Revisiting Big Ten Basketball Preseason Predictions
Purdue basketball completed a truly remarkable turnaround after a 5 loss nonconference slate, finishing 12-6 in the Big Ten and securing Matt Painter’s first 20 win season of the post-Baby Boilers era. That was a really fun sentence to type. But it had me thinking back to preseason predictions of the Big Ten conference. I distinctly remember that most analysts didn’t think much of Purdue, despite the return of AJ Hammons and the addition of two four-star recruits in Isaac Haas and Vince Edwards to an already-deep roster. In their defense, the talent at the top was lacking…but it seemed like Purdue was being hugely overlooked by many outlets. We at BS are the skeptical Purdue fans, and even we had Purdue slated to finish at around 7th in the Big Ten.
So, with endless thanks to Inside the Hall, I looked back at a bunch of preseason predictions and wanted to know where they went wrong. Also thanks to StatsbyLopez, a great sports stats blog, for spurring this idea with his NFL Pundit Rankings.
Rankings were pulled from from 10 publications/websites linked below. (Again, thanks to Inside the Hall for some of these.)
Sporting News, Lindy’s, Athlon Sports, Blue Ribbon Yearbook, ESPN Insider, ESPN’s Big Ten Coaches Poll, RealGM, BT Powerhouse, BTN.com Beat Writer’s Poll, and CBSSports.com.
Below is how each publication ranked Purdue before the season began. Warning: it’s ugly.
Not great.
Common concerns that were voiced included the team’s lack of any senior leader (most were published before Jon Octeus graced us with his presence), AJ Hammons’ inconsistency, Coach Painter’s inability to forge a post-Hummel/Johnson/Moore identity, and the depth of the Big Ten.
But…those concerns were largely debunked by many of their own predictions, with most forecasting All-Big Ten Second/Third Team and Defensive Player of the Year honors for Hammons (which was pretty close to spot-on) and a breakout season for a wing like Kendall Stephens (not as much). It seems like most places believed in the individual pieces, but were skeptical that Painter could put it all together.
RealGM explained it best: “When you get down to the thirteenth best team in a conference, you are often talking about a terrible team. Purdue isn't terrible; the Big Ten is just deep.”
Now, the conference wasn’t as good as most initially thought, but the talent differential from 2-13 (if we exclude the outliers Wisconsin and Rutgers) ended up being pretty close.
Here’s a complete compilation of each prediction (on the right), contrasted with how things really turned out (on the left).
Every outlet I looked at got Wisconsin and Rutgers right, but those weren’t too hard to predict before the season. To visualize how each team performed compared to their expectations, I calculated the Mean Absolute Error for each team…or, the average difference between each publication’s projection and their final Big Ten ranking. Here’s what we have, arranged in from largest to smallest MAE order:
Purdue’s #1!!! Finally!
So this just means “experts” screwed up Purdue’s ranking the most, and got Wisconsin and Rutgers’ spots on the nose. Without scaling anything, we see that Nebraska and Maryland performed very differently when compared to their preseason expectations, and Michigan/Ohio State/Minnesota varied enough to surprise a few pundits.
I also calculated the Mean Squared Error, to highlight anywhere the projections were off by a relatively large margin. Here’s what we have, arranged in order of the Big Ten’s final standings:
This makes the takeaway a lot clearer: Pundits thought Maryland and Purdue would be terrible, while Nebraska would be very good. Turns out, not so much. The differences seen with Ohio State, Michigan, and Minnesota do not seem as significant as it looked, especially considering the middle of the Big Ten has historically been fairly muddled.
Now, I’m not calling for anyone to be fired over this. I’m not even calling for anyone to be given twitter-hate over this. It’s more just to illustrate how stupid preseason rankings are. Some significant changes are made that predictions cannot possibly account for, and in Purdue’s case I would point to the addition of Octeus as a huge reason for the upswing in performance.
The other obvious change: injuries, suspensions, and dismissals of high-impact players. Here’s a list (as complete as I could make it) of legitimate Big Ten players missing a chunk of games:
There’s no way to predict something like that. Illinois and Michigan (and maybe even Michigan State) should have been significantly better, but they had injuries to key players that derailed their seasons. Michigan’s injuries were particularly brutal, as LeVert and Walton Jr were both All-Big Ten level players. Purdue, meanwhile, is a pleasant surprise in conference play partly because our Boilermakers didn’t suffer any injuries all season.
Expectations are often adjusted based on these meaningless preseason predictions. More times than not, each team competes more with their preseason expectations more than opposing conference teams. That can be problematic, especially when you’re using these expectations to color an important decision…let’s say, an NCAA tournament berth.
Illinois looks to suffer the most from expectation comparisons…but losing Rice, Cosby, and Abrams for significant chunks of the season will be lost on voters that might not have paid enough attention to the news coming from Champaign. Instead, all they’ll see is a .500 anonymous Big Ten team that hasn’t generated one headline all season. Thus, NIT.
Purdue and Maryland, meanwhile, gets major boosts for wildly outperforming minimal expectations…even though both teams are still fairly flawed. Unless something drastically changes, it looks as though voters will like this Purdue storyline enough to grab a 10-seed, even though resume-wise Illinois and the Boilers are much closer than we want to admit.
Long story short: down with the click-drawing preseason prediction posts.
(They’re still super fun to read, though. Dammit. I’m part of the problem.)
