APBRmetrics

tsherkin · Joined: 31 Jan 2005 Posts: 247

Stop me if this has been done in any great detail before, but has anyone looked at if simply adjusting for pace is actually a viable method of diminishing stats from one era to the next?

I was having a conversation with someone on RealGM and this came up.

We were discussing scoring talent between Duncan and Kareem, and the idea came up that Duncan scored at such-and-such a rate in the early 20s, he was evidencing as much or more scoring talent as Kareem did in the 70s because the difference was only pace-related.

This doesn't pass the laugh test for me. I watched Duncan and I've watched a lot of Kareem and to me, it seems qualitatively clear that Duncan had nothing like Kareem's offensive talent and putting it down to pace is a disservice to KAJ. But I'm looking for a way to back it up without resorting to the age-old "Nuh-uh!!" I know that Kareem had more range and I know that he used both sides of the paint more effectively because his left-handed hook was pretty damned good.

But the conversation took to per-possession scoring rate, which seemed to indicate that although Kareem's shot volume in, say, 76-77 (when he took 18.7 FGA/g) wasn't unreachable in the modern era (compared to someone like Al Jefferson, who managed just under 18 FGA/g over a full season but two years ago, before he was injured), that Kareem's offense was pace-inflated and he couldn't replicate it in the modern era.

What do people here think?

Qualitatively, it seems laughable. It seems sensible to suggest that, as it did in his own career, KAJ's FG% would fluctuate anywhere from the low 50s to as high as 60%, depending on the season, but would he really be incapable of replicating his scoring rate on similar efficiency now, despite facing doubles and triples all night in his own career and still getting it done?

We're not talking the scoring he managed in his first 7 seasons; he was playing 40-44 mpg in those years and it seems unlikely that he'd do that now. But in 77, he played 36.8 mpg, 18.7 FGA/g, 6.5 FTA/g and while it's true that he played at a pace of 104.7, which is much higher than most teams play at today, does that matter?

Magic's numbers didn't seem to diminish much when he went from playing at 103.7 in 83-84 (18/7/13, 62.8% TS) to 94.1 in 90-91 (19/7/12.5, 62.3% TS).

It seems more likely that the scoring depth would diminish rather than the individual, particularly since we've seen recent examples of players who could reach a similar shot volume, particularly with that kind of skill set (range, equal usage of both sides of the key, high/mid/low post skills, passing ability, etc).

So what do you guys think? Does this make sense? Can we just directly apply the pace adjustment and see what happens, is that sufficient?

DSMok1 · Posted: Tue Nov 09, 2010 2:31 pm Post subject:

The first way to attack it would be to look at pace, but the game was also different back then. Perhaps comparing to league average would be useful.

Hmmm... If we're actually talking effectiveness, then TS% and USG% would be the first place to go, and again compare TS% to the league as a whole.
_________________
GodismyJudgeOK.com/DStats

acollard · Joined: 22 Sep 2010 Posts: 56 Location: MA

I think pace adjustment does make sense, to some degree, because higher percentage scoring opportunities only come by so often. If the game is played at a faster pace, higher percentage scoring opportunities will occur more often.

You see this happen a lot with bigs who have a very high FG% on a low usage. The first impulse is to ask, why don't they just shoot more if they are so efficient? But often they shoot at such a high percentage because most of their shots are put-backs or dunks off of feeds from a penetrating guard. The simple fact is there are only so many chances for them to take these high percentage shots. It seems if there are 10% more possessions per game, it would follow that these high percentage chances (and other high percentages situations) would occur roughly 10% more often.

I'm not saying Kareem only scored points that way, but the number of chances he got per game in those favorable positions probably was affected by pace.

DSMok1 · Posted: Tue Nov 09, 2010 2:59 pm Post subject:

Thinking about it, yes, you MUST adjust for pace. If you're talking point totals, that must be done.

Consider also, though: the talent pool was different. There were fewer teams, then, so the level of competition would be higher--except the population from which the NBA players were being taken was smaller also.

As for Duncan vs. Kareem... At age 30, the first season for which we have all of their numbers, Duncan appears to have been close to Kareem's equal in the scoring department. I'd look at TS% and USG% as the primary measures.
_________________
GodismyJudgeOK.com/DStats

tsherkin · Joined: 31 Jan 2005 Posts: 247

The thing I'm struggling with in this specific comparison is that Kareem wasn't getting a lot of "easy" scoring chances by comparison to what he'd get today. He slowed the game down on his possessions and got a lot of half-court iso stuff with multiple defenders draped all over him and he did it at similar ranges to what guys with post skills can get today... and occasionally stepped out farther because he was a skilled high post player.

The thing that I have trouble believing is that Duncan is a comparable or better scorer than Kareem with a more limited skill set. He's a weaker FT shooter with less range and has always done a lot more work in the PnR than did Kareem. Some type of concession to pace makes sense in principle, but it doesn't strike me as completely valid when you're talking about a low-post iso player like KAJ. Duncan's a very good player, but apart from 06-07, he doesn't seem to be quite as good from an efficiency standpoint and his ORTG has tended to be lower than Kareem's as well.

Now, the usage does seem to reflect that Kareem had the ball less, but if there are more possessions overall, than Kareem using a similar number of possessions would be reflected as a lower overall usage rate, would it not? Is usage a pace-adjusted stat? Because if it's not, then the comparison along that angle kind of falls apart. It's supposed to be a percentage of team plays used while on the floor, right? So it would make sense then that it's something impacted by total plays used while that player is on the floor and thus not adjusted to account for pace.

I guess it'd be useful to do a direct comparison of possessions used per game for a more accurate picture, because if that number is similar, then the USG% is irrelevant.

Does anyone see where I'm coming from here? This is a stats forum, so it feels a little awkward but this specific comparison is bothering me because it doesn't seem to add up.

DSMok1 · Posted: Tue Nov 09, 2010 3:55 pm Post subject:

I disagree on the usage issue. If a player uses 20% of his team's possessions, which equals 20 possessions, that is not the same as a player using 30% of his team's possessions, which equals 20 possessions. We must evaluate each player in the context of trying to help his team win. If a player averages 1.2 PPP in each of those 2 cases, the second player helps his team significantly more. If each of those teams average 1 PPP in all other possessions, the second team would have an efficiency of 1.06 PPP while the first would have an efficiency of 1.04 PPP.

Something you are referring to in regards to PnR vs. ISO: you are implying that Duncan was "assisted" more. In that case, more of the credit for his numbers should be directed to his PGs rather than him. Unfortunately, we don't have "assisted" numbers for Kareem. I will disagree with you on your evaluation of Duncan here, though--he has had fewer of his FGs assisted than any other post player (well, close to the black hole known as Zach Randolph) in the last 5 years--hovering around or just above 50%. He's below average for the league as a whole, counting all of the guards! Of course, Kareem was probably even lower.
_________________
GodismyJudgeOK.com/DStats

Crow · Joined: 20 Jan 2009 Posts: 806

This old thread discussed some of the same ground

viewtopic.php?p=14969&sid=04dc243bd7c1b9b659408e8604f14ca9

The term league rating is out there, though, in the usage I am recalling, it is just league average efficiency.

Mike Goodman's comparison methods and results are worth finding and reviewing. There is some discussion and linkage in the above thread.

tsherkin · Joined: 31 Jan 2005 Posts: 247

DSMok1 · Posted: Tue Nov 09, 2010 7:23 pm Post subject:

What I'm trying to explain is that the constant on the team is the percentage of the team's possessions. It is not the number of possessions. We should measure each player and team on a per-possession basis (that is one of the basic tenets of the APBRmetric philosophy). A player with a usage of 20% is league-average in how often he is used, no matter the pace the team plays with.

An example from now: Minnesota plays at 104 poss/game. Charlotte is playing at 90 poss/game. A player that has a 20% usage for each, and scores on exactly the same percentage of his possessions, will score more for Minnesota. That doesn't mean the player is better; in fact the player is playing exactly equally. The same case is true here, though there are other differences (opponents are not the same, league-average is not the same).
_________________
GodismyJudgeOK.com/DStats

tsherkin · Joined: 31 Jan 2005 Posts: 247

DSMok1 · Posted: Tue Nov 09, 2010 9:16 pm Post subject:

Basically, you feel that there is a relationship between pace and usage--post players tend to take a higher proportion of shots in slow-paced games and guards and wings a higher proportion in fast-paced games. I could see that. I wonder if a study could be formulated to figure out if that's the case?

I note that there weren't any players with really high usage percentages in the 80s, compared with the 2000s. Other than Jordan, of course. See: http://www.basketball-reference.com/play-index/tiny.cgi?id=b3EMt

Mike G · Posted: Wed Nov 10, 2010 7:18 am Post subject:

Tim Duncan in 2002 averaged 30 points per 48 minutes, for a team that gave up 90 pts/48. So he scored 33.3% of what the rest of the league scored in games he was in.

Kareem (Alcindor) in 1971 avg'd 37.8 per48, and Mil opponents avg'd 105.7 .
He was therefore scoring 35.8% of the totals of 5-man opponent lineups.

Kareem had a better TS% in these years of comarison, especially relative to the rest of their respective leagues. Including such considerations, I rank 6 of Kareem's scoring seasons better than Duncan's best.
_________________
`
36% of all statistics are wrong

DSMok1 · Posted: Wed Nov 10, 2010 9:03 am Post subject:

gabefarkas · Joined: 31 Dec 2004 Posts: 1313 Location: Durham, NC

schtevie · Joined: 18 Apr 2005 Posts: 409

If the goal is to make a meaningful comparison of two players, playing different positions, in distinctly different eras, over which time so many aspects of the game were dramatically transformed, my first and best piece of advice would be to abandon the goal.

Failing that, the best you can do is compare the players in terms of rankings within their own era, e.g. Kareem had an X%ile TS% in year Y vs. so and so for Duncan. Not as sexy a result, but at least such a comparison has some nutritional value.