Tony Dungy shares some views on analytics

Getty Images

In recent years, a growing number of coaches have been making in-game decisions based more on math than other real-time factors that potentially influence football decisions.

On Thursday night, Colts coach Frank Reich decided to go for it on fourth and seven with three minutes left from the Houston 47 while trailing by three and holding two timeouts (he burned one before the fourth-down play). On Friday morning, Hall of Fame coach Tony Dungy addressed the decision.

“That’s why I have a problem with analytics,” Dungy said on PFT Live regarding the decision not to punt and play defense. “Well, percentage-wise this is gonna happen, and this is what you should do. . . .  I think Frank Reich was watching this game and he saw his defense really struggling to contain the Texans offense, and I don’t think he felt good about getting the ball back [if he’d punted], and that’s what happened. After they went for it on fourth down, they had their timeouts, needed the stop, and couldn’t get it. I think it was really based on what he had seen by Deshaun Watson and that offense that he went for it and they just didn’t come up with it.”

Dungy elaborated on the intersection between analytics and the realities of a given game.

“You’re sitting there and you’ve got a chart that says 90 percent of the time on this fourth down and one you’re going to make it and win the game,” Dungy said. “Okay, yeah, but you might be playing against the ’85 Bears. Or you might be playing against a team that doesn’t have any offense that you can kick the ball to them and you feel very confident that you’re going to get the ball back in two minutes. So it depends on what’s happening in the game, how you’re playing, all of those things — field conditions, weather. To me, you just can’t go by a book and say, ‘This percentage of the time, X happens so I’m going to do that.'”

He’s right. And that’s not an anti-analytics take. It’s a pro-reality take. The percentages are what they are, and the percentages are shaped by the many unique situations in which paast teams made decisions that did or didn’t work based on factors completely and totally unrelated to the math.

The best coaches are aware of the numbers, and they use them as a factor in each decision that is made. And for the same reason the numbers shouldn’t be ignored, they also shouldn’t be the only factor.

14 responses to “Tony Dungy shares some views on analytics

  1. Had the Texans defender not grabbed Brissett’s ball hand while he tackled him, that’s a first down! Good play by the defense and almost a great play by Brissett. I say Frank Reich made a good call that just came up short. That, and credit to the defender.

    This is way more fun that talking about reviews, play clocks, etc.

  2. There is a misconception about the use of analytics in general and specifically in sports. Data and analytics are used to inform our decisions, not make our decisions. If used correctly, they are a valuable tool in both business and sports.

  3. The whole idea of analytics is to stop punting 4th and 1 on the other guy’s 40.
    Joe Gibbs said he was terrified going for it it with Riggins in the Super Bowl.
    What it comes down to is when a team fails and loses, coaches get fired if they went for it, and nobody remembers what happened when they punt.

  4. Analytics is about tendencies and cumulative outcomes over time. It is not at all about likelihood of an individual outcome in a specific part of a specific game.

    There are so many unique variables at work in any given game situation that analytics can’t possibly be responsibly used as the justification for any individual decision. It just helps give you a framework for overall strategy, which is essentially the toolbox you have available when making decisions. The tool you pick is what’s best for what is in front of you in the moment.

    I think Reich made a smart choice based on what he could control. The bad outcome was about what he couldn’t control: the element of chance in the flow of play. Which is what gives the game its life and excitement.

  5. We have computer programs that have re-created our actual universe, complete with a simulation of the physics and chemistry of its expansion…

    but Dungy doesn’t believe today’s computers accessible to a billion-dollar industry could possible include input of Jeff Driskell vs Aaron Rodgers as the opponent, when calculating all possible outcomes, and how often each will occur.

    Sigh. Incredibly fortunate that he made the HOF. He was the Jason Garrett underwhelming playoff coach of the 90s.

  6. Reminds me of when a manager pulls a pitcher late in a game he is dominating but may have a runner in scoring position with a 1 run lead, then proceeds to lose by 5.

  7. I’m not a fan of analytics. To me it ruining sports. Sports to me is great because there is a ‘human factor’ to it. Sports are played by human beings, coached by human beings, and officiated by human beings. Analytics is designed to take out that human factor. Taking out that human factor, is relying on a computer to do your thinking for you. People over 40 like thinking for themselves. People under 40 have always relied on a computer doing the thinking for them. It’s a generational thing. You guessed it, I’m over 40.

    To me it’s BS to say analytics just makes suggestion on tendencies. Too many coaches are afraid to go against the analytics, and several coaches simply use analytics for everything they do. When it comes to a situation that is not covered, they are simply lost and have no idea what to do. I can name more than 1 coach that does this.

    I think it was back in the early 70’s when somebody fed a bunch of statistics in a computer about different eras of football teams. They created Super Bowls of teams maybe something like the ’64 Chiefs and the ’56 Bears and several other. They were on radio. I listened to the first game. The though of legendary NFL team against each other was interesting at first. Then the thought it’s wasn’t real kept creeping in my mind. It was just a computer fantasizing about a ‘what if’ game. It wasn’t real. It didn’t have a human factor. I never listened to another game. Apparently I wasn’t the only one who didn’t care for a computer playing a fantasy football game. Computerize football didn’t last long.

    You could make a case for analytics could be a useful used only a guide for tendencies only, but it’s not being used that way. It’s became the way you play the game according to a computer. Give me the human factor every time!

  8. Very good analysis by Dungy. But that being said analytics is the future. Remember it’s a percentage based on the data. If it’s not taking into account who you are playing (i.e. the 85 Bears) then you shouldn’t be using it. Every game that is played provides additional data and increases the accuracy of the percentage. But the coach still needs to put it in perspective and if the 1, 5,10, etc percent chance of not making it is not worth the risk then you shouldn’t be chancing it.

Leave a Reply

You must be logged in to leave a comment. Not a member? Register now!