Friday, May 26, 2017

Stats Explained

Author's Note: This is the second part of a 36,000-word project I drafted earlier this year and have long been considering releasing as an e-book. The introduction is available by clicking hereAny feedback is appreciated.



This book relies heavily on statistics. It is my belief that statistics demonstrate very well what a player has done on the field. Statistics were not invented in 1990 by fantasy sports nerds. They have a purpose.
The eye test, when available, can still fail, both for the uninformed and sometimes even the quite well-informed. The only true eye test doesn’t work without seeing all 22 players on the field, and that is not available except the last couple seasons. Not everyone can watch the games, either, and those who do have memories that can fail or fade over time.
Plus, I love numbers. If you have them, I will use them.
Every football fan owes a tremendous debt of gratitude for the folks at Sports Reference and Pro Football Reference for their fantastic work. Their scoring summaries, game logs, and Play Index provide a great deal of information in this book beyond just total season statistics. I’m sure almost every football fan is familiar with their work, but head on over to www.pro-football-reference.com anyway to check them out. The Play Index (at www.pro-football-reference.com/play-index) is a particularly great tool for finding cool statistics.
That being said, here are the rate statistics that appear throughout the book.
Passer Rating
Passer rating is the official NFL metric of passing efficiency. It creates a single number out of a passer’s completions, yards, touchdowns, and interceptions per attempt. It was invented in 1973 and immediately implemented by the league. Passer ratings in seasons prior to 1973 became official retroactively.
Adjusted Yards per Attempt
Adjusted yards per attempt (AY/A) takes a passer’s yards, credits them 20 yards for touchdowns, demerits them 45 for interceptions, and averages yards per attempt. It is generally considered a more accurate measure of a quarterback’s performance than passer rating, and is used here through the 1968 season.
Adjusted Net Yards per Attempt
Adjusted net yards per attempt (ANY/A) is AY/A but with sacks taken into account. QB sack data is available for 1969 and beyond. This book uses AY/A through 1968 and ANY/A for 1969-present.
All else equal, if passer rating points to one quarterback, and AY/A or ANY/A to the other, I will lean towards the QB with the better ANY/A.
Fourth Quarter Comebacks and Game-Winning Drives
These are abbreviated as 4QC and GWD. The numbers are based on research by Scott Kacsmar, which Pro Football Reference adopted. A fourth quarter comeback means the team was trailing and came back; if the final score is a tie, a QB can still get credit for a fourth quarter comeback. A game-winning drive can come in a tie game, but must win the game, obviously.
Defense-adjusted Yards Above Replacement (DYAR) and Defense-adjusted Value Above Average (DVOA)[1]
These numbers come from www.footballoutsiders.com and are all publicly available on their website, which I recommend to you. The site has been around for over a decade and is very well established and respected. They describe their signature stat, DYAR, thusly: “the value of the quarterback 's performance compared to replacement level, adjusted for situation and opponent and then translated into yardage.” Meanwhile, DVOA “represents value, per play, over an average quarterback in the same game situations. The more positive the DVOA rating, the better the player's performance. Negative DVOA represents below-average offense.”[2] DVOA is depicted as a percentage above average, so 0.0 is exactly average.
Football Outsiders has calculated DYAR and DVOA for every season from 1987-2016. It is used here when available. For the seasons from 1987-2005, it is the only defense-adjusted statistic mentioned. That changes for 2006 because…
Total Quarterback Rating (QBR)[3]
This is ESPN’s stat. More accurately, it is the ESPN Stats & Information Group’s stat, so it’s not just a made-up number created by a bunch of talking heads on a stage.
QBR can be a controversial stat, but passes the smell test more often than not.[4] It is a rate stat, comparable to passer rating, ANY/A, and DVOA. It includes rushing plays as well.
The main benefit of QBR is its attempt to split up a quarterback’s contributions to his own stats from those of his receivers.[5] QBR is available from 2006 onward and, like the Football Outsiders stats, it adjusts for strength of opponent.
There are multiple numbers ESPN makes available that go into QBR. Because QBR is a rate stat, I will also use Points Added, which is a cumulative stat that ESPN describes as the “number of points contributed by a quarterback over the season, accounting for QBR and how much he plays, above the level of an average quarterback.” It is unclear whether this number is adjusted for opponent.
QBR operates on a 0-100 scale, with a 50 being average.[6]
Use of Stats
I did not create any sort of algorithm that tries to incorporate passer rating, ANY/A, DYAR, and QBR into a single number and just go with that each season. Instead, every factor is considered, including the raw completion, yard, and touchdown totals. 

The selections are thus ultimately subjective, which is why you're allowed to disagree!

Note: If you missed it in the introduction, I've also published my review of 1977 as a sample chapter, available here.




[1] All Football Outsiders numbers in this book are freely available on their website.
[2] These definitions are found on every one of their QB leaderboard pages, e.g. http://www.footballoutsiders.com/stats/qb2016. Such pages are also, of course, my source for these numbers.
[3] All QBR numbers from 2006-present can be freely accessed through the URL http://www.espn.com/nfl/qbr.
[4] There have been 15 qualified quarterbacks to put up a QBR of 80 or better in a season: Peyton Manning four times, Tom Brady three, Aaron Rodgers and Drew Brees two each, and Tony Romo, Dak Prescott, David Garrard, and Matt Ryan once each. Garrard’s season, in which he threw 18 touchdowns vs. only 3 interceptions, had by far the fewest plays in the group; in a larger sample, he might well have fallen below 80. “NFL Total QBR - All-Time Best Seasons,” ESPN.com, accessed March 1, 2017, http://www.espn.com/nfl/qbr/_/type/alltime-season.
[5] For more on this, see http://www.espn.com/nfl/story/_/id/6833215/explaining-statistics-total-quarterback-rating. This article was ESPN’s introduction of QBR. The most significant change to QBR since then is that it now contains an opponent adjustment, a change announced in 2016 at http://www.espn.com/nfl/story/_/id/17653521/how-total-qbr-calculated-explain-our-improved-qb-rating.
[6] Although, over the years QBR’s have seemed to me to be rising. I haven’t been able to find or get an answer from ESPN Stats & Info as to why.

Monday, May 22, 2017

Introduction to Most Valuable Quarterbacks

UPDATE! This project is now available on Amazon at http://amzn.to/2vjKAxG
Author's Note: This is the introduction to 36,000 words I drafted earlier this year and have long been considering releasing as an e-book. Any feedback is appreciated.

Welcome to Most Valuable Quarterbacks: 1951-2016. This is not a book about careers. It is about seasons. The best seasons quarterbacks have ever had. Or at least, the best season a quarterback had each year.
I hope I will bring to light some great seasons that weren’t recognized at the time, and I hope you will enjoy all the stories, numbers, and stories told by numbers within.
Multiple media organizations vote on the subject of the best quarterback every year for their All-Pro teams. The Associated Press produces by far the most popular and widely-cited of these teams. The AP announced All-Pro teams before 1951, but did not have a spot for quarterbacks before then, hence the starting point of this project. Sometimes, but not always, the alternatives to the AP, the most famous being Pro Football Weekly, will produce a different result than the AP. That openness for disagreement was part of the inspiration for putting this thing together.
The MVQB selections are all my opinion. I didn’t take a poll or create a stat. If you find yourself disagreeing with me at times, that means this book is generating critical thought and succeeding. If you find yourself disagreeing with me all the time, I don’t know whom to blame—I don’t think any of my picks are too adventurous, even if a couple of them might be seen that way by some people. And if you find disagreement disagreeable, feel free to interpret my arguments as devil’s advocacy. If you even want to engage in a good-faith discussion on a specific selection, feel free to email me (address at the end of the book).
Here’s the layout. Each year is discussed, but the order isn’t exactly chronological.
For the most part, the main foci of the discussion are the Associated Press 1st Team All Pro picks at quarterback. I will compare and contrast my own Most Valuable Quarterback (MVQB) selections with those the AP picked for their All-Pro team. First I discuss years in which I deviate from the AP, but not easily. Then there are years I stay with the AP, also not easily. Next come four years where I believe the AP got things very wrong.
Then, there are years so historic, everyone agrees who the best quarterback was; for these seasons, I discuss some other quarterbacks who had great years and might have been All-Pro or MVQB in other seasons. These three segments cover about half of the period from 1951-2016. The other seasons are briefly discussed in reverse chronological order; in all of these, I agree with the AP selections.
There are two additional sections; one discusses various players, teams, and trends related to the MVQB award, and then there is a conclusion [content TBD].

The MVQB award, like the All-Pro (and MVP) awards, is for the regular season only. It would make for a pretty silly comparison if I were creating an award that accounted for playoffs as a “replacement” for a regular season-only award.

All throughout, there will be all sorts of interesting tangential information in footnotes, so I hope some of those stats are fun.

Anyway, that’s what you’re in for. I hope you enjoy.

A final note: It occurred to me more than 19,000 words into this project that something like this had been done before with NBA MVP’s by Bill Simmons as a chapter of his Book of Basketball. I note that only because the baseball website FanGraphs (www.fangraphs.com) once met backlash after its first annual trade value column, similar to a thing Simmons had previously done with the NBA, albeit in a much different style. Like that, this is hardly a proprietary concept, but let the similarity be duly noted.
However, I was not consciously mimicking Simmons when I set out on this project. I have not even reread that section of his book since recalling it, so as not to consciously or subconsciously adopt any part of his approach. And I do remember enough to suggest that when Simmons “corrects” the record on the NBA MVP, part of his shtick involves the conceit that he is definitely right. I try not to give any such impression. There are times where it will even seem like I’m apologizing for my choice.[1]
Okay, with that disclaimer out of the way, let’s dig in.

Note: For a sample chapter, click here




[1] I also use footnotes for non-citation reasons, in case it’s claimed Simmons invented those too.