should ncaab BPM be adjusted for SOS?

[ampersand5]

I didn't realize that BPM had been put out for NCAAB until this week. I've had a lot of fun playing with it, but disliked some of the results I was seeing.

Players from bad teams with poor SOS were doing very well. Obviously, their numbers are a product of the quality of competition, which skews the result.

BPM works in the NBA without regard for scheduling because every team plays in the NBA, so its all pretty consistent. However, there are massive discrepancies in the quality of NCAAB teams.

is it possible to rectify this?

[Crow]

SOS adjustment makes more sense to me for college than not having one.

( Even in for NBA BPM. Over 1.5 pt difference from easiest schedule to hardest. That is a lot, if one is trying hard to be fair and precise across the league. Celtics had the easiest schedule until 2-3 days ago.)

[DSMok1]

All versions of BPM, both NBA and NCAA, are completely adjusted for strength of schedule.

[Crow]

I should have checked before tacitly accepting that BPM is not strength of schedule adjusted.

But when I got around to it now, it took a bit more effort than it should as the BPM explanation is not currently a link offered on the BR-about page or the glossary. It has an about-BPM URL but no link from the about page.

No link http://www.basketball-reference.com/about/
No link http://www.basketball-reference.com/about/glossary.html
If you were on a team page, saw BPM and wondered what went into it, the glossary link would not currently help you. You'd have to know there was a BPM explanation page and hunt manually for it like I did. Accidental or is something up? Will this be corrected?

Link http://www.basketball-reference.com/about/bpm.html

BPM is both strength of schedule adjusted and adjusted for player tendencies when up or down on scoreboard... is that done thru two adjustments or just one? I assume it is two separate adjustments; but on quick re-read and from afar I can't 100% tell for sure, so this time I ask. Either way is it possible that doing both leads to results different than "expected", without saying if expected or the adjustment(s) are "right" or "wrong" yet, overall or especially for certain teams with low SOS?

[ampersand5]

All versions of BPM, both NBA and NCAA, are completely adjusted for strength of schedule.

I thought the factor was adjusting for being ahead/behind, my mistake.

[Crow]

You can have a strong / tough team SOS or a weak one and played ahead more or behind more. Between these two factor adjustments, if they are separate, you'd have four groups of players, one adjusted "up" twice, two adjusted up and down but perhaps different magnitudes of adjustment, and one group adjusted down twice, right? That is a lot of adjusting and difference in adjusting, worth highlighting and maybe looking at again I think or at least talking about further, even if it arises this way.

These general adjustments assume all players play better or worse when facing tough / easy opponents and team up or behind. How strong is the behavior of individual players in these directions? Do 80-90% follow these patterns or just it more like 60-70%? Regardless, players who don't actually follow the general pattern are going to get general adjustments not consistent with their individual behavior. And the application of the general adjustment may affect the accuracy of NCAA to NBA projections, especially if the team characteristics of the NCAA and NBA teams of a particular player vary from one extreme of adjustments to the other (2 up to 2 down or vice versa). For good or not. With time, this should be reviewed, on average and for particular player groups, types and individuals.

General adjustments are easier to work with, and might be sounder on average than no adjustment; but if you have to know individual performance when facing strong / weak SOS and being ahead or behind to find the general trends would there be a way (Bayesian?) to incorporate the individual data with the general trends instead of defaulting to the general trend entirely? For advanced users, especially teams, would it be better to present an unadjusted BPM, the adjusted, and how the individual players data movement compares to the general adjustment trends? Depth of description over tidiness and simplicity?

And as for the general trend for playing ahead or behind... is that SOS adjusted? Ideally should it be? Are these adjustments sound enough? Other stuff got kept out of BPM for not meeting a significance test right? Did these adjustment take and meet a similar test? Resoundly, nice enough, barely or didn't undergo such an explicit test?

[DSMok1]

The overall Adjusted Team Rating is already adjusted for strength of schedule and then I adjust it for the effects of playing with the lead. When the player ratings are adjusted so they equal the adjusted team rating, this therefore includes both strength of schedule and playing with lead adjustment.

[Statman]

The overall Adjusted Team Rating is already adjusted for strength of schedule and then I adjust it for the effects of playing with the lead. When the player ratings are adjusted so they equal the adjusted team rating, this therefore includes both strength of schedule and playing with lead adjustment.

That's how my player college player ratings are done also, just my team ratings & stat weights are a little different than BPMs. The obvious difference between my results & BPM is my ratings will rank high rebound & low assists rate guys (or vice versa) better or much better than BPM.

I don't see any problem with how college player BPM "looks" in terms of pace/SoS. The results make a ton more sense than WS, and except for the outliers I mentioned (the reb*ast outliers) correlate pretty closely to my results.