Nathan Sliver here with sevenfortyseven.com. In the PBSL we get a lot of great trash talk between GMs. We also get a lot of great critical media articles. Lately I have been wondering if I could measure how good a coach a GM is. To do so I started with my dataset of every PBSL team ever (excluding last year). I added a few variables to the arsenal. I recorded the GM for the first time ever and I calculated the record of a team's conference minus the team's wins against the other conference as a measure of strength of schedule. I also calculated the sum of each player attribute for the eight players that played the most minutes for a team in a given year. I then used a stepwise regression (minimum BIC) of these variables to determine the best set of control variable with wins as the dependent variables. I then added a dummy variables for GMs. Xist ended up being the control because the program defaulted to alphabetical order and I'm not really sure that makes sense but I didn't think about that until now so I'll let 42 tell me why having Xist as the control either screws it up or doesn't matter. Anyway, here are the results for my control variables.
- Control Results.GIF (7.98 KiB) Viewed 1890 times
You could look too much into this and try to figure out which attributes lead to success but I'll just say that I'm happy everything is heading in the right direction. The residual plot looks good and an adjusted r squared of .736 is encouraging.
- Residual plot.GIF (19.68 KiB) Viewed 1931 times
And now, for the moment you all have been waiting for. Unfortunately most of the results are not significant but I think we can still get a good idea of who is getting more out of their players than other GMs.
- Coach Table.GIF (29.81 KiB) Viewed 1892 times
Obviously Snuglife struggled. . . A lot. 10 fewer wins a season because of a coach's struggles is rough. I was really surprised that kucoach was significantly positive after his recent stretch of mediocre results with loaded teams. I don't think anyone is surprised to see who is at the top of this list. I always thought Bow got way more out of his teams than others could have and the same goes for Darth, FTB, Balls, and Nola. No offense to Aburns but that surprised me. Congratulations, your coaching is apparently worth 7+ extra victories a year. Obviously there are some gaps in the analysis but, as I said, I think this provides some fairly interesting insight. Let me know what your think.