When Mike McCarthy was hired as the next head coach of the Dallas Cowboys, it came with the promise of an analytical approach to the game as a means to reinvent himself. It took some time, but McCarthy did build out the analytics team he had talked about, and the Cowboys’ thrilling comeback win over the Falcons last year demonstrated that he did indeed know what he was talking about when it came to analytics.
Ironically, the day before the Cowboys’ season opener in Tampa Bay, the analytics team at Next Gen Stats dropped an article breaking down the head coaches who made the best decisions as it relates to fourth down and two-point conversion attempts. And while Browns head coach Kevin Stefanski and Ravens head coach John Harbaugh ranked significantly ahead of the pack, McCarthy sat there in sixth place just slightly behind Colts head coach Frank Reich.
However, Dallas had one of the worst fourth-down conversion rates last year, though that was largely due to all of the injuries on offense. This article from Next Gen Stats suggests that McCarthy was indeed making better decisions with analytics in his toolkit.
Which is why McCarthy’s decision making Thursday night was so confounding.
There were three different occasions where the Cowboys settled for a field goal in spots where they shouldn’t have. Before we get ahead of ourselves, the Cowboys’ final field goal was not one of these occasions. The obvious sentiment there is that Dallas had to eat up more clock because everyone knows exactly what Tom Brady is going to do with that much time left. But the offense stalled out, they were in field goal position down two, and had just two timeouts on the wrong side of the two minute warning. It’s a bad situation to be in, but kicking the field goal was the right call there.
Now, onto the mistakes. The first one came early in the second quarter, and it resulted in Greg Zuerlein’s first missed field goal. Dallas had the ball on fourth and three on the Buccaneers’ 13-yard line. A cursory knowledge of the Bellman Equation suggests to go for it already, since the result of failing to convert gives the other team the ball with a very long field and backed up close to their own goal line. Ben Baldwin’s fourth down decision bot suggested to go for it with a medium level of confidence.
---> DAL (7) @ TB (14) <---— 4th down decision bot (@ben_bot_baldwin) September 10, 2021
DAL has 4th & 3 at the TB 13
Recommendation (MEDIUM): Go for it (+1.6 WP)
Actual play: G.Zuerlein 31 yard field goal is No Good, Wide Left, Center-J.McQuaide, Holder-B.Anger. pic.twitter.com/XSkuwR2pGv
As you can see from the chart, the Cowboys win percentage was calculated to have no difference between failing the fourth-down attempt and failing the field goal attempt, but jumped seven percentage points with a successful fourth-down conversion attempt. Instead, McCarthy opted to kick, and ultimately got what he deserved: zero points.
The second mistake was easily the more egregious. The Cowboys were set up with fourth and goal just three yards out from paydirt. They were down five and about a third of the way through the third quarter. McCarthy kicked and Zuerlein made it, which cut the deficit from a one-score game to a one-score game. Big whoop. Any guesses on what the Baldwin Bot said?
---> DAL (16) @ TB (21) <---— 4th down decision bot (@ben_bot_baldwin) September 10, 2021
DAL has 4th & 3 at the TB 3
Recommendation (STRONG): Go for it (+4.2 WP)
Actual play: G.Zuerlein 21 yard field goal is GOOD, Center-J.McQuaide, Holder-B.Anger. pic.twitter.com/as9qhH5ovw
Correct. A strong recommendation of going for it with a positive swing of 4.2 units of win probability. The crazy part here is the win percentages split by result: a successful fourth-down conversion made the Cowboys 18% more likely to win than a successful field goal. When dealing with win percentages, 18% is a pretty good swing. Also it’s only three yards for a touchdown! Why not trust the offense, the only part of the team that was consistently playing well all night?
The third mistake McCarthy made was attempting that 60-yard field goal that inevitably fell short. There’s a certain logic in trying to find points in unexpected ways, but there’s also an inherent danger in giving Brady the ball at the 50-yard line with any amount of time left. For the record, the bot suggested punting there, but it wasn’t an overwhelmingly strong case. The fact that the Buccaneers failed to capitalize mitigates the blunder, but a blunder it still was.
All in all, this was probably the worst we’ve seen McCarthy in terms of decision making since he came to Dallas. Plenty of fans will point to certain failed conversion attempts last year, but many of those were justified decisions that were let down by the men on the field. Thursday night was the exact opposite: the men on the field (meaning Dak and the offense) weren’t letting anyone down, but McCarthy repeatedly ignored the analytics and fell into old habits. The result was a loss.
It’s a potentially troubling development, especially if it continues throughout the year. Thursday night’s game proved the Cowboys can hang with the best in the NFL so long as their quarterback is on. The defense and special teams will get there with time, but if McCarthy isn’t able to carve out an edge with his decision making then things could get ugly.