Duke does decline, objectively speaking

After Duke beat UNC last week, I got an e-mail from one of my friends containing his thoughts from the game. Included was this comment: “Scheyer, Smith and Singler playing 40, 39 and 40 minutes. Sure it was UNC on the road, but this is a trend and they will break down.”

Among Duke fans and People That Talk About College Basketball On Television (Hi Pat Forde!), it’s become the gospel truth that Duke head coach Mike Krzyzewski plays his best guys too many minutes, leading his team to collapse from exhaustion down the stretch every season.

But like Giordano Bruno, who was burned at the stake, I don’t buy the gospel truth.

Let’s break this up into its parts and approach it scientifically. (That’s right, I’m about to blind you with science. For a more in-depth discussion of my methods, click here.) 1. Duke collapses down the stretch. 2. Duke’s stars play too many minutes.

First we’ll tackle the idea that Duke collapses down the stretch.

It’s easy to see why people would believe that Duke struggles in the latter part of the season. The 2006 team—ranked No. 1 nearly all year—lost its final two conference games and then fell to LSU in the Sweet 16. The 2007 Blue Devils climbed as high as fifth in the polls before losing eight of their last 12 games, including their last four. In 2008, Duke peaked at No. 2 in the rankings, and then promptly finished 6-5.

But three years of data make up what scientists call anecdotal evidence (and TV commentators call gospel truth). So I went back and looked at Duke’s game-by-game results in all ACC games since the team’s last trip to the Final Four in 2004. Using the database at, I determined efficiency margin (offensive points per possession minus defensive points per possession) for every ACC game, including the conference tournament. I used only ACC games to keep quality of opponents relatively controlled, and used per-possession statistics so that I could compare efficiency without regard to a team’s style.

I expected to shine the bright light of “statistics” and “math” on the anecdotal idiocy around me. I thought I’d debunk the idea that Duke had a tendency to fade down the stretch.

But I was wrong.

As it turns out, Duke does have a tendency to fade down the stretch. The Blue Devils’ efficiency margin dropped, on average, by 0.01 points per possession per game. That doesn’t seem like much, but it adds up. Assuming 67-possession games (the national average), the Blue Devils declined by 13 points relative to their opponents over the course of the 19-game conference season (0.01 x 67 posessions x 19 games = 13 points). January’s double-digit win became March’s nail-biter—or even worse.

You can make the argument that as the calendar turns from January to February to March, the Blue Devils’ conference opponents gain an increasing sense of urgency—needing a big win to make the Tournament while Duke’s postseason fate is sealed. But other traditional powers like North Carolina and Michigan State have historically maintained constant efficiency margins throughout their conference schedule.

While Duke does have a tendency to fade down the stretch, that still doesn’t prove that these late-season slides are due to playing its players too many minutes.

Yes, Jon Scheyer, Kyle Singler and Nolan Smith are averaging 36.6, 35.5 and 35.2 minutes per game this season. Yes, they’re second, third and fourth in the ACC in minutes played per game. Yes, Duke has had at least two players average over 32 minutes per game in every season except one since 2005. Yes, this year’s five Blue Devils earning the most minutes (these are nominally the “starters,” though they might not start every game) play 75.4 percent of the available minutes—among the highest percentages in the country. And yes, Duke annually ranks near the top in this statistic.

But it’s not like playing more than 35 minutes in a basketball game is some sort of great feat. Forty-nine NBA players are currently averaging more than 35 minutes per game, and no one ever talks about Tyreke Evans—who should be a sophomore at Memphis this year—collapsing down the stretch. And it’s not like Scheyer, Singler and Smith didn’t know that the coaching staff expected them to put up big minutes this year. They were three of four perimeter players on the roster this year, and of course, Coach K has a history of pushing his best players to play big minutes.

Plus, with eight television timeouts plus halftime during every game, it’s not like these guys are running for-plus minutes straight anyway.

Still, there’s no escaping the fact that Duke’s top players historically play an awful lot of minutes. And, anecdotally, everyone remembers National Player of the Year J.J. Redick averaging 37 minutes per game in 2006 before struggling against LSU in that Sweet 16 upset loss.

So I compared, year-by-year, the slope of the decline in Duke’s efficiency margin to the percentage of minutes played by Blue Devil starters. As it turns out, there’s really no relationship. Even in years where Coach K has played a deep bench and kept his stars rested, his team still declined. If anything, years that Duke’s stars played more minutes tended to be more successful.

So why would this be?

Maybe, when Duke has better players, those players are counted on to play big minutes. And maybe, when Duke has better players, those players are better able to stave off a late-season decline.  

Or maybe that’s not it. Maybe Duke’s Hall of Fame coach and high-character players give their best effort all season, leading them to overachieve at the start against more talented teams that don’t turn it on until later.

But while recent Duke teams have asked their starters to play big minutes and have declined in performance at the end of the season, these two phenomena don’t seem that closely related. If anything, those teams working their stars harder have had better late-season results.

And, just in case you’re interested, despite Scheyer, Singler and Smith’s big workload, this year’s efficiency margin is trending upwards.

Trust me, it’s science.


Share and discuss “Duke does decline, objectively speaking” on social media.