Mike McCarthy - "The Analytics Guru That's Not"
For most of his career, Mike McCarthy as a head coach in the National Football League was a mediocre, old-fashioned type, a "mid-2000’s NFL head coach."
He was a dinosaur – but not a terrible dinosaur.
Today, McCarthy as a head coach is a horrendous pseudo-sabermetrician.
And he may be the worst coach in the NFL.
Despite his Super Bowl XLV run, which started from the 6th seed in 2010, McCarthy in Green Bay was not a great pass play designer and had zero creativity in the run game. He also had a lack of understanding of how much a running game can help out a quarterback – help out any quarterback, no matter how talented that quarterback was.
He was eventually fired, and the perception was that the league had passed him by.
How do you fix that perception?
Well, according to McCarthy himself, he dove into “the analytics” during his off-year, re-inventing his philosophy to help best prepare a team to win in the 2020's.
Let’s look at the results.
First the obvious: the Cowboys – a top six Super Bowl favorite entering the year – are 2-5 SU and 0-7 ATS. Now let's dive into the how, and the why. There are three areas specifically where I think McCarthy's intent to be “analytically-driven” is exactly in line with what makes the Cowboys so bad at football.
2020 Dallas Cowboys Rankings:
- Pace of Play: 1st (21.7 seconds per snap, 3.3 seconds faster than anyone)
- Pass/Run Split: 3rd Heaviest Passing Team (65% passing, behind Jacksonville & Chicago)
- 4th Down Attempts: 2nd (13 attempts in 7 games, only behind the Atlanta Falcons' 16)
"Play fast, throw a lot, play aggressive"
Sounds good right?
McCarthy and the Cowboys are doing all the things that the "Barnie-at-the-Bar" types will tell you – with some awful sense of authority – what the numbers say are the right things to do.
When they awaken from their mid-day nap, these guys will point you to EPA to validate their claims:
EPA per play: Expected Points Added per Play, and it's all the buzz these days. On its surface, I admit, it is a very useful tool, but it's not end all be all. At most, EPA is half of the battle. Scoring 60 instead of 30 does not give you extra wins if you shut out your opponent.
Expected Wins Added Per Play is the "God-Metric." We just haven’t divined its calculus quite yet.
Let’s break down why Dallas' supposedly analytics-driven methods have backfired so magnificently:
Pace of Play: 1st
Is it possible that it's not a coincidence the Cowboys lead the NFL in both giveaways AND pace of play - and lead both by a wide margin?
Dallas is getting plays off 3.3 seconds quicker than ANY team over the last 3 years.
Running a quick regression analysis (YouTube + Excel is a phenomenal learning tool, my people) of the last three years of data, we also see that there is a strong correlation between Pace of Play and Giveaways Per gGame. (P Value is <.00001, meaning there is almost no chance the correlation is coincidental).
About 10% of giveaways can be explained by the pace with which you play. Moreover, for every 10 seconds quicker you snap the ball, you average one more turnover per game. So, it should be no surprise to us that Dallas is in fact leading the NFL in giveaways while playing at this blistering speed.
Think about it. How much better and calmer are you when you take just one more deep breath before executing something difficult? Is there any basketball player anywhere that does not take a deep inhale right before shooting a free throw? I had a friend nail a job interview recently.
Her secret? She said she just took one extra second to breathe and think before answering their questions. Take a second and breathe. What a genius concept!
But Chip Kelly and the "new Mike McCarthy" will tell you that they are gaining some advantage by hurrying up and not allowing the defense to substitute. For a given possession, maybe it does help – but one extra turnover every 10 possessions means that you lose almost every time.
Now let's look at the other side of the table: the Green Bay Packers, 32nd with a bullet in pace of play, are also 32nd with a bullet in giveaways (meaning they're #1 in protecting the ball).
Maybe it's a coincidence, but Matt LaFleur has often been called out for being a bad play-caller because of his run-heavy style, yet he sits at 18-4 overall as an NFL head coach.
The 40-year old LaFleur (even though he looks 18) is somehow the dinosaur here.
Yet somehow the real dinosaur is burning the countryside.
Meanwhile, the old slovenly looking guy (McCarthy) is getting worked while thinking he is keeping up with the times. What once was old is new again. And what once was a bad head coach is now horrendous.
Pass/Run Split: 3rd
The 2020 Cowboys are 3rd in the league in pass percentage, throwing on 65% of their possessions, and they are losing: 0-7 ATS. I could draw out that correlation more, but I think this comparison of McCarthy'd former team is more illustrative.
If you look at the 2018 Green Bay Packers (6-9-1 under McCarthy) and the 2019 Green Bay Packers (13-3 under LaFleur), almost all of their offensive stats ended up about the same on a per play basis.
Yards Per Play, Yards Per Rush and Yards Per Pass were almost identical between the two years, and the Green Bay defense was about the same too.
So what explains the seven more wins that the 2019 Packers had?
The biggest difference between these two seasons was the fact that the Packers went from the #1 most pass-heavy team in the league (67%) to a league average pass/run split team (59%).
No EPA stat could explain the difference in gameplay here, but any "old" head football coach can tell you this: If you can run the ball effectively and maintain the same offensive efficiency – you will have fewer turnovers AND have more control of gameplay (the ability to end the half with the ball or with a score).
You will also put your pressure on your opponent to do more with less time having the ball.
If you knew a possession was going to score, then how long would you want the possession to be? The answer is the entire half. You don’t want your opponent to have the ball, ever.
Let's have a thought exercise: Which is better? Scoring a touchdown with 65 seconds left for your opponent to score before the half OR score a field goal with 0 seconds left for your opponent to score before the half? Despite the fact that every fan loves to see his team punch it in - I figure those two scenarios (in a yellow wood) are analytically about the same. (I realize the irony in that I'm making this designation without having any data to support the claim, yet nonetheless I'm confident of its veracity through logic and ingrained football watching experience. i.e. "The Tape").
Time matters people. Use it. Claim it. Own it.
4th Down Attempts: 2nd
Mike McCarthy’s Cowboys are 2nd in total 4th down attempts with 13 (small sample size) and 22nd in success rate (54%). Of the three metrics I am looking at here, this one has most to do with game script.
If you have a lot of 4th-and-1 plays, then you should lead the league in 4th down attempts. For that reason, I will talk less about the Cowboys and more about the concept of 4th down attempts in general.
Everybody and their mom knows that in almost every scenario it's wiser to go for it on 4th-and-Goal from the one-yard line than to kick a field goal. Why? The expected points of going for it (~50%+ of 7 points) is slightly more than that of an attempted FG (~2.9 points or whatever).
If you miss, you gain more expected points on defense because you have the opponent backed up on their goal line with the possibility of a safety in the mix.
Analytics yay! Roll the Johnny Damon/Jonah Hill montage.
What about 4th-and-Goal from the 2? Same thing right? Sure.
What about 4th-and-Goal from the 5? Here’s where people who have no idea what they're talking about get on their high horse.
Example: Cardinals head coach Kliff Kingsbury got killed in his very first road game for settling for a field goal on 4th-and-Goal from about the five-yard line twice in a loss at Baltimore (Week 2 of 2019).
EVERYBODY was on the same group-think train here:
"Isn't Kingsbury supposed to be the young analytics guy? And he kicks a field goal on 4th and Goal? Twice?!? Guess he chickened out when he got in the big chair..."
Please. They ran six passing plays from inside the five and no one was open on any of them. If I do not know that 4th-and-Goal from the five by itself is a sketchy play - with the game playing out right in front of his eyes, then Kingsbury’s decision was a no-brainer.
Yet all of somehow 95% of people who get paid to talk about football disagreed with the guy paid to actually execute the game. It frustrates me to no end, the immediacy with which "everybody knows better" than the 32 guys hand-selected to coach in the NFL.
God knows I disagree with these guys all the time. And I voice my opinion, too. I just to try to back it up with something more than a vague assumption based on some number hidden under a desk somewhere.
If you are that emphatic about a team going for it on 4th-and-Goal from the five, then show me the work.
Show me the team-specific and non-team-specific conversation rate on 4th downs with five yards to go and explain to me why their EPA of the Cardinals going for it on 4th-and-5 was better than taking the points. Did you factor in that the Ravens were largely a running team less likely to get sacked for a safety?
I didn't think so.
I don’t see any analysis of the kind anywhere - I rarely do - instead just universal booing from the hoard of "Monday Morning Head Coaches" scattered across social and mainstream media.
Let’s take one last/recent example of a similar coaching decision much maligned by the media.
Romeo Crennel and the Texans decided to go for two, in order to take a nine-point lead, with three and a half minutes to go in the fourth quarter agains the Tennessee Titans.
This was obviously the right decision (in my opinion), and here’s where "number-crunchers" get lazy.
Elite analysts, like a Warren Sharp for example, know that matchups and team-specific factors play a huge role in these decisions. Second level “numbers guys” want to act like that stuff doesn’t matter, and that their spreadsheet will even out in the end. But there's a big 'maybe' there.
The reality is that the Texans were giving up nine yards per play and 10 yards per rush on the defensive side. Those are astronomically-poor numbers for any defense in either category, and they had a 50/50 chance to win the game on offense right in that moment. Houston had a much lower chance of stopping the Titans at any point in the game, and Derrick Henry was getting stronger as the game wore on; the opposite for Houston’s defensive line. The Texans had played zero defense up to that point, and having more than three minutes left is more than enough time for several NFL possessions.
Kicking the XP and going up 8 in that instance guaranteed nothing.
“But it would guarantee you overtime!”
As it so happened, the Texans did go to overtime. And they lost immediately. Because - again - they had no shot at that point in the game stopping the Titans and Henry.
Look, I am no expert at football or numbers - I just learned how to run regression analysis in excel earlier this morning. That said, however, I do know enough of both to realize that the amount of nonsense construed as expert opinion these days. It’s always been laughable to me, but at this point, the "non-sense" is nearing the point of tragedy, calamity and devastation.
Let’s at least try to see the whole chessboard here – this is all I’m saying. We have the tools. We have the collective IQ. Groupthink and bandwagoning at easy targets will never help anybody.
I’ll leave you with this: How many Expected Points Added did the Falcons' Todd Gurley gain by faling into the end zone last Sunday against Detroit? More than zero, right?
How many Expected Wins Added? A big fat negative number.