Support and Data
I have really found tha contacting support at GitHub is useful. I state my problem and send screenshot of what is going on and they normally get back to me within 24 hours.
What data is available? Their is data on each individual game that I have access to.
What does that data look like? The data is clean. It is stats for Chris Mullen who is a Hall of fame basketball player. It shows his points, assists, rebounds, field goal precentage, etc. All of this is important to determine good or bad performance.
How assumptions can we make about that data? I think that the data can help me determine how good Chris was. My initial thought was to try to figure out what made his best year his best year and why his performance went down and what were the determining factors. I know he played for Run TMC that got their name from Run DMC. They were bad boys! But I think I may see if some players although they are not in the great 50-40-90 club that they may still shoot it better and may just have bad stretches or a stretch when they were hurt that takes them out of that elite club. I remember I was averaging 20 in college and then I wasnt feeling well one game and I new I couldnt play but the coach tried to put me in anyway. I played like 2 minutes and had 0 points. It hurt my average a bunch!
How additional data might be useful to validate our assumptions? I am not sure right now but I will continue to think about it.
How can we turn our assumptions into a model? I think we can just find the top stretches of performance and determine how well they do when they are on and shooting well. Then look at their performance when they are off and shooting bad. Then try to determine if Mullen is a better shooter than one of the 8 players that have shot 50-40-90. Chris almost shot 50-40-90 for his whole career. He was over 50, shot almost 39 from the 3 point line, and shot 87 percent from the free throw line for his career!
That is really really impressive.