May 30, 2014; Miami, FL, USA; Indiana Pacers head coach Frank Vogel reacts as he walks off the court during the first half in game six of the Eastern Conference Finals of the 2014 NBA Playoffs against the Miami Heat at American Airlines Arena. Mandatory Credit: Steve Mitchell-USA TODAY Sports
With the Opening Night just around the corner, Nylon Calculus is previewing the upcoming season by taking you through some of the most important numbers to know, numbers that help tell stories about players, teams and league-wide trends.
4-25-1.
That was the record against the pointspread last season for the Indiana Pacers from February 7, through April 6. The legal betting markets generally do a very good job of staying in synch with team quality. Without much warning, Indiana’s performance level dropped so dramatically that the market couldn’t catch “down.”
It wasn’t a bad week…or a bad stretch over a few weeks. That’s almost two months of seeming imposters wearing Pacers uniforms. After starting the season as clear championship threats, Indiana was suddenly performing at a level more associated with teams expected of tanking. The best way to more fully express that is with this sequence of average victory margins…
Indiana’s first 40 games: +10.7 (fantastic for half a season)
Indiana’s next 18 games: +4.9 (Houston finished at +4.6)
Indiana’s next 20 games: -7.9 (Milwaukee finished at -8.2)
The betting markets kept pricing Indiana like a superpower when they had dropped/regressed toward being merely playoff caliber. Then, they still kept pricing them like a championship contender when they played like Milwaukee for a month (that’s the Bucks team that finished 15-67). A not-unexpected regression had finger-snapped into a collapse.
It’s not often you see the market that far off the mark with a team for so long. And, we’re probably not likely to see something that extreme in this coming season (particularly for a playoff team!). But, it’s important for the field of analytics to remember that quants and statheads were part of the market (and media) influences MISSING this story. Indiana turned into Milwaukee for a month and hardly anyone noticed until very deep in the process.
- Indiana’s won-lost record still looked great (frontloaded by a 33-7 start)
- Indiana’s “season-to-date” margin average still shone
- Indiana’s “season-to-date” stats still impressed
Concerns in the analytics field about small sample sizes leading to faulty conclusions have created a danger of letting larger sample sizes carry too much weight. Things had changed in Indiana. Hopefully someday a book will be written about all that was going on behind the scenes.
One major lesson of 4-25-1 is that the field of analytics still has to work on ways to recognize real course changes on the fly. How can we spot when teams have lost focus or intensity? How can we recognize when chemistry and camaraderie has been altered? What mix of game film and numbers make it clear that it’s time to de-emphasize a large past sample size in favor of a smaller one that’s more in synch with the current reality?
Of course, Indiana managed to settle the ship to a degree. And, they were able to grind their way through the weaker Eastern Conference for an extended postseason stay.
Frank Vogel discussed this on Zach Lowe’s podcast recently (October 9):
"As much as everybody wants to talk about the epic collapse…from February 1 on we were 21-15. That’s not horrible. That’s a pretty good winning percentage…to go 21-15 down the stretch, to me, isn’t that bad. So, even at our worst, we were still pretty good and still able to get to the conference finals."
Technically it was 21-16…and Vogel had to add an 11-3 start, pre-Milwaukee impersonation to make it look that good. At their worst, they spent the most of March and early April playing like a doormat. The Pacers did ultimately play for the Eastern Crown, but they were not the same team we saw early in the season. (Could they have won a best-of-seven from anyone in the West?)
We’ll check in periodically with developments keyed by the legal betting markets through the season here at Nylon Calculus. Pointspreads are a representation of analytics in and of themselves, as a lot of analytic approaches go into the process of determining where a line opens and closes. And, as we saw last season, they can sometimes be a key element within an early warning system for dramatic changes in team form.