RealGM

Hi, I've been a lurker on this site for over a year and finally decided to make an account and post. I've been studying up on SRS and trying to understand how it works, and it seems to underrate defensively slanted teams. In a blog post from Mike Lynch, he writes "(SRS) is slightly biased toward offensive-minded teams - because it considers point margins instead of point ratios, it treats a 50-30 win as more impressive than a 17-0 win." Do we have any sort of way to adjust for this, like giving a team a slight increase in SRS for having a lower defensive rating? Thanks.

From what I recall SRS = (Points) Margin of Victory (MOV) adjusted for Strength of Schedule (SOS).

You could try to use Points Ratios instead, but how do you handle scorelines containing 0, for example the 17-0 you mentioned. This shouldn't be an issue in basketball though.

You could use %score difference, I guess.

Rishkar wrote:Hi, I've been a lurker on this site for over a year and finally decided to make an account and post. I've been studying up on SRS and trying to understand how it works, and it seems to underrate defensively slanted teams. In a blog post from Mike Lynch, he writes "(SRS) is slightly biased toward offensive-minded teams - because it considers point margins instead of point ratios, it treats a 50-30 win as more impressive than a 17-0 win." Do we have any sort of way to adjust for this, like giving a team a slight increase in SRS for having a lower defensive rating? Thanks.

You could multiple the SRS by INVERSEpace.

Take the 2023 Cavaliers for example. They played at a 95.7 Pace in a 99.1 Lg Avg Pace.

((LEAGUE Pace)/(TEAM Pace))*SRS
((99.1/95.7)*5.23)

This changes their SRS from 5.23 to 5.42...Maybe not the difference you were looking for, but accounting for Pace would account for MOV.

But, circling back to your original point, is a 17-0 win more impressive than a 50-30 win or a 100-80 win?

Thank you all for the suggestions, I'll play around with it

Rishkar wrote:Hi, I've been a lurker on this site for over a year and finally decided to make an account and post. I've been studying up on SRS and trying to understand how it works, and it seems to underrate defensively slanted teams. In a blog post from Mike Lynch, he writes "(SRS) is slightly biased toward offensive-minded teams - because it considers point margins instead of point ratios, it treats a 50-30 win as more impressive than a 17-0 win." Do we have any sort of way to adjust for this, like giving a team a slight increase in SRS for having a lower defensive rating? Thanks.

I don’t know if I’d consider a 100-80 win more impressive than a 120-100 win personally though

This shouldn’t be an issue in the NBA because the realistic lower bound for scores is not close to 0.

I actually agree with OP. Obviously the lower bound isn’t 0, and we’re not going to be looking at dramatics SRS changes, but a point differential based formula which does not account for pace is notable, especially in an era where rate based statistics are so popular even among casual fans. And yes, a 100-80 win is objectively more impressive/ meaningful from a statistical standpoint than a 120-100 win— you’re scoring 125% of your opponent’s total vs 120%, and in a slower paced game, it’s harder to win by 20 than in a faster paced game. Of course, not all defensive teams play at a slow pace. Granted, it’s hard to adjudicate the “impressiveness” of any two wins without a lot of context, but over the length of a season, I think it’s absolutely a salient distinguisher.

Point% seems to me to be the most effective way to evaluate in this context. An inverse pace formulation would show the merits of your classic Spurs slow it down and play iron defense, but it wouldn’t account for the sort of run and gun teams which are actually much better defensively than offensively (like many young teams with great athletes who aren’t great team offensive players). Those teams may in the end have lower scores even though the pace is higher, just because both offenses will be inefficient in those games.

acheck13 wrote:I actually agree with OP. Obviously the lower bound isn’t 0, and we’re not going to be looking at dramatics SRS changes, but a point differential based formula which does not account for pace is notable, especially in an era where rate based statistics are so popular even among casual fans. And yes, a 100-80 win is objectively more impressive/ meaningful from a statistical standpoint than a 120-100 win— you’re scoring 125% of your opponent’s total vs 120%, and in a slower paced game, it’s harder to win by 20 than in a faster paced game. Of course, not all defensive teams play at a slow pace. Granted, it’s hard to adjudicate the “impressiveness” of any two wins without a lot of context, but over the length of a season, I think it’s absolutely a salient distinguisher.

Point% seems to me to be the most effective way to evaluate in this context. An inverse pace formulation would show the merits of your classic Spurs slow it down and play iron defense, but it wouldn’t account for the sort of run and gun teams which are actually much better defensively than offensively (like many young teams with great athletes who aren’t great team offensive players). Those teams may in the end have lower scores even though the pace is higher, just because both offenses will be inefficient in those games.

I understand the point here, but looking at it its hard for me to say it'll make any difference in the grand scheme - if we want just a raw differential we can look at netRtg instead, as that is pace-adjusted and will paint a clearer picture.

WestGOAT wrote:From what I recall SRS = (Points) Margin of Victory (MOV) adjusted for Strength of Schedule (SOS).

You could try to use Points Ratios instead, but how do you handle scorelines containing 0, for example the 17-0 you mentioned. This shouldn't be an issue in basketball though.

The author was obviously referring to football. However, the point perhaps isn't without merit (in fact it may be even MORE relevant to basketball).

For example if we see a game in our modern space n' pace high-scoring 3pt era where the outcome is 135-119........is that really a more relevant or "bigger" margin of victory than a 89-78 victory in the slogging defensive grudge-matches of circa-2000?

SRS would say yes; but as to OP's question, I'm not sure it really is. Point ratio calculation is an intriguing concept, imo. I'd honestly never thought about it before.

trex_8063 wrote:

WestGOAT wrote:From what I recall SRS = (Points) Margin of Victory (MOV) adjusted for Strength of Schedule (SOS).

You could try to use Points Ratios instead, but how do you handle scorelines containing 0, for example the 17-0 you mentioned. This shouldn't be an issue in basketball though.

The author was obviously referring to football. However, the point perhaps isn't without merit (in fact it may be even MORE relevant to basketball).

For example if we see a game in our modern space n' pace high-scoring 3pt era where the outcome is 135-119........is that really a more relevant or "bigger" margin of victory than a 89-78 victory in the slogging defensive grudge-matches of circa-2000?

SRS would say yes; but as to OP's question, I'm not sure it really is. Point ratio calculation is an intriguing concept, imo. I'd honestly never thought about it before.

I mean, while on the subject, I think that overtimes should be an automatic 1 point win for SRS/MOV purposes. Seems kind of silly that a team that tied in 48 minutes but gets smacked by 10 points in OT does worse than one that just lost by 8 in regulation.