinals actually rank 8th in rushing attempts this year. So this formula sees a quarterback with a great defense and a run-heavy offense, and concludes that Palmer’s life is pretty easy. That’s a half-truth: more accurately stated, Palmer’s life is easy because of a great defense and Palmer (and the rest of the passing offense) make Palmer’s days pretty simple.

One quarterback I was curious about was Alex Smith. The Chiefs rank 2nd in net points allowed, which does jive with the eye test. But Smith’s usage rate is perfectly average.

There have been 15 games this season where a team has thrown for fewer than 150 yards and won, and Smith’s Chiefs have three of those games. That sort of statistic is why it’s easy to think Smith would rank high in the QB Help Index. But here’s something I didn’t know until today: Kansas City has run the second fewest plays in the NFL. That’s because the Chiefs play at the slowest or second-slowest pace in the NFL, depending on your preferred choice of metric.

So perhaps Smith is underrated by conventional stats that look at his low number of attempts. Once you factor in the slow pace at which the Chiefs offense deliberately operates, and Smith’s running — 8.4% of all Kansas City plays are Smith runs4 — he doesn’t stand out as quite a game manager. Thought of another way (and I realize I’m stacking the deck here), Kansas City running backs rank just 26th in carries by running backs! (And Arizona running backs rank 2nd in carries.) How much of a game manager can Smith be if his running backs have just 303 carries?

Finally, did you notice that none of the top 10 teams in quarterback help are going to make the playoffs? That’s not too surprising, given what goes into the formula, but it’s still interesting to note. In fact, New England is the only team with a positive QB Help Index that is going to make the playoffs. Given Tom Brady’s 68% usage rate, that’s not surprising, but it is noteworthy.

What do you think? What stands out to you? How would you tweak the formula?

1. Last time, I did not go through the exercise of converting Points Allowed into net points allowed, so score one for not being a good enougher! []
2. Last time around, I used pass ratio, but this is a better — and more precise — methodology. []
3. St. Louis ranks 13th in points allowed, but jumps to 8th in net points allowed. That’s because the Rams have four return touchdowns and the offense threw two pick sixes. []
4. Of course, this does include kneels. []
• disqus_FxNEckmMhR

I think if this is going to be categorized under the general term “qb help” then wr/te group and offensive line would be very informative to provide a more complete picture. It may be too difficult to gauge and convert into numbers though. Just thinking of the Jags and Bortles being #1 here, and it makes sense with the usage number you calculated and the poor defense of the Jags, Bortles has to do a lot for the Jags to win. The thing that is missed is that Bortles has two studs on the outside in Robinson and Hurns that is among the best wr duo in the league, and also Julius Thomas (disappointing this year because he was out of shape but still a pretty nice target) makes his life a whole lot easier. Contrast that with Teddy and the Vikings, who have a strong defense and AP still being a stud mean Teddy doesn’t have to put it up as much. That WR/TE group is severely lacking in talent though, and that offensive line situation is atrocious, so while Teddy may not be used as much, his degree of difficulty when he is dropping back is pretty high. So I think this data is interesting but I don’t ultimately know what to make of it because there are more things that factor into QB help.

In general, QB’s are evaluated by two methods: stats and wins. The type of help you’re referring to affects a QB’s stats, while the help Chase is referencing affects a QB’s win/loss record. Teddy Bridgewater’s “help” just happens to fall on either extreme, depending on which angle you’re looking from.

• disqus_FxNEckmMhR

This isn’t true at all though, are you saying the receiving corp and o line don’t affect wins and losses? Of course it does, Bridgewater and the Vikings are a more extreme case because the defense and AP make up for a lot of the deficiencies of the offense, but that doesn’t mean those other groups don’t affect a qb’s win/loss record. To your other point, this article makes it pretty clear that it is silly to evaluate qbs on w/l record. All qb’s aren’t asked to do the same thing, and in fact are asked to do very different things and have very different circumstances, and overall only can control a limited part of the game. Look at Philip Rivers, he’s had a fine year, but he has been in possibly the worst situation possible so looking at the W/L record is pretty uninformative.

• sacramento gold miners

If the QB didn’t handle the football on virtually every offensive play, I would be fine with taking more of a baseball approach, since starting pitchers usually only last six innings. But I’m struggling to think of another sport where a single position is so important. So yes, wins and losses must always be paramount when evaluating QB play. Teddy Bridgewater is better than Christian Ponder, and his performance has also lifted the defense because of his efficiency. Ponder made more mistakes which put the Vikings at a defensive disadvantage.

Philip Rivers is in a tough spot, but the Chargers have lost several tight games, so he must take part of the blame for those defeats. Also keep in mind the relocation issue may be a factor in that locker room.

• disqus_FxNEckmMhR

Measuring qbs by wins and losses basically says the qb is the only thing that affects the game.QBs have a huge impact but there are so many other things that can go into whether a team wins or not, it’s simply silly to evaluate them based on their w-l record. Not to say it doesn’t matter, but unless you are going to include the proper context when citing the record, it’s useless on it’s own. If Philip Rivers didn’t have such a bad offensive line and a team riddled with injuries they likely aren’t in such tight games.

So do you blame Drew Brees for the Saints’ losing season, or do you acknowledge that he’s played with a historically bad defense? Did Rex Grossman go 13-3 in 2006 because he’s clutch or because the Bears had a fantastic defense and kick returner to back him up?

• sacramento gold miners

Brees is one of six HOF locks at the QB position today, but does share some responsibility for the Saints 2015 season, age is beginning to show. Rex Grossman was the 2000 Trent Dilfer for the 2006 Bears, reality was going to catch up with him sooner rather then later. Those QBs had good seasons, but weren’t game-changers. The good to great QBs are sustainable, and time has the final judgement.

If anything, the changing nature of the game will dictate wins and losses are even more important. I think it’s impossible to assign a numerical value, but a QB will always play a key role. Some of his contributions just can’t be measured, and that’s why it’s so difficult. Jay Cutler is the Jeff George of this era, and his lack of leadership after exiting a tight playoff game versus the Saints years ago spoke volumes about why he hasn’t been a consistent winner.

I should have been more clear; of course I agree that receivers and o-line affects W/L record. I was merely pointing out that fans often bemoan a QB’s lack of weapons (meaning skill positions) while ignoring the advantages he has in other areas.

I am vehemently against using W/L records to evaluate QB’s. It’s a team sport, and the QB is only responsible for roughly 20% of any given game outcome (estimating). But most people, including so-called experts, trot out QB wins as if it’s a reliable indicator of leadership, grit, clutch ability, and other intangibles.

Great stuff, Chase. When people talk about a QB’s supporting cast, they’re generally referring to his skill position teammates. But in reality, a QB’s entire team affects his play (whether directly or indirectly), so it’s nice to see this quantified. How much work is it to calculate this metric? I’d love to see this going back historically.

• John

There’s no such thing as an individual stat, all stats are team stats. Maybe sacks, but I’ve seen countless times a Reggie white or Bruce smith get double or triple teamed which gave a teammate free run to a defenseless qb where they get the sack. Yea player a technically got a credited with a sack but Reggies presence made that happen

• Trepur

I think you should make it percentage of a teams yards rather then percentage of a teams plays. That way efficient QBs aren’t penalized for their efficiency (though I would add you should treat sacks as positive yardage [or make it a base ~7 yards per sack value], not negative, for the purpose of the formula, since a QBs usage probably correlates positively with sacks).

• That problem with that is passing yards gain so many more yards per play than rushing yards, I think it would distort things. I didn’t crunch the numbers, but my hunch is it would narrow the variance. But it’s something to consider.

• Trepur

But the lower standard deviations mean z-score would be unchanged.