Q&A: Statistician Jeffrey Rosenthal on why the math points to a Raptors’ victory—and why Drake gives us a better shot
At the beginning of the NBA finals, U of T statistics professor Jeffrey Rosenthal created a mathematical model to predict the Raptors’ chances of winning this year’s NBA Finals. We spoke with Rosenthal about the frailty of human expertise, the (statistically demonstrable) Drake factor and why the odds are increasingly—89 per cent (!!!)—in our favour.
Going into the finals, you gave the Raps a 51 per cent chance of victory. I presume our odds are better now?
They sure are. Before the finals started, the Raptors and the Warriors were quite evenly matched. I compared their home numbers with their away numbers, and the home court advantage was just enough to push the Raptors over the edge. Obviously a lot has happened since, so now it’s looking like the odds are 89 per cent in favour of victory.
How does your model work?
It’s a fairly simple model based on performance over the course of the season. Each time a team wins, their odds go up. When they lose, their odds go down. The Raptors’ odds of victory were at their lowest during the previous series against Milwaukee. After they lost the first two games, their probability of winning the series went down to 19 per cent.
Do your odds factor in variables, like whether or not Kevin Durant plays?
No. There are other models that are a lot more detailed, but those can be tricky for a few reasons. First, you don’t know if and when an injured player is going to come back. Second, when a player comes back after an injury it’s not always easy to predict the outcome, since he might not be fully recovered.
Before the finals, almost all the sports experts were predicting the Warriors would win, while most statistical and AI models were predicting Raptors. What are these machines seeing that human experts can’t?
It’s not just the human experts; the betting markets in Vegas were also heavily in favour of the Warriors. The statistical models aren’t necessarily picking up things that humans don’t, but they’re also not vulnerable to factors that can influence humans—for example, the fact that the Warriors have won a number of championships. That thinking isn’t totally misguided, but you can’t really use it to compare to this season, since a lot of key Raptors weren’t even on the team last year. My model looks specifically at this season and the players who are playing now. Models don’t understand things like branding or reputation. Sure, the Warriors have a really great brand: they won all these championships, they have all these stars, and people think, “Of course they’re going to beat this team from Toronto that nobody’s ever heard of.”
Do they understand the Drake factor?
That’s an interesting question because it relates to the home court advantage, which increases the odds of winning by about 10 per cent. There are many reasons for that: the players might get a better sleep in their own beds, the court is familiar, and the crowd is cheering for you. Drake psyches up the home crowd and trash-talks our opponents. It’s all kind of silly, but it does make a difference.
Can statistics take into account things like will and heart?
Those things are hard to measure and they cut both ways. For example, the Warriors have experience, but maybe they aren’t as excited because they’ve already won a number of times.
A hand can only hold so many championship rings.
Can stats predict whether an individual player will have a great game or a lousy game?
It’s not easy, since there are so many factors. This year, for example, Fred VanVleet rushed back to Illinois for the birth of his son between games 3 and 4 of the Eastern Conference Finals. People thought, “Oh, he’ll be off, he missed a night’s sleep.” And then he came back to shoot the best three-pointers of his career.
All thanks to little Freddie Jr.
Another interviewer said the same thing, but his response was, “Oh, no, this was all me!”
Are you a Raptors fan aside from all the number crunching?
I was only a mild fan before playoff fever hit. Now, I’ve jumped on the bandwagon like everyone else.
Have you placed any bets?
I haven’t. I guess it would seem like a smart thing to do. I’m just not much of a betting man.
Where will you watch Game 5?
From a resort on Vancouver Island. I booked the vacation well before I knew the Raptors would make the finals. I was invited to do an interview today from the middle of Jurassic Park, which would have been pretty cool. Maybe next year.
Do you have odds on the Raptors making it again next year?
I did look at how often the Eastern Conference championship winners were back in the finals the next year. The chances are around 47 per cent, which is pretty good.
This interview has been edited for length and clarity.