At this point I only have limited faith in the model (despite decent performance last year, out-pacing FiveThirtyEight). The baseline predictive models could be improved, which would lower the variability in the simulations. Other forecasters are moving to more nuanced models while mine is still based on BPM.
Gimme them predictions
Fetch the CSV from github.
|Team||Projected Wins||2017-18 Wins||Improvement||Playoff Probability|
|New York Knicks||26.2||17||9.2||9%|
|Team||Projected Wins||2018-19 Wins||Improvement||Playoff Probability|
|Golden State Warriors||55.9||57||-1.1||91%|
|Los Angeles Clippers||47.8||48||-0.2||74%|
|Portland Trail Blazers||45.2||53||-7.8||62%|
|Los Angeles Lakers||40.6||37||3.6||46%|
|San Antonio Spurs||38.9||48||-9.1||41%|
|New Orleans Pelicans||38.1||33||5.1||36%|
|Oklahoma City Thunder||38.0||49||-11.0||37%|
- Interesting to see GSW as the favourite, somehow, yet again. Steph Curry is very, very good. BPM likes him - but it also likes Draymond, Russell, Cauley-Stein, and Looney. The model also has some preference for prior season performance.
- Although everyone (including my own intuition) is forecasting Milwaukee and Philly to dominate the East, the regular season standings might not work out that way. Toronto and Boston have a lot of talent, and Milwaukee might regress a little bit. BPM isn’t high on any Milwaukee players other than the Giannis Antetokounmpo. They were great last year, but they were also healthy. Philly is a strong team, but it’s a top-heavy one, which makes them vulnerable to injury.
- The regular season might not be as sunny as forecasted in LA. The Clippers are strong, but Paul George and Kawhi Leonard are not projected to play ironman minutes. The Lakers have some players on the wrong side of the ageing curve.
- Balance. Most of the league has at least an outside shot of making the playoffs or missing the playoffs. This should make for some exciting basketball!
If I were a betting man…
I would hesitate to bet on this model. This year I’m shadow betting: calculating optimal bets based off of the model, so that I can evaluate its performance after the season. I took over-under odds for win totals from one site and calculated average expected returns, according to the simulations.
I came up with three NPV-positive OVER bets:
- Charlotte over 23.5 wins
- Golden State over 48.5 wins
- Washington over 26.5 wins
And four UNDER bets:
- Chicago under 31.5 wins
- Lakers under 51.5 wins
- Milwaukee under 57.5 wins
- San Antonio under 46.5 wins
Charlotte doesn’t have many great players to work with, but 24 wins isn’t a big number to reach, and they’re in the junior conference. Same goes for Washington. Washington might be more path-dependent than Charlotte, in the sense that they have more talent to trade away if they decide to be less competitive.
I love the Spurs, but 47 wins is a lot. 58 is a lot for Milwaukee too - that would be nearly no regression to the mean, despite losing Malcolm Brogdon.
I’m looking forward to watching this season. It’s a year-long celebration of the Raptors’ championship!