How Do You Evaluate Player Defense Statistically?

[ty 4191]

Hi Everyone!

Several questions:
--Which defensive metrics do you all use, for current players, to evaluate player defense? Which are most important and valid and which should be discarded? Why?

--Which metrics do you use for pre "Databall Era" (1996-1997) players, to evaluate defense?

--Which metrics do you use for pre steals/blocks/turnovers/off vs. defensive rounds, (etc.) players (pre 1973-1974)?

Thanks so much for all your expertise and feedback. Greatly appreciated. I'm here to learn!!

[Dutchball97]

For the current season I'm mainly going off EPM (which just released for 22/23) and LEBRON (which should be coming any day now). For earlier seasons in the databall era RAPM and mainly PI RAPM at that is my go to for defense. Before that it gets a lot more difficult so it's necessary to look at how well their team did defensively and how much of that is likely to be attributed to certain players. DBPM, DWS and defensive rating are all very unreliable and shouldn't be used where superior metrics are available but even then it's still useful to get an overall feel for players that were around before more accurate metrics came around. When a player ranks highly in all these flawed metrics it isn't a certainty they were good defenders but most good defenders do generally perform well in these metrics. So if their boxscore metrics are positive and their team was strong defensively (or decent defensively with poor defensive teammates in the case of guys like Mutombo and KG) it's likely they were good defenders.

[AEnigma]

Looking at team performance when a key defender is added or misses time is also informative. In the databall era, we can say 2000 Dikembe had massive defensive impact even though the overall roster result was dreadful. You cannot do the same as easily without that type of data, but looking at someone like Thurmond and the Warriors, he would have them as a top defence and then they would crater whenever he missed time. Same applies to Bill Walton. Or what happened to team defences when guys like Dave DeBusschere or Paul Silas were added to a roster. Those “WOWY” (with or without you) results match contemporary reputation and what our eye test can tell us.

On that note, contemporary reputation can be potentially useful too. Most of us know the flaws with guidebooks and all-defensive teams, but they still can help you figure out who to consider more closely. Going down the line, a guy like Bob Lanier has been specifically maligned at times because his Pistons were frequently bad on defence, yet guidebooks from the time were complimentary of his defence, and his teams routinely collapsed whenever he missed time; he never made all-defensive teams, so we can say the perception of him was never that of a top defensive big, but we can confidently say that he was better than his team’s defensive ratings might suggest.

[70sFan]

AEnigma nailed it. I would only add that you need to do a lot of tape analysis to get a clear picture of why some players had clear positive impact, while others did not. You can't make it work without watching games.

[ty 4191]

Looking at team performance when a key defender is added or misses time is also informative. In the databall era, we can say 2000 Dikembe had massive defensive impact even though the overall roster result was dreadful. You cannot do the same as easily without that type of data, but looking at someone like Thurmond and the Warriors, he would have them as a top defence and then they would crater whenever he missed time. Same applies to Bill Walton. Or what happened to team defences when guys like Dave DeBusschere or Paul Silas were added to a roster. Those “WOWY” (with or without you) results match contemporary reputation and what our eye test can tell us.

Where do you find on/off defensive data for people like Bill Walton and Nate Thurmond?

[AEnigma]

Looking at team performance when a key defender is added or misses time is also informative. In the databall era, we can say 2000 Dikembe had massive defensive impact even though the overall roster result was dreadful. You cannot do the same as easily without that type of data, but looking at someone like Thurmond and the Warriors, he would have them as a top defence and then they would crater whenever he missed time. Same applies to Bill Walton. Or what happened to team defences when guys like Dave DeBusschere or Paul Silas were added to a roster. Those “WOWY” (with or without you) results match contemporary reputation and what our eye test can tell us.

Where do you find on/off defensive data for people like Bill Walton and Nate Thurmond?

There is none, but what you can do when you have players like Walton or Thurmond who miss large chunks of the season is look at the team’s performance in games they played and in games they did not play and take an approximate measurement of how much worse the team (or the team’s defence specifically) becomes. With/without versus on/off.

[ty 4191]

There is none, but what you can do when you have players like Walton or Thurmond who miss large chunks of the season is look at the team’s performance in games they played and in games they did not play and take an approximate measurement of how much worse the team (or the team’s defence specifically) becomes. With/without versus on/off.

Like, going through box score by box score, and looking at teams points allowed and FG against? Or is there a less time consuming way?

In any case, this is all quite helpful. Thank you!

[OhayoKD]

There is none, but what you can do when you have players like Walton or Thurmond who miss large chunks of the season is look at the team’s performance in games they played and in games they did not play and take an approximate measurement of how much worse the team (or the team’s defence specifically) becomes. With/without versus on/off.

Like, going through box score by box score, and looking at teams points allowed and FG against? Or is there a less time consuming way?

In any case, this is all quite helpful. Thank you!

basically you look at how well the team plays with and then without, or in this case how well the defense holds up without and how it looks with.

Then you adjust for context.

[tsherkin]

Honestly, I don't really do a lot of player defensive evaluations using unified stats or +/-. I try to watch as many games of their play as possible and track them individually, or at least I do when I have time. Then you can look at smaller things like deflections, quality rotations and so forth. Then you have to step back and evaluate what was part of the scheme and just didn't work (or gave up a secondary option) and what was actually a mistake, etc. I suppose stuff like PI RAPM and EPM and whatever are all decent beginnings, though, especially relative to reputation and team defensive performance. On/Off numbers, etc.

[cupcakesnake]

While I definitely check whatever numbers are available, it's more for fun than analysis. If I see a catch-all metric really loving a certain player, it's a guide for who to watch to see if I can figure out why. (EPM loves Justise Winslow and Dean Wade this year so far so it makes me curious to see what they're doing out there).

But I truly feel you need to see what kind of play types a player faces, and how they perform in those, and what the defensive structure around them is like. Even great defenders usually have weaknesses or types of plays they're less adept at. Sometimes those defenders play in defensive structures that mask that while other defenders are stuck playing in defenses where that's always exposed. Gobert playing without any secondary help defender, 2016 Kawhi playing in Tony Parker/Pau Gasol lineups as examples that jump to mind. Nothing in the numbers helps me understand that Utah has Bojan Bogdanovic as the rim protector when Rudy defends the perimeter, and nothing in numbers will help me understand how opponents would remove Kawhi's man from the offense to attack the Parker/Gasol pick & roll. You can look at lineups and make assumptions, but you have to see it to understand it.

I think even with injuries and on/off, there's so much context to understand. For years, Joel Embiid had no realy back up center, and the on/off made him appear like a god. The for 1 year, Al Horford was his backup, and the on/off made him look mortal. Embiid was probably the same level of defender in all of those years. Sometimes a guy gets injured, and their team is well equipped to survive because of a good defensive backup at their position. Sometimes that happens to be a weakness in the roster, but I don't think the player should get extra credit for that.