Can SRS be calculated for a limited set of games?

[rrravenred]

Hi, just a question about calculating SRS for teams. Now as I understand it, SRS has an iterative component where you adjust and adjust based on team strength (which then alters the relative strength of everyone else in the set) until you arrive at the final number.

Is there a way in which you can do this for a reduced game set (e.g. missing key players or coaching changes) in a way which gives an accurate final figure? And if so, is this doable by the layman with limited statistical experience...

The reason I asked is that on the PC board, the point arose about the relative strength between the 2011 and 2012 Miami teams and as to whether 2012 was the better team once injuries were taken into account...

[mysticbb]

Well, SRS is OLS, a rather simple linear regression. In order to calculate the strength of one team only for a couple of games, you can just remove all games you don't want to have included from the whole sample. Then run the regression.

The formula would be:

MOV = HCA + a_1*tm_1 + a_2*tm_2 ... + a_30*tm_30

a_x are the coefficients (SRS of the specific teams), which will be calculated by the regression, tm_x is either 1 for the home team, -1 for the away team and 0 for the teams not playing, while HCA is homecourt advantage, which is something around 3 (I use 3.2, which gave the best fit in my tested sample). You can use a tool like R to do such thing.

That would be the exact way to do it. But we can make a little bit simpler by just using the calculated SRS of the opponents (SOS, sum of all SRS devided by the amount of games), calculate homecourt advantage ((amount of away game - amount of home games) * 3.2 / amount of all games) and calculate scoring margin in average. Then an approximated SRS becomes: MOV+HCA+SOS. That should be rather easy to do in Excel.