≡ Menu

Boxscore vs. PFF: Era in Review

Adam Steele is back with more analysis of traditional box score stats versus Pro Football Focus’s big time throw and turnover-worthy play metrics. And we thank him for it.


A couple of weeks ago, I compared TD/INT and BTT/TWP numbers for the 2021 season. Today we’ll be looking at the entire Pro Football Focus era going back to 2006.

Before compiling the data, I hypothesized that TD/INT and BTT/TWP would track in relative lockstep, though perhaps the upward slope of the PFF metrics would be less severe. That turns out to be true for 2006-07 and 2014-21, but oh boy was there some wackiness taking place in between. In the graph below, you’ll see league TD-INT difference in blue and league BTT-TWP difference in red: [continue reading…]

{ 0 comments }

Guest Post: Improving on TD:INT Ratio

Adam Steele is back with the crazy notion that we should stop using stochastic, binary events in ratio form as the basis for judging quarterbacks. Fancy that. We thank Adam for his ideas and analysis.


The most commonly cited quarterback stats in mainstream analysis are touchdown passes and interceptions, usually presented as TD/INT ratio. This essentially functions as shorthand to compare the quantity of a player’s great plays against his terrible plays. But this is quite unfortunate since both stats are very noisy and situation dependent. TD/INT ratio not only lacks important information but it can be downright misleading at times.

Luckily for us the good folks at Pro Football Focus have come up with a much better alternative: Big Time Throws (BTT) and Turnover Worthy Plays (TWP). These stats are tabulated by watching film so they capture far more signal than the process-blind box score numbers. Passers get credited with a BTT when they make a throw that goes well beyond what’s expected on a given play, and this includes passes which are dropped or wiped out by penalty. Meanwhile a TWP is charged when a throw is made that has a good chance of being intercepted (whether it’s actually picked or not), or when the QB gets careless with the ball during his dropback and fumbles when such an error could’ve been avoided. [continue reading…]

{ 2 comments }

The penultimate week of the season was an odd one from a statistical standpoint. QBR and PFF mostly agreed (for once), but some of the boxscores straight up lied to us about how well quarterbacks played. We saw 45 touchdown passes and 32 QB turnovers, and that’s standard fare for a late season week in today’s NFL. However, according to PFF graders, quarterbacks registered 36 big time throws and a whopping 55 turnover worthy plays!

Let’s look at the week 17 rankings then take a closer look at some of these misleading statlines: [continue reading…]

{ 0 comments }

Pick a QB, any QB: there are no right answers.

The 2018 NFL Draft was supposed to change the landscape of the NFL at the quarterback position. Maybe not right away, of course, but in a few years — say, 2021? — the five quarterbacks selected in the first round of the 2018 NFL Draft would be the stars of the day. Instead, Josh Rosen flamed out immediately, Sam Darnold proved to be underwhelming under three different coaches, and Baker Mayfield’s stock fell dramatically in his fourth year. Even Lamar Jackson, the 2019 AP MVP, has fallen off; after a notable dropoff in play from 2019 to 2020, he fell further in an injury-plagued 2021. At this point, only Josh Allen is an unimpeachable franchise quarterback, but even he has seen a significant decline in passing efficiency this season.

All told, the 2018 first round quarterbacks as a group have been decidedly below average as passers this season, with three of the four starters (excluding Rosen) being in the bottom five of the NFL in interception rate.

This made me curious: which draft classes have been the most productive in 2021? With 17 weeks in the books — a traditional NFL regular season — here’s what I did. [continue reading…]

{ 0 comments }

I’m short on time right now so this week’s QB rankings will be presented without commentary. [continue reading…]

{ 0 comments }

Adam Steele is back to give us his thoughts on this week’s quarterbacks. It’s been fun to follow his progression from hopeful fan to ranting madman over the course of just fifteen weeks.


This is starting to sound like a broken record but we just witnessed yet another week of terrible quarterbacking. Only 9 of 32 qualifiers even cracked a QBR of 50! We can’t even blame this on backup QB’s dragging down the average as the bottom 10 were all regular starters aside from Mike Glennon. If anything, the backups outperformed the starters with Tyler Huntley taking the week 15 crown and Nick Mullens placing eighth. [continue reading…]

{ 0 comments }

Adam Steele is back with another quarterback recap. He has a broken arm and a detached retina, and he’s ready to win it.


This may have been the least interesting week of the 2021 season for overall game quality (favorites were 12-2 with some totally noncompetitive matchups), but it was still a fascinating slate for analyzing quarterback performance. [continue reading…]

{ 0 comments }

As has been the case for nearly two months now, the steady decline of NFL offenses continued in week 13. Scoring has dropped from 24.8 points per game last year to 22.9 this year. The league is currently averaging 11.0 yards per completion; if this holds it will be the lowest in NFL history. There’s also been a marked shift in touchdown passes vs. interceptions. Early in the season there were 3.5 TD passes for every INT; that ratio is now below 2 to 1. For the first time in several years the NFL has found a nice equilibrium between offense and defense. It’ll be interesting to see if the competition committee devises rule changes to boost offense again in 2022.

Here are the week 13 rankings: [continue reading…]

{ 0 comments }

This week I’m excited to introduce a new data set into my quarterback rankings, courtesy of Pro Football Focus. I’d like to give a helmet knock to our good buddy Bryan Frye for working out some details behind the scenes to make this possible. [1]Note from Bryan: all I did was ask.

From this point forward, I will be using two metrics to rate quarterbacks: QBR and PFF offensive grades. This makes me giddy because both systems attempt to isolate the QB’s contribution from that of his teammates. That’s a significant step up from ANY/A, DYAR, and EPA which simply assign team offensive statistics to the QB taking the snaps. I can live with that for historical comparisons where we don’t have anything better, but in today’s world of robust data there’s no reason to settle for such a high degree of entanglement.

As neither QBR nor PFF grades account for workload, I needed to make an adjustment to prevent low usage QB’s from hogging the top of the rankings. After experimenting with a few ideas I settled on adding a z-score for play count (based on qualifiers only) and giving it half weight compared to the z-scores for the two metrics. It’s not perfect but it gets the job done without too many arbitrary decisions.

Let’s see how the new system looks for week 12: [continue reading…]

References

References
1 Note from Bryan: all I did was ask.
{ 0 comments }

Adam Steele is back, and he’s eschewing the expository fluff (which I am re-adding, right here). Enjoy, friends.


Here are this week’s quarterback rankings: [continue reading…]

{ 0 comments }

As with the rest of these recaps, the ideas and analysis presented here are from Adam Steele. I’m just a dunce with admin rights.


We’ve seen a steady decline in quarterback play across the last month, and this week hit a new low. The unweighted average QBR for week ten qualifiers was a dismal 44.7. That would have ranked 27th in the league last year! This isn’t a surprise as offense tends to decline in the second half of every season as defenses jell and the weather starts to make an impact.

Here are this week’s numbers: [continue reading…]

{ 0 comments }

Adam Steele is back, continuing to refine his methods in real time, and we get to benefit from it. Thank you Adam, for working it out publicly.


After tabulating the numbers for week nine I realized I needed to make one more tweak to the formula. Since EPA per play and QBR are both agnostic to volume, QB games with a low number of plays were disproportionately clustered at the top and bottom of the rankings. Obviously it’s harder to maintain an extreme performance over a larger sample than a smaller one. My solution was to regress EPA/P by adding 20 plays of 0.1 EPA (roughly league average) to everyone’s stat line before calculating their z-scores. This fix strikes a nice balance between efficiency and volume.

Onto this week results: [continue reading…]

{ 0 comments }

Adam Steele is back with a brand new convention, and we thank him for keeping the site humming.


With the NFL trade deadline taking place this week, I decided to make a trade of my own. I’m ditching DYAR in favor of EPA per play. After Davis Mills placed seventh in DYAR by turning into Dan Marino down 38-0, I knew I had to switch to a metric that filters out garbage time.

Thanks to Ben Baldwin and his nifty site rbsdm.com, it’s easy to query EPA/P with various amounts of garbage time removed. After some experimentation I settled on a 4% filter; plays which occur when win probability is below 4% or above 96% are thrown out. The vast majority of plays are still counted but nonsense like the Davis Mills experience is rightfully ignored. To wit, Mills drops from -.149 to -.474 EPA/P with this filter applied.

Here are the week eight numbers: [continue reading…]

{ 0 comments }

Guys, gals, and nonbinary pals, Adam Steele is back with his quarterback recap. And we thank him for it.


 

I’m going to keep the commentary short and sweet today, so here are the week seven rankings: [continue reading…]

{ 0 comments }

After offenses lit up the scoreboard in week five, we were treated to the poorest quarterbacking of the year in week six. Both DYAR and QBR hit their low points in the 2021 season. What’s shocking is that this happened with the Jets on their bye week!
[continue reading…]

{ 0 comments }

Finally, a week in which the best QB games were more extreme than the worst. The league saw DYAR and ANY/A reach their high points for any week in 2021. Oddly, this was the second worst week for QBR despite the meteoric conventional stats. There seemed to be an unusual number of highlight reel catches and long completions on busted coverages, both of which are likely discounted in QBR.

Fittingly, old man Brady tops the chart in the same week he became the all time leader in total DYAR. Brady and Lamar Jackson had the two best games of the year according to DYAR but fared much worse in QBR. In Brady’s case it’s likely because he faced almost zero pressure vs Miami, while Jackson gets taken to the woodshed for his goal line fumble (another example of QBR overweighting running plays).

Josh Allen had the opposite result – dominant in QBR but merely good in DYAR. He was very successful on his runs, completed his average pass a whopping 13 yards downfield, and didn’t have enough plays to pump up his counting stats. [continue reading…]

{ 0 comments }

With it being the (almost) quarter point of the 2021 season, I’m going to skip commentary on the week four games and focus on quarterback performance across the first month of the season. [continue reading…]

{ 0 comments }

Adam Steele is back with his quarterback recap for week 3, 2021. Thank you, Adam, you beautiful man.


 

The theme of week three is the same as week two: terrible rookies. Remarkably, the eleven worst QB games this season have all been more extreme than the single best game. Anyone reading this is well aware of Justin Fields‘ spectacularly awful sack-fest performance against Cleveland. But according to DYAR, that wasn’t even the worst game this week!

Here are the week three numbers: [continue reading…]

{ 0 comments }

The following post attributes authorship to Bryan Frye, but everything under the line comes from the mind of Adam Steele. We thank him for his contributions to the site and to football discussion.


 

For the second week in a row, the worst quarterbacks had more extreme performances than did the best. By absolute value, the eight worst games this season have been more extreme than the single best game. I don’t have a good explanation for this other than sheer randomness. [continue reading…]

{ 0 comments }

The following post attributes authorship to Bryan Frye, but everything under the line comes from the mind of Adam Steele. We thank him for his contributions to the site and to football discussion.


 

With Chase being such a busy man, I have taken over posting weekly passing reviews during the 2021 season. While the classic ANY/A formula has served its purpose over the years, I’m going to tap into a couple of more advanced metrics to rank quarterbacks on a weekly basis.

Each week, qualifying QB’s will be scored using Football Outsiders’ DYAR and ESPN’s QBR metrics. I think this will give us a nice balance between play-by-play and charting stats, as well as a balance between counting stats and pure efficiency. The qualifying players will have their z-score calculated for DYAR and QBR then averaged to create their overall score.

Here are the week 1 results: [continue reading…]

{ 0 comments }

We’ve come to the end of the line. After several posts ranking and reranking, thinking and rethinking, quarterbacks with Total Adjusted Yards per Play and its descendants, this is the one I imagine most readers really want to see. Today, we are looking at measured performance in the regular season and playoffs combined. This is where guys like Y.A. Tittle, who feasted in the regular season but nearly always faltered in the postseason, see their positions fall down the list. Where passers like Jim Plunkett, whose regular season performances left much to be desired but went full tilt bozo in the playoffs, rise up the ranks. As far as the NFL record book is concerned, the playoffs don’t count toward career stats or win-loss totals. While I understand not rewarding players for getting to participate in more games, I can see the argument that it is equally unfair not to reward them for playing well enough to continue the march toward a championship. In order to balance those ideas, I have only counted playoff performances that measured above average by TAY/P.

A quick word on the numbers I’m using. You can find more detail in previous articles in the series, but this should be sufficient to introduce the rookies and refresh the veterans. [continue reading…]

{ 0 comments }

We’re back at it with quarterback rankings based on Total Adjusted Yards per Play and its abundant offspring. This time, we’re getting into combined regular and postseason stats for single seasons. For the purposes of this article, I will refer to this as a full season. None of the stats are new and have been explained in what I hope is sufficient detail in previous posts. [1]Here are links for the base methodology, the introduction of Z Value and positive value, the methodology and refinement of championship leverage, and a brief explanation of retroactive leverage. Fun … Continue reading As fun as it would be to call this “the greatest quarterback seasons in history!” or something like that, I seem to have a deeply held grudge against page views and web traffic, because I can’t get behind calling it anything of the sort. This is one measure of how much quarterbacks produced in a given full season. I believe it is the best measure when trying to compare across eras in which superior metrics don’t exist, but that’s about as far as I can go on the hubris tip. Anyway, these are my numbers. I hope you like them. [continue reading…]

References

References
1 Here are links for the base methodology, the introduction of Z Value and positive value, the methodology and refinement of championship leverage, and a brief explanation of retroactive leverage. Fun fact: with the addition of a game to the schedule, championship leverage will increase for the 2021 season!
{ 0 comments }

We have already looked at postseason performance in single games and single seasons. Today, we’re finally having a go at full playoff careers. All of the metrics I am using today have been explained, in great detail, in the previous four posts, so I am not going to belabor the point here. I will, however, remind the reader that Total Adjusted Yards per Play, and its many variants, is just one approach to measuring quarterback performance. It doesn’t account for weather, and it is not adjusted for the strength of opposing defenses (not yet, at least). Moreover, these numbers are based on box score stats and do not include more granular information, like time on the clock, field position, and yards to go on a set of downs. A four yard pass on 1st and 10 is much less valuable than a four yard pass on 3rd and 3, but TAY/P treats them equally. This is by design, because the goal of this metric is to do the best possible job of comparing quarterbacks across eras. I can’t do that with DVOA or EPA/P, because the play by play data just don’t go back far enough. However, when I looked into the correlations of TAY/P with more granular metrics, the r value tended to land between .93 and .94 (even for ESPN’s QBR, with its often wacky use of win probability). This suggests, to me, that most of these issues smooth themselves over in the long run. [continue reading…]

{ 0 comments }

Yesterday, we looked at playoff performance in individual playoff games, as measured by Total Adjusted Yards per Play (and its copious derivatives). Today, I’m taking the next logical step and looking at performance over a full postseason in any given year. In other words, instead of examining who produced the most in the Super Bowl last year, I am determining who produced the most over the entire 2020 playoffs. By that, I mean who produced the most as measured by this particular set of numbers. They happen to be my preferred numbers for comparing across eras, but your mileage may vary.

For the uninitiated, here is a brief rundown of the metrics used:

Total Adjusted Yards per Play is like ANY/A with rushing included. It is (pass yards -sack yards + rush yards + 20*pass TDs + 20*rush TDs – 45*interceptions – 25*fumbles) / (passes + sacks + rushes). This version of TAY/P doesn’t include first downs, since I only have reliable first down data back to 1991 and want to make the playing field as level as possible when comparing back to 1936. [continue reading…]

{ 0 comments }

Before the 2021 season starts, I figured I would try to add a little more to the Total Adjusted Yardage information I have been posting. Today, I want to discuss the same stats as before, as well as a few additions, for every postseason game in which a quarterback had at least five action plays. Not all information is complete, as sacks are unavailable prior to 1948. However, I’m working with what I have and not looking back. [1]If you want to look at all the raw data, including quarterbacks with just one plays, you can check out this Google sheet. Because this is the postseason, it is inherently worth more with regard to both earning a championship and establishing one’s legacy. Therefore, I am going to include championship leverage in the discussion. I don’t have much to say, so let’s get to the numbers. [continue reading…]

References

References
1 If you want to look at all the raw data, including quarterbacks with just one plays, you can check out this Google sheet.
{ 0 comments }

Yesterday, we looked at the best (and worst) regular season performances, as measured by Total Adjusted Yards per Play and its many derivatives. Today, it’s time to look into total career values. Keep in mind, these figures don’t include the postseason, where many legends cemented or defined their legacies. We will get to that later, I promise.

Because I apparently hate driving traffic to the site, I will not title this anything to do with the greatest or best quarterback. Instead, I want to be honest about the fact that the results below are simply one measurement of career performance and are not meant to be definitive. I do believe it is the best approach I have seen when it comes to using numbers to compare quarterbacks across eras, but it isn’t perfect. When you see “Johnny Unitas,” what you are really seeing is Unitas, throwing to Raymond Berry, John Mackey, Lenny Moore, Jim Mutscheller, and Jimmy Orr, handing off to Alan Ameche, and standing behind Jim Parker and Bob Vogel, while glancing over at Weeb Ewbank and Don Shula standing on the sidelines. When you see “John Elway,” what you are really seeing is Elway throwing to a ragtag group of receivers, playing behind a ho hum offensive line, and under the tutelage of an unimaginative head coach during his prime, before getting basically the opposite of that late in his career. The average reader at Football Perspective has a good grip on both history and stat and should have little trouble contextualizing the numbers presented today. [continue reading…]

{ 0 comments }

It has been a while since I published anything specifically relating to my niche homebrewed metric, Total Adjusted Yards per Play (TAY/P). It has been longer, still, since Chase has posted his latest entry into the corpus of his legendary QBGOAT series. I figured I would kill one bird and dizzy another with one stone. This post is specifically about TAY/P and its derivatives. It is not a treatise on the greatest quarterbacks of all time. Instead, it is a look at how several quarterbacks have fared in a few different variations of a single measurement since 1932. If you want to view it as a GOAT list, you are free to do so, though I would not encourage it. I believe statistics should be used to support your argument rather than serve as the entirety of your argument.

I am not under the delusion that this is the premier metric to use when evaluating quarterbacks. I prefer a holistic approach that includes everything from simple box score stats to convoluted algorithms and subjective grades. I consider EPA/P, DVOA, and Total QBR to be superior measurements. [1]CPOE is a useful stat when trying to get a better idea, from the numbers, who was more or less accurate than their actual completion rate may have indicated. ANY/A is good but only looks at … Continue reading The only problem is that they don’t cover a large enough portion of NFL history to make comparisons. Thus, I continue to use TAY/P because it uses simple box score numbers to create a metric that can compare quarterbacks dating back as far as we have box scores. [2]With some caveats. We have full stats dating back to 1967. Prior to that, we don’t have full sack and sack yardage information for the AFL. We have data for sack yardage lost in the NFL dating … Continue reading [continue reading…]

References

References
1 CPOE is a useful stat when trying to get a better idea, from the numbers, who was more or less accurate than their actual completion rate may have indicated. ANY/A is good but only looks at dropbacks. Success rate is usually instructive when looking at how a quarterback leads an offense, but it is defined differently by different entities and, thus, can be difficult to discuss without first defining the term. I prefer to count plays with positive EPA successful, rather than the 40/70/100 division or some variation thereof.
2 With some caveats. We have full stats dating back to 1967. Prior to that, we don’t have full sack and sack yardage information for the AFL. We have data for sack yardage lost in the NFL dating back to 1947, but we don’t know the number of sacks themselves prior to 1963. We have precious little sack information for the AAFC. In the NFL, we don’t have fumble data earlier than 1945, and we don’t have fumbles or for the AAFC at all. Prior to 1936, NFL teams didn’t even play the same number of games, which makes serious analysis tricky. And before 1932, we only have touchdowns. This all ignores the fact that yards are awarded a whole numbers, even when only half yards are gained. A touchdown from the one inch line still counts as a one yard run, by rule. On one play, that is a small deal, but over the course of a long career, it can add up (or take away). Though it mostly evens out.
{ 0 comments }

Update:

The 2021 NFL Draft saw 156.7 points of draft value spent on quarterbacks, the 2nd-most in NFL history behind only 1999.  As expected, Lawrence and Wilson were the first two picks, with Lance going third overall.  The fourth quarterback didn’t go off the board until the 11th pick, which was a bit of a surprise: that was Fields to Chicago, and the fifth quarterback (Jones) went to New England at 15.  Trask, Mond, and Mills were indeed seen as the best of the rest, and went in the span of four picks in the 60s.  The margin was so close that had Fields been selected with the 8th overall pick, and the rest of the quarterbacks were chosen in the same spot, then the 2021 Draft would have exceeded the ’99 Draft in terms of draft capital spent on quarterbacks.

The rest of the original article is below.

The 2021 NFL Draft looks to be extremely quarterback heavy. The Jaguars are going to select Clemson quarterback Trevor Lawrence with the first overall pick, and the expectation is that the Jets will draft BYU passer Zach Wilson with the second selection.  The 49ers will likely draft a quarterback with the third pick, too, and the rumors are that it could be Alabama QB Mac Jones, or Ohio State’s Justin Fields, or even North Dakota State’s Trey Lance; regardless, all five quarterbacks are expected to go in the first round, and perhaps even all in the top ten!  Three other quarterbacks — Stanford’s Davis Mills, Florida’s Kyle Trask, and Texas A&M’s Kellen Mond could have early picks used on them, too.

The most QB-heavy draft in NFL history was not the famous 1983 Draft — which featured three HOF passers and six quarterbacks selected in the first round — but rather the 1999 Draft.  That year, quarterbacks were taken with the first three picks, and two more were drafted in the top fifteen; second, third, and two fourth round picks were also used on quarterbacks.

I looked at every NFL Draft since 1950 and calculated how much draft capital was spent on quarterbacks each year.  The picture below shows those results, using the Football Perspective Draft Value Chart.

Let’s say that in the 2021 Draft, Lawrence/Wilson/Fields go with the first three picks, Lance and Jones get selected at 7 and 9, and as The Athletic’s Mock Draft by Dane Brugler provides, Mills goes to the Patriots in the middle of the second round, Mond is taken by the Bears a few picks later, and Trask is a fourth round pick by the Vikings.  If no other quarterbacks are selected in the top 224, that would mean that the draft capital spent on quarterbacks in 2021 was equal to 159 points using the Football Perspective Draft Value Chart.  That would, by a hair, make this the most quarterback-heavy draft in NFL history, as the ’99 Draft had 158.6 points of draft capital spent on quarterbacks. And if quarterbacks go 1-2-3-4 to start the Draft, that would up the allocation to 162.6 points. It seems very likely that 2021 will be either the #1 or #2 quarterback draft in NFL history, at least according to draft capital spent on the position.

{ 0 comments }

History of the Career Passing Yards Record

This is an update and revision to a post I wrote after Peyton Manning’s retirement following the 2015 season. I originally penned it to celebrate Manning’s triumph over the record books and look back at the history of the record. Since then, Drew Brees and Tom Brady have broken that record, and both look to push it to new heights as they battle for the crown. Brees beat Brady to the mark and hasn’t looked back yet. The Saint has a history of beating the odds. He has been lauded by an adoring media and legions of fans, and deservedly so. However, with all the attention given to modern players, we often fail to properly remember former greats – legends of the game who paved the way for the sports celebrities of today.

More than relics from days past, these men were trailblazers who helped legitimize the sport we have grown to love. Unlike the iconic sports figures of today, many of these players were actual heroes, serving in the military and coming home to work full-time jobs to support their families. Before quarterbacks were millionaires, they were mostly indistinguishable from the everyman. Except on Sundays. On Sundays, they became giants. [continue reading…]

{ 1 comment }

Over the last few years, I have been updating the career passer ratings for NFL quarterbacks to adjust for era. Over the last 100 years, the NFL has consistently approved rules changes to make passing easier, and as a result, passer rating has consistently spiked:

[continue reading…]

{ 1 comment }
Previous Posts