RealGM

Strength of Schedule I mean. I'm asking for the purpose of calculation of SRS. I know that BBR says you can take the average SRS of the opposing teams but just for curiosity's value, that can't work, can it? I mean, you need to calculate SOS first to get SRS, so it's a paradox. It may be a stupid question but it's driving me up the wall.

SRS is based on the regression on the team results for a given season (ordinary least square is used, but ridge should pretty much give the same result). Therefore, SRS is per se adjusted for the strength of schedule. The coefficients of the regression are giving you the SRS for each team, while the intercept gives you the HCA for the respective season.

SOS = MOV - SRS

where MOV is the average scoring margin for the respective team.

Now, in order to determine the SOS for a given sample, you can easily take the average SRS of the opponents and adjust that for the HCA (which is about 3 to 3.5 for the last 30 years).

If you want to make it accurate, you could run a regression on a specific sample. For example: You want to know how good teams played in January and February? Take all games during that time and run the regression in order to determine SRS and SOS.

It' can seem like a paradox but this is how it works. You basically just do multiple runs, always updating your "current SRS". In the first run everyone's "current SRS" is just their point differential. From that you can easily compute SOS. Now you compute each team's new "current SRS" using point differential and the SOS you just computed. You then compute SOS again.
And so forth, until everything's stable

Well, using matrix algebra makes more sense than going through such loop ...