Pete Carroll vs. the challenge flag, Part 2

In Part 1 we learned two key lessons:

1) Pete Carroll is willing to risk a single timeout on a whim, wins 45 percent of his replay challenges, and yet remains extremely competitive with his NFC West peers since joining the Seattle Seahawks in 2010, because...

2) Good coaches didn’t usually win more than half their challenges either. Jim Harbaugh and Sean McVay scored lower than Carroll; Kyle Shanahan and Bruce Arians higher.

Based on the second discovery, I threw one theory into the ring on Monday: success with the red flag is not connected to overall football game success. It may feel like a doomed challenge wrecks a drive and thus contributes heavily to a loss, but the greater picture doesn’t yet suggest that correlation. That’s why I expanded our sample to the entire league, all active head coaches, to test the theory.

Remember Carroll’s row of digits? One more time, for those in the back, then:

I want to highlight, in advance of the numbers behemoth below, the 111 and 6.5 columns. Those are going to matter more than I thought when embarking on this journey.

Deep breath. 29 active coaches have thrown the red flag at least once. Sorted by challenge win percentage:

(Note: three coaches are rookies and have not yet tasted the thrill of officially telling officials just how wrong they were. That’s why only 29 men are listed.)

Nine takeaways and two questions lie ahead. Plus your insightful (?) comments. The time for over-analysis is upon us.

Takeaway 1: Uh oh. Carroll is 18th on the list. With only 11 coaches behind him. Maybe he is below average.

Takeaway 2: Not so fast. Sample size matters. There’s a lot of noise at the extremes.

Eliminating everyone with less than 20 challenges results in a much less noisy chart, full of experienced coaches who’ve all won something, be it Super Bowls, conferences, or multiple division titles. You can find all these guys in the middle, which hints that regression to the mean is a strong force in challenge flag statistics.

Anyway, in this new 17-coach sample, Carroll ranks 9th.

Go with 30 or more challenges, and Pete is 6th of 12.

Make the bar 50 challenges, and you’ll find him again in the middle, 4th of 8.

And as we saw yesterday, he’s right near the average again with the winningest active coaches, so when you increase to 100 throws, you guessed it, Carroll is 3rd of 6.

Accounting for small samples, he’s at the median or better. Most veteran coaches are worse than him at challenging plays.

Takeaway 3: How about those bottom five coaches. That’s a real list.

It’s one GOAT, one promising but unproven wunderkind (McDaniel), one enigma (Smith), and two coaches any GM would immediately hire if available (McVay, McDermott).

It sure doesn’t look like being bad at challenges automatically costs you very much when those are the worst five.

Takeaway 4: Not exactly a Murderer’s Row of coaches at the top, either.

Correct. Only eight active coaches have a career win-loss record under .500, and three of them are among the best challengers. Saleh and Bowles have won more challenges than they’ve lost but are a combined 45-73 in actual football wins and losses.

Takeaway 5: Imagine Taylor and McDermott on the same staff. That roster would go 15-2 and lose almost all their challenges. I joke, but no better way to illustrate how high-end talent can pretty easily overcome a coach’s occasional miscue.

Question 1: What role might the recent replay rules play, in which every score and change of possession is automatically reviewed? Didn’t used to be like that.

That’s a real wild card. I am super curious to see how that discussion develops in the comment section, because I can see effects in multiple directions.

Takeaway 6: Some coaches are timid, others not so much. You see everything from three challenges a year to eight-plus.

The data has led me to wonder if some guys are valuing the sheer volume of challenges won (individual large rewards) over percentage. It’s entirely possible they have calculated internally that one challenge victory is worth more than one timeout lost, so they can afford to gamble.

Because they toss the flag eight times a year, tops in the league, Payton and Harbaugh have each won 54 reversals for their team. They’re exchanging many timeouts in favor of three or four challenge wins. Unlike a lot of the people at the top of the chart, who challenge with half the frequency.

By the way, Carroll’s in the very next tier at 6.5 yearly challenges, still well above league average, with only four men ahead of him.

Takeaway 7/Question 2: Man, is there any football thing Andy Reid’s not good at?

No.

Takeaway 8: McVay and Tomlin are twinsies.

• One SB win and one SB loss apiece, three division titles each in their first six years
• Same .419!
• Same frequency of challenge
• Same career W-L record above .600

How cute! Good thing we already love the Steelers and Rams so much around here.

Takeaway 9: I just don’t see a correlation. Not on the surface. Between winning challenges and winning games, that is. If Belichick’s .406 challenge win percentage wasn’t enough to convince, the younger generation’s numbers will clinch it. McVay sits at .419; McDermott’s a lousy .259; everyone’s darling Taylor manages a putrid .355; McDaniel has prevailed once in six tries. Sirianni and LaFleur and Shanahan may all appear gifted with the red flag in the early going, but let’s see the Eagles head man sustain it, LaFleur has a lot to prove with Aaron Rodgers gone, and Shanny was still carrying a W-L record under .500 this time last year.

So yes, you’d like to Carroll to make fewer doomed game-day challenges, I’d like that too, especially since you’re never going to win a spot-of-the-ball challenge — but in the end it’s in his DNA to go on impulse sometimes. I wouldn’t want him any other way because he wouldn’t be himself otherwise, and this version of himself wins a lot of football games. Especially since real wins and losses do not look driven by challenge proficiency.

And he’s not even bad with the flag anyway; once you cancel the noise at both ends of the spectrum, he’s slightly above average.