San Antonio early offseason in review (HW/bondom34/dbrandon)

[HartfordWhalers]

RAPM doesn't have anything to say about individual defense. You can literally play only with four great defenders and have a great RAPM even if you're a sieve. By adding stops and forced turnovers, you force a situation where the player has to tangibly contribute to his team's success. It's certainly better from a narrative standpoint. Like, "Pau's team doesn't do all that well defending when he's on the floor, but he blocks shots and rebounds." Because if all you need from him is to block shots and board, then you get a more honest view of his defensive ability.

If you have a player who plays with 4 great defenders and the result is a team defense worse than when those 4 great defenders play with others, you will have a negative RAPM.

Very roughly put, RAPM solves Team 1 scoring = {O1 + O2 + O3 + O4 + O5 } - {D6 + D7 + D8 + D9 + D10}.

So, if D6-D9 are superb defenders, it can still assign a negative to the last guy so long as it sees those 4 play with enough other people that it knows who the good 4 of the 5 are.

A team's Drating on the other hand suffers from the problem you say, where the worst defender on the best defensive team might have a better rating than an extremely good defender on a bad team.

Thanks for the explanation.

But that assumes the "others" aren't sieves themselves.

No. I answered the question using that as the example because you said if the others are all great. But the principle is the same if lets say 2 of the other 4 are bad, or all 4 are.

Re-paraphrased:
Very roughly put, RAPM solves Team 1 scoring = {O1 + O2 + O3 + O4 + O5 } - {D6 + D7 + D8 + D9 + D10}.

So, if D6 and D7 are good defenders, and D8 and D9 are bad defenders, it can still assign a negative/positive to the last guy so long as it sees those 4 play with enough other people that it knows who is good and who is bad.

RAPM and RPM have a bunch of faults, but it definitely isn't that it makes assumptions like have been suggested.

[bondom34]

Butler was 43rd in multiyear, which is the more reliable metric. Single year is really noisy.

To add, Tony Snell is 20th. So yeah, they've got fine perimeter defenders in Chicago.

They are fine, hence why Gasol was a good defender when you take into account his blocks, steals and rebounds. I fully expect him to be a better defender in SA unless the Spurs have even worse injury luck.

Also, if the Spurs are running an offense through Pau, great, but again you'd better run a crazy good offense, because again, he's giving up a ton more defensively. So whatever your excuse, unless he's putting up great scoring numbers he's killing the defense.

According to RAMP, which again is not set in stone as the best stat to judge players. Stats like RPM think he'll be fine on defense and won't need to do too much to be a positive. It's one thing to say you like one stat more than any others. It's another to act like no one can argue against that stat.

And HW covered the last point, RAPM does account for other players and considering its the main component of RPM, this argument doesn't make much sense. You can look at RAPM and blocks/steals separately.

Why do you WANT to look at them separately, though? If you're conceding that blocks and steals matter, then it doesn't make sense to have a more limiting stat be the standard-bearer. RPM is a suggestion on how much all those things matter. It could be wrong, but I'm not willing to dismiss it simply because it cares about stats.

Except blocks/steals aren't really an indicator of good defense .

And RPM is about the only stat that thinks that he'll be "fine" on defense.

And the entire point of RAPM was to remove the box score. RPM is an adulterated version of it. The point is that a box score doesn't measure anything really useful ultimately, its simplified and won't tell you who's really helping a team win a lot of the time. Hence some players w/ great box score numbers aren't great in plus/minus. See Corey Brewer's defense. Or Drummond. See a bunch of players.

[Andre Roberstan]

Butler was 43rd in multiyear, which is the more reliable metric. Single year is really noisy.

To add, Tony Snell is 20th. So yeah, they've got fine perimeter defenders in Chicago.

They are fine, hence why Gasol was a good defender when you take into account his blocks, steals and rebounds. I fully expect him to be a better defender in SA unless the Spurs have even worse injury luck.

Also, if the Spurs are running an offense through Pau, great, but again you'd better run a crazy good offense, because again, he's giving up a ton more defensively. So whatever your excuse, unless he's putting up great scoring numbers he's killing the defense.

According to RAMP, which again is not set in stone as the best stat to judge players. Stats like RPM think he'll be fine on defense and won't need to do too much to be a positive. It's one thing to say you like one stat more than any others. It's another to act like no one can argue against that stat.

And HW covered the last point, RAPM does account for other players and considering its the main component of RPM, this argument doesn't make much sense. You can look at RAPM and blocks/steals separately.

Why do you WANT to look at them separately, though? If you're conceding that blocks and steals matter, then it doesn't make sense to have a more limiting stat be the standard-bearer. RPM is a suggestion on how much all those things matter. It could be wrong, but I'm not willing to dismiss it simply because it cares about stats.

There's actually a decent amount of evidence that blocks don't really matter that much as a defensive metric (the DREDGE article the other day on NylonCalculus pointed this out). Rim protection as a whole does, but that's tremendously dependent on context (e.g. Serge Ibaka is a fantastic rim protector, but since the rise of the stretch 4 is usually not near the basket). Steals do have a ton of value, though.

[Chinook]

If you have a player who plays with 4 great defenders and the result is a team defense worse than when those 4 great defenders play with others, you will have a negative RAPM.

Very roughly put, RAPM solves Team 1 scoring = {O1 + O2 + O3 + O4 + O5 } - {D6 + D7 + D8 + D9 + D10}.

So, if D6-D9 are superb defenders, it can still assign a negative to the last guy so long as it sees those 4 play with enough other people that it knows who the good 4 of the 5 are.

A team's Drating on the other hand suffers from the problem you say, where the worst defender on the best defensive team might have a better rating than an extremely good defender on a bad team.

Thanks for the explanation.

But that assumes the "others" aren't sieves themselves.

No. I answered the question using that as the example because you said if the others are all great. But the principle is the same if lets say 2 of the other 4 are bad, or all 4 are.

Re-paraphrased:
Very roughly put, RAPM solves Team 1 scoring = {O1 + O2 + O3 + O4 + O5 } - {D6 + D7 + D8 + D9 + D10}.

So, if D6 and D7 are good defenders, and D8 and D9 are bad defenders, it can still assign a negative/positive to the last guy so long as it sees those 4 play with enough other people that it knows who is good and who is bad.

RAPM and RPM have a bunch of faults, but it definitely isn't that it makes assumptions like have been suggested.

But that didn't actually answer what I said. Let's say -- and yes, this isn't realistic at all -- that Players 1-4 only played with Players 5 and 6, and Player 5 only played with 1-4, and 6 only played with 1-4. Then 7-11 played in a line-substitution afterward. And 1-4, 5 and 6 only played against 1-5 of the opponents, while 7-11 always played 6-10 of the other team. And there's either only one other team or 29 identical teams.

In that completely contrived scenario, there'd be no way to compare anyone besides 5 and 6 on the first team. With so many things controlled for, it would work great to compare them, but obviously, it doesn't do anything for anyone else.

Now, again, that's not realistic, but you still get a skew the closer to get to that scenario (the more the same lineups play together, the less fluid your substitutions, the more frequently you play the same teams, yadda yadda). I don't know how well it's accounted for, but it's probably not as pure as you're suggesting.

[Chinook]

Butler was 43rd in multiyear, which is the more reliable metric. Single year is really noisy.

To add, Tony Snell is 20th. So yeah, they've got fine perimeter defenders in Chicago.

They are fine, hence why Gasol was a good defender when you take into account his blocks, steals and rebounds. I fully expect him to be a better defender in SA unless the Spurs have even worse injury luck.

Also, if the Spurs are running an offense through Pau, great, but again you'd better run a crazy good offense, because again, he's giving up a ton more defensively. So whatever your excuse, unless he's putting up great scoring numbers he's killing the defense.

According to RAMP, which again is not set in stone as the best stat to judge players. Stats like RPM think he'll be fine on defense and won't need to do too much to be a positive. It's one thing to say you like one stat more than any others. It's another to act like no one can argue against that stat.

And HW covered the last point, RAPM does account for other players and considering its the main component of RPM, this argument doesn't make much sense. You can look at RAPM and blocks/steals separately.

Why do you WANT to look at them separately, though? If you're conceding that blocks and steals matter, then it doesn't make sense to have a more limiting stat be the standard-bearer. RPM is a suggestion on how much all those things matter. It could be wrong, but I'm not willing to dismiss it simply because it cares about stats.

There's actually a decent amount of evidence that blocks don't really matter that much as a defensive metric (the DREDGE article the other day on NylonCalculus pointed this out). Rim protection as a whole does, but that's tremendously dependent on context (e.g. Serge Ibaka is a fantastic rim protector, but since the rise of the stretch 4 is usually not near the basket). Steals do have a ton of value, though.

I read that Pau was actually among the best centers in the league in rim-protection. It as a graph on Spurstalk. I'll try to see if I can find it.

[HartfordWhalers]

Thanks for the explanation.

But that assumes the "others" aren't sieves themselves.

No. I answered the question using that as the example because you said if the others are all great. But the principle is the same if lets say 2 of the other 4 are bad, or all 4 are.

Re-paraphrased:
Very roughly put, RAPM solves Team 1 scoring = {O1 + O2 + O3 + O4 + O5 } - {D6 + D7 + D8 + D9 + D10}.

So, if D6 and D7 are good defenders, and D8 and D9 are bad defenders, it can still assign a negative/positive to the last guy so long as it sees those 4 play with enough other people that it knows who is good and who is bad.

RAPM and RPM have a bunch of faults, but it definitely isn't that it makes assumptions like have been suggested.

But that didn't actually answer what I said. Let's say -- and yes, this isn't realistic at all -- that Players 1-4 only played with Players 5 and 6, and Player 5 only played with 1-4, and 6 only played with 1-4. Then 7-11 played in a line-substitution afterward. And 1-4, 5 and 6 only played against 1-5 of the opponents, while 7-11 always played 6-10 of the other team. And there's either only one other team or 29 identical teams.

In that completely contrived scenario, there'd be no way to compare anyone besides 5 and 6 on the first team. With so many things controlled for, it would work great to compare them, but obviously, it doesn't do anything for anyone else.

Now, again, that's not realistic, but you still get a skew the closer to get to that scenario (the more the same lineups play together, the less fluid your substitutions, the more frequently you play the same teams, yadda yadda). I don't know how well it's accounted for, but it's probably not as pure as you're suggesting.

My apologies.

If I had known you had asked about a completely unrealistic hypothetical I wouldn't have answered you at all.

[bondom34]

Which is great, but if the team still gives up a boatload of points elsewhere doesn't really matter.

[Chinook]

Except blocks/steals aren't really an indicator of good defense

Not without context, which is theoretically what RPM does. Ignoring them altogether in a stat doesn't seem like the answer.

And RPM is about the only stat that thinks that he'll be "fine" on defense.

I don't think being top 10 in the league is "fine". It seems completely workable for the Spurs, given the offensive advantages.

And the entire point of RAPM was to remove the box score. RPM is an adulterated version of it. The point is that a box score doesn't measure anything really useful ultimately, its simplified and won't tell you who's really helping a team win a lot of the time. Hence some players w/ great box score numbers aren't great in plus/minus. See Corey Brewer's defense. Or Drummond. See a bunch of players.

But that's not a universally accepted assumption. I love advanced stats as much as anyone, but most of them don't ignore the measurables at all. We have to decide how much they matter. Because it's clear that ignoring them leads to people thinking Tim was a defensive presence against Golden State, which he wasn't.

[Chinook]

My apologies.

If I had known you had asked about a completely unrealistic hypothetical I wouldn't have answered you at all.

Your apology is accepted. That is a flaw in the logic of the system, and as I said, it doesn't have to be that extreme to matter. I was looking at the limit, if you want to use some calculus term, but it's not like you can't see the trend before you get there. I don't claim to know exactly how much it matters or how well it's combated by other parts of the equations. But it's there.

[Chinook]

Which is great, but if the team still gives up a boatload of points elsewhere doesn't really matter.

But that's just an assumption based on RAPM.

[Texas Chuck]

There's actually a decent amount of evidence that blocks don't really matter that much as a defensive metric (the DREDGE article the other day on NylonCalculus pointed this out). Rim protection as a whole does, but that's tremendously dependent on context (e.g. Serge Ibaka is a fantastic rim protector, but since the rise of the stretch 4 is usually not near the basket). Steals do have a ton of value, though.

This is great stuff. Steals in fact are by a fairly good margin the best possible defensive result because the other team doesn't score on that possession and with a live-ball turnover, you generate the most efficient offensive scoring options.

The caveat when looking at a defender is if he takes himself out of good defensive positioning chasing steals. Older Kobe was a great example of this--his overall defense suffered because he did a ton of ball watching trying to anticipate for steals letting his man get a lot of open looks/cuts to basket. Chris Paul is a great example of a player who gets a lot of steals, but remains a very effective overall team defender.


And of course the greatest rim-protector of all-time, Mr. William Russell talked frequently about the efforts he made when blocking shots to try and control the ball either himself or to a teammate as opposed to doing the DeAndre spike it into the 8th row because it looks cool(not that there isn't some value in the deterrence he brings) and thus despite the Celtics of his day being relatively mediocre offensively, they beat teams by all of the extra possessions they got and excelled in transition.

[bondom34]

Except blocks/steals aren't really an indicator of good defense

Not without context, which is theoretically what RPM does. Ignoring them altogether in a stat doesn't seem like the answer.

And RPM is about the only stat that thinks that he'll be "fine" on defense.

I don't think being top 10 in the league is "fine". It seems completely workable for the Spurs, given the offensive advantages.

And the entire point of RAPM was to remove the box score. RPM is an adulterated version of it. The point is that a box score doesn't measure anything really useful ultimately, its simplified and won't tell you who's really helping a team win a lot of the time. Hence some players w/ great box score numbers aren't great in plus/minus. See Corey Brewer's defense. Or Drummond. See a bunch of players.

But that's not a universally accepted assumption. I love advanced stats as much as anyone, but most of them don't ignore the measurables at all. We have to decide how much they matter. Because it's clear that ignoring them leads to people thinking Tim was a defensive presence against Golden State, which he wasn't.

Actually, ignoring them generallyu isn't a bad idea. Steals leaders are strangely not all good defenders, nor are blocks leaders.

And Carlos Boozer was top 10 in DWS a few years ago, I assume he was a fine defender that year too even though it was the single stat showing he was "fine".

And apparently you don't really love much in the way of any analytics, or you'd realize how awful the idea of Pau vs. GSW is. I'll wait for a Bulls fan to chime in maybe.

[Mykhyn]

1. People are underrating Dedmon. Dude is gonna be a beast with a good coaching staff and a large role that will be handed to him(defensive big)

2. Gasols main strength for us will be his passing which we will need direly and will force him to see more minutes than is ideal. We lost Diaw and Tim. Parker and Manu are even older. I expect Gasol to take over a lot of the bench playmaking with Manu playing more time with the starters in important matchups(unless Anderson impresses). Gasols defense isnt an issue against GSW comparatively to Duncan since Duncans best spot against GSW was on the bench. Yes Gasol will be worse against most teams, but against the teams we expect to face at the end of the season hes an equal to Duncan

3. No Green shouldn't have been traded.

4. Dejounte Murray was a steal. I'm salivating at a lineup of Murray-Simmons-Kawhi-Anderson-defensive big in 2 years

5. Losing Duncan hurts, but not really against GSW which is likely the only team that matters. Dedmon will be an upgrade for that matchup. I fully expect Dedmon to be our starter over Gasol.

6. Lee doesnt need to be good defensively because he'll being playing against scrubs more often than not.

7. I expect our counter to Curry/Klay/Iggy/Durant/Green to be Manu/Green/Anderson(maybe Simmons instead)/Kawhi/Dedmon



With all that said I gave us a B due to not addressing our PG issue.

[Texas Chuck]

Dedmon will be an upgrade for that matchup. I fully expect Dedmon to be our starter over Gasol.

I think there is almost zero chance of this.

[Chinook]

Actually, ignoring them generallyu isn't a bad idea. Steals leaders are strangely not all good defenders, nor are blocks leaders.

Does it have to 100 percent to matter at all?

And Carlos Boozer was top 10 in DWS a few years ago, I assume he was a fine defender that year too even though it was the single stat showing he was "fine".

I feel you're getting to the point where you're really overstating your case. Gasol has a lot of good defensive metrics. He was 15th in the league in DWS, 18th in DBPM and 38th in DRTG -- and none of those are taking out small-sample-sized players. 82games.com has him as a mediocre defender by PER allowed (though elite by net PER) and with him having pretty neutral on-off numbers (they allow slight fewer points with him off the court but with a higher DFG%).

And apparently you don't really love much in the way of any analytics, or you'd realize how awful the idea of Pau vs. GSW is. I'll wait for a Bulls fan to chime in maybe.

Here are the aforementioned charts on rim-protection.

[Andre Roberstan]

1. People are underrating Dedmon. Dude is gonna be a beast with a good coaching staff and a large role that will be handed to him(defensive big)

2. Gasols main strength for us will be his passing which we will need direly and will force him to see more minutes than is ideal. We lost Diaw and Tim. Parker and Manu are even older. I expect Gasol to take over a lot of the bench playmaking with Manu playing more time with the starters in important matchups(unless Anderson impresses). Gasols defense isnt an issue against GSW comparatively to Duncan since Duncans best spot against GSW was on the bench. Yes Gasol will be worse against most teams, but against the teams we expect to face at the end of the season hes an equal to Duncan

3. No Green shouldn't have been traded.

4. Dejounte Murray was a steal. I'm salivating at a lineup of Murray-Simmons-Kawhi-Anderson-defensive big in 2 years

5. Losing Duncan hurts, but not really against GSW which is likely the only team that matters. Dedmon will be an upgrade for that matchup. I fully expect Dedmon to be our starter over Gasol.

6. Lee doesnt need to be good defensively because he'll being playing against scrubs more often than not.

7. I expect our counter to Curry/Klay/Iggy/Durant/Green to be Manu/Green/Anderson(maybe Simmons instead)/Kawhi/Dedmon



With all that said I gave us a B due to not addressing our PG issue.

I like Dedmon a lot for you guys, but you're fooling yourself if you think Pau signed anywhere he wasn't going to be a starter. It was an issue in Chi with Jo, it will be an issue in SA if he doesn't start.

And Iggy would light Anderson's slow ass on fire.

You guys can quote me on this at the end of the season: Gasol's best place against GSW will be on the bench as well.

The issue I have with Gasol isn't Gasol, who's a perfectly fine player. It's the opportunity cost incurred by prioritizing an older, slower frontcourt player over guys who can start taking over playmaking duties from Manu and Tony or more athletic bigs who can stay on the floor against fast frontcourts. Dedmon helps, but he's the only one. Unless Tony finds the fountain of youth, I don't think the backcourt is good enough any more to compete against GSW.

[Chinook]

Dedmon will be an upgrade for that matchup. I fully expect Dedmon to be our starter over Gasol.

I think there is almost zero chance of this.

Me too, but that's what Smitty reported.

[bondom34]

Actually, ignoring them generallyu isn't a bad idea. Steals leaders are strangely not all good defenders, nor are blocks leaders.

Does it have to 100 percent to matter at all?

And Carlos Boozer was top 10 in DWS a few years ago, I assume he was a fine defender that year too even though it was the single stat showing he was "fine".

I feel you're getting to the point where you're really overstating your case. Gasol has a lot of good defensive metrics. He was 15th in the league in DWS, 18th in DBPM and 38th in DRTG -- and none of those are taking out small-sample-sized players. 82games.com has him as a mediocre defender by PER allowed (though elite by net PER) and with him having pretty neutral on-off numbers (they allow slight fewer points with him off the court but with a higher DFG%).

And apparently you don't really love much in the way of any analytics, or you'd realize how awful the idea of Pau vs. GSW is. I'll wait for a Bulls fan to chime in maybe.

Here are the aforementioned charts on rim-protection.

No, but if James Harden is a steals leader, and Whiteside a blocks leader, and Miami's defense is actually better when Whiteside is on the bench, they're probably not a great indicator.

And again, you're stating all box score defensive stats. Which again, are largely useless.

And finally, as I said, rim protection is great, but he's a bad defender.

[bondom34]

Actually, ignoring them generallyu isn't a bad idea. Steals leaders are strangely not all good defenders, nor are blocks leaders.

Does it have to 100 percent to matter at all?

And Carlos Boozer was top 10 in DWS a few years ago, I assume he was a fine defender that year too even though it was the single stat showing he was "fine".

I feel you're getting to the point where you're really overstating your case. Gasol has a lot of good defensive metrics. He was 15th in the league in DWS, 18th in DBPM and 38th in DRTG -- and none of those are taking out small-sample-sized players. 82games.com has him as a mediocre defender by PER allowed (though elite by net PER) and with him having pretty neutral on-off numbers (they allow slight fewer points with him off the court but with a higher DFG%).

And apparently you don't really love much in the way of any analytics, or you'd realize how awful the idea of Pau vs. GSW is. I'll wait for a Bulls fan to chime in maybe.

Here are the aforementioned charts on rim-protection.

No, but if James Harden is a steals leader, and Whiteside a blocks leader, and Miami's defense is actually better when Whiteside is on the bench, they're probably not a great indicator.

And again, you're stating all box score defensive stats. Which again, are largely useless.

And finally, as I said, rim protection is great, but he's a bad defender.

[Chinook]

No, but if James Harden is a steals leader, and Whiteside a blocks leader, and Miami's defense is actually better when Whiteside is on the bench, they're probably not a great indicator.

The Miami/Whiteside thing was disproved with a large sample size. And yes, you aren't actually making an argument by saying this. Those guys are better defenders that they would be if they didn't do those things. Hardly a counter.

And again, you're stating all box score defensive stats. Which again, are largely useless.

In your opinion. This isn't fact just because you believe it to be so.

And finally, as I said, rim protection is great, but he's a bad defender.

But the Spurs have a way to making it so the bigs' man job on defense is to protect the rim once guys get funneled to them.