I was inspired by LVBruin's fanpost to determine some +/- calculations for certain lineups. I noticed that our own official athletics site provides detailed play-by-plays of each game, which I hypothesize are generated in an automated manner. I programmed a python script to parse these play-by-plays and calculate the +/- for every possible combination of our players out on the floor. Fortunately, all the play-by-play data is provided in a basic HTML page, which has been fairly painless to process. Sadly, there are a few errors in the data. I once found that the play-by-play indicated that Malcolm Lee was subbed out and he subsequently made an offensive rebound.
LVBruins said that some of his data was lost in spreadsheets, so I thought I might remedy that by writing an automated script. This is my first try at attempting a program like this, so it's possible it may be inaccurate. I did not test it thoroughly but I did note that the minutes played by all possible rotations seen during the game add up to 40. Also, it does not handle overtime situations but I can add that functionality later if it happens.
Find the stats I collected after the jump.
Below are +/- stats for every 5-man line up that played at least 2 minutes together on the floor.
It is listed as the names of the players in the rotation, the +/- per minute, the actual +/-, and the minutes played.
@ Oregon State:
Honeycutt, Jones, Lee, Nelson, Smith: 0.26, 2, 7.77 minutes
Honeycutt, Jones, Lane, Lee, Nelson: 1.4, 1, 7.33 minutes
Anderson, Honeycutt, Lane, Lee, Smith: 0, 0, 3.05 minutes
Anderson, Honeycutt, Lee, Nelson, Smith: 0.0, 0, 2.93 minutes
Jones, Lamb, Lane, Lee, Smith: 1.34, 3, 2.23 minutes
Anderson, Honeycutt, Jones, Lamb, Nelson: 0.0, 0, 2.22 minutes
Anderson, Honeycutt, Jones, Lee, Smith: 1.32, 13, 9.87 minutes
Anderson, Honeycutt, Jones, Lee, Nelson: -0.31, -2, 6.42 minutes
Anderson, Honeycutt, Lamb, Lee, Smith: 1.81, 5, 4.23 minutes
Anderson, Honeycutt, Lane, Lee, Smith: 0.0, 0, 3.7 minutes
Anderson, Honeycutt, Jones, Lane, Lee: -2.39, -8, 3.35 minutes
Honeycutt, Jones, Lee, Nelson, Smith: -1.69, -4, 2.37 minutes
Honeycutt, Jones, Lane, Lee, Nelson: -1.85, -4, 2.17 minutes
Naturally the +/- in the Oregon game is higher for lineups without Reeves because he was scoreless that night. The stats look pretty awful for the game against OSU though, thanks to our terrible second half.
Let me know if you have any suggestions or ideas on how I can improve the data collection process. Should I present it visually like LVBruin did? Maybe you guys would like to see an aggregate +/- for every combination over the course of the season so far? That'll take a bit more time on my end to code up, but right now, I can provide +/- for the lineups seen in individual games. I could also redo the program in Java to I can make a standalone application for other users to download and use, which would take some time but would likely be good practice for my career.