RealGM

Hi, I have been thinking lately about the best players with big drops in career averages from RS to PS. In many cases, it could be explained by on-court issues, but sometimes comparing career averages can be very misleading, because postseason sample isn't representative of their overall career sample.

This is especially true for two of the top 15 players ever who often hear criticism about their postseason play - Kevin Garnett and Wilt Chamberlain. Wilt Chamberlain played most of his postseason games in LA, even though it's only a small portion of his overall career. Meanwhile, Garnett played most of his postseason games as a Celtic, because he missed playoffs in some years during his prime and his postseason runs are short.

I want to use these two players to show flaws at comparing RS vs PS career numbers. Let's start with raw numbers:

Wilt in RS: 30.1/22.9/4.4 on 54.0 FG%, 51.1 FT% and 54.7 TS% in 45.8 mpg
Wilt in PS: 22.5/24.5/4.2 on 52.2 FG%, 46.5 FT% and 52.4 TS% in 47.2 mpg

At first look, Wilt's numbers took a notable hit in the playoffs. The problem with this comparison is that his RS scoring numbers are highly inflated by 1961-63 period, but he didn't play many postseason games in that span (only 3 in 1961, none in 1963). On the contrary, he played a lot of postseason games in LA when he was no longer a high volume scorer.

If we actually calculate a weighed RS averages, based on how many playoff games Wilt played each, the outcome would be notably different:

Wilt in RS (weighed): 26.8/21.8/4.7 on 58.5 FG%, 48.3 FT% and 55.4 TS% in 45.0 mpg
Wilt in PS (weighed): 22.5/24.5/4.2 on 52.2 FG%, 46.5 FT% and 52.4 TS% in 47.2 mpg

As we can see, the lower volume and efficiency is still noticeable, but the difference isn't as drastic as raw ppg can suggest.

You can also do it for 5 years "peaks":

1964-68 Wilt in RS (raw): 30.6/23.6/6.1 on 55.6 FG%, 46.8 FT% and 55.1 TS% in 46.2 mpg
1964-68 Wilt in RS (weighed): 29.8/23.5/6.3 on 58.2 FG%, 45.7 FT% and 55.5 TS% in 46.0 mpg
1964-68 Wilt in PS: 27.0/27.0/5.8 on 54.3 FG%, 44.2 FT% and 53.4 TS% in 47.9 mpg

You can do the same thing with KG, Robinson, Hakeem and pretty much all players that don't have a lot of postseason games during their prime seasons.

Do you think it's useful if someone wants to use career averages to compare RS vs PS numbers? This is not an attempt to make Wilt (or any other player) look better, but to show these averages in a better context.

Well, much like in the regular season, all great players (even pantheon level) have their playoff primes/peaks and "valleys." Lots of reasons for this. Different rosters/coaches/opponents, etc.

How are you calculating “weighted” averages here?

70sFan wrote:Do you think it's useful if someone wants to use career averages to compare RS vs PS numbers? This is not an attempt to make Wilt (or any other player) look better, but to show these averages in a better context.

This is an obvious reference to my recent thread on Malone and Garnett.

In short, this doesn't apply to Garnett as he played eight straight playoffs with Minnesota. He went further with Boston, as anyone does with a much better team, but the large sample size is there. Chamberlain not so much and he is wrongfully judged on skewed playoffs mostly due to older age and situation, such as bad coaches.

On here, Robinson and Malone are harshly criticised for their career playoffs but Garnett isn't and is wrongfully rated higher than them, which is absurd.

I do think career averages are useful most of the time. To use the example of Karl Malone I think it paints a pretty accurate picture. He made the post-season every year of his career, didn't take long to become good and aged gracefully. However, if we just focus on small samples we'd miss the message that he wasn't terrible but definitely less effective in the play-offs than the regular season as he was pretty inconsistent year to year. Picking a certain sample could intentionally or unintentionally paint him as better/worse than he actually was. It really is going to depend on the situation. Someone who rarely misses the play-offs and frequently has deep runs will have more meaningful play-off career averages than someone who missed the play-offs for a significant amount of years.

KG and Oscar are the best examples of play-off averages dropping because of some missed post-seasons in their prime. Jokic is an example of someone who has an exaggerated jump. Jokic has a regular season average of 19.8 ppg and a post-season average of 26.4 ppg but he didn't make the play-offs his first 3 years. In the years Jokic did make the post-season, his regular season averages were 20.1, 19.9, 26.4 and 27.1 and by dividing total points over those 4 seasons by his games played you get 23.3 ppg. A 3 ppg increase in the play-offs is still great but even then you still need to account for his post-season ppg being inflated by his 2022 ppg average being inflated from just being 1 series that happened to be a shootout with limited defense. It's similar for assists and rebounds as in those last 4 seasons he generally never had below his career averages so the actual relevant averages are higher than currently listed.

So ramblings aside, career averages are useful but only with proper context.

No-more-rings wrote:How are you calculating “weighted” averages here?

Weights are set by the number of postseason games played in a given season. It means that when you calculated weighted average for Wilt, you give 1962 numbers a weight of 12, but 1963 has a weight of 0.

migya wrote: This is an obvious reference to my recent thread on Malone and Garnett.

It's not, I have been thinking about it a few days and today I finally found time to make calculations for Wilt.

In short, this doesn't apply to Garnett as he played seven straight playoffs with Minnesota. He went further with Boston, as anyone does with a much better team, but the large sample size is there.

Garnett played 47 playoff games with Minny and 96 after the trade. It definitely has a massive impact on his total averages, in fact the difference is even more drastic than in Wilt's case.

On here, Robinson and Malone are harshly criticised for their career playoffs but Garnett isn't and is wrongfully rated higher than them, which is absurd.

As I said, I don't care about forcing any narratives here. This thread is about a new way to look at RS vs PS production, leave your agenda to your own threads.

Dutchball97 wrote:I do think career averages are useful most of the time. To use the example of Karl Malone I think it paints a pretty accurate picture. He made the post-season every year of his career, didn't take long to become good and aged gracefully. However, if we just focus on small samples we'd miss the message that he wasn't terrible but definitely less effective in the play-offs than the regular season as he was pretty inconsistent year to year. Picking a certain sample could intentionally or unintentionally paint him as better/worse than he actually was. It really is going to depend on the situation. Someone who rarely misses the play-offs and frequently has deep runs will have more meaningful play-off career averages than someone who missed the play-offs for a significant amount of years.

KG and Oscar are the best examples of play-off averages dropping because of some missed post-seasons in their prime. Jokic is an example of someone who has an exaggerated jump. Jokic has a regular season average of 19.8 ppg and a post-season average of 26.4 ppg but he didn't make the play-offs his first 3 years. In the years Jokic did make the post-season, his regular season averages were 20.1, 19.9, 26.4 and 27.1 and by dividing total points over those 4 seasons by his games played you get 23.3 ppg. A 3 ppg increase in the play-offs is still great but even then you still need to account for his post-season ppg being inflated by his 2022 ppg average being inflated from just being 1 series that happened to be a shootout with limited defense. It's similar for assists and rebounds as in those last 4 seasons he generally never had below his career averages so the actual relevant averages are higher than currently listed.

So ramblings aside, career averages are useful but only with proper context.

Yeah, that's why it could be useful to compare weighted averages instead of raw ones. Jokic, KG and Oscar are all very good examples when it can be very misleading.

70sFan wrote:

migya wrote: This is an obvious reference to my recent thread on Malone and Garnett.

It's not, I have been thinking about it a few days and today I finally found time to make calculations for Wilt.

In short, this doesn't apply to Garnett as he played seven straight playoffs with Minnesota. He went further with Boston, as anyone does with a much better team, but the large sample size is there.

Garnett played 47 playoff games with Minny and 96 after the trade. It definitely has a massive impact on his total averages, in fact the difference is even more drastic than in Wilt's case.

On here, Robinson and Malone are harshly criticised for their career playoffs but Garnett isn't and is wrongfully rated higher than them, which is absurd.

As I said, I don't care about forcing any narratives here. This thread is about a new way to look at RS vs PS production, leave your agenda to your own threads.

Garnett went to Boston at age 31, Malone's mvp seasons were after that age.

Is not anyone's agents, it's asking others for their reasonings.

migya wrote:

70sFan wrote:

migya wrote: This is an obvious reference to my recent thread on Malone and Garnett.

It's not, I have been thinking about it a few days and today I finally found time to make calculations for Wilt.

In short, this doesn't apply to Garnett as he played seven straight playoffs with Minnesota. He went further with Boston, as anyone does with a much better team, but the large sample size is there.

Garnett played 47 playoff games with Minny and 96 after the trade. It definitely has a massive impact on his total averages, in fact the difference is even more drastic than in Wilt's case.

On here, Robinson and Malone are harshly criticised for their career playoffs but Garnett isn't and is wrongfully rated higher than them, which is absurd.

As I said, I don't care about forcing any narratives here. This thread is about a new way to look at RS vs PS production, leave your agenda to your own threads.

Garnett went to Boston at age 31, Malone's mvp seasons were after that age.

Is not anyone's agents, it's asking others for their reasonings.

I don't compare Malone and Garnett in this thread. You have your own thread to do that.

70sFan wrote:

No-more-rings wrote:How are you calculating “weighted” averages here?

Weights are set by the number of postseason games played in a given season. It means that when you calculated weighted average for Wilt, you give 1962 numbers a weight of 12, but 1963 has a weight of 0.

I’m sorry but i’m not really quite following you. I understand that more games have more weight in averages, but bballref already accounts for this when you do multi year averages.

I could be missing something entirely though.

My guess is he is smoothing them over so it does not matter that Wilt played double the number of postseason games as a Laker. All that matters is the singular postseason-to-postseason average.

CharityStripe34 wrote:Well, much like in the regular season, all great players (even pantheon level) have their playoff primes/peaks and "valleys." Lots of reasons for this. Different rosters/coaches/opponents, etc.

Right...and they are all different, which is why--what 70sFan is presenting--is controlling for the differences between the hills and valleys.

No-more-rings wrote:

70sFan wrote:

No-more-rings wrote:How are you calculating “weighted” averages here?

Weights are set by the number of postseason games played in a given season. It means that when you calculated weighted average for Wilt, you give 1962 numbers a weight of 12, but 1963 has a weight of 0.

I’m sorry but i’m not really quite following you. I understand that more games have more weight in averages, but bballref already accounts for this when you do multi year averages.

I could be missing something entirely though.

Imagine a player plays 2 seasons, Season A averaging 30/10/10 in the regular season and Season B with averages of 20/5/5. In the post-season, the player doesn't make the playoffs in Season A but makes it in Season B averaging 22/6/6.

If we simply average the two regular seasons together, they would average 25/7.5/7.5 in the Regular Season and 22/6/6 in the Post-Season--a clear drop in production...but wait, with weighted averages, 100% of the weight would be on Season B since it is the only post-season, and we would conclude this player Increased their production in the post-season.

The same concept and results with a different approach (and longer approach) would simply being to sum the Total of the statistics in the seasons in which a player made the post-season, divide that Sum by the total number of games, and do the same for post-season. Weighted Averages is a short cut to this methodology with an identical answer.

No-more-rings wrote:

70sFan wrote:

No-more-rings wrote:How are you calculating “weighted” averages here?

Weights are set by the number of postseason games played in a given season. It means that when you calculated weighted average for Wilt, you give 1962 numbers a weight of 12, but 1963 has a weight of 0.

I’m sorry but i’m not really quite following you. I understand that more games have more weight in averages, but bballref already accounts for this when you do multi year averages.

I could be missing something entirely though.

Postseason numbers are taken from bballref straight, but it's about RS weighed averages. I will try to explain it with this example:

Wilt Chamberlain averaged 41.7 ppg in 1960-64 period.His average for the playoffs is 34.6 ppg. A quick comparison of these numbers don't take into account that Wilt didn't play a single game in 1963 playoffs and that he played a different number of games each postseason. To make it more apples to apples comparison, I use weighted average for RS numbers:

1960: 37.6 ppg (in RS) * 9 games (in PS)
1961: 38.4 ppg (in RS) * 3 games (in PS)
...
1964: 36.9 ppg (in RS) * 12 games (in PS)

You divide the sum by the numbers of playoff games played in 1960-64 period and you get the expected postseason averages based on RS numbers. In short, this way of calculating expected numbers include the fact that Wilt didn't play in the playoffs in 1963 and that he played most of his games in 1962 and 1964.

Why is it helpful? It gives you a good estimation of what numbers you should expect from Wilt in 1960-64 postseason sample and how much he truly regressed. Of course, you can play with numbers and adjust them to opponent faced, pace etc, but that's just a rough idea.

It makes sense to do for players who didn't play many playoff games in their primes - Oscar or KG for example. Is it clear now?

So basically we are trying to place a hypothetical average for missed postseasons then? That’s how I interpret this.

No-more-rings wrote:So basically we are trying to place a hypothetical average for missed postseasons then? That’s how I interpret this.

More like we try to exclude RS numbers when someone misses postseason. We also take into account that not all PS runs have the same length.

Take a look at this example:

Someone has 2 seasons when he averages:

30/6/6 in season A playing 82 games
20/4/4 in season B playing 82 games

his averages are: 25/5/5.

If he plays 2 PS games in season A but 20 PS games in season B, while averaging 24/5/5, you may say he underperformed, but with weighed averages it's clearly not the case, because we should expect him to average lesser numbers than what averages suggest.

RS averages: 25/5/5
Weighed averages: 21/4.2/4.2

PS averages: 24/5/5

It's clear that he didn't really underperformed during that time.

Of course, for two years samples it's not very useful, but when you look at the averages from 15 years samples, it could help. Look at Oscar Robertson example:

RS averages: 25.7 ppg
Weighed averages: 22.5 ppg
PS averages: 22.2 ppg

Oscar scored in the playoffs as we should expect from him based on the sample, his PS sample just isn't representative for his whole career.

Good thread. Although with the Wilt example, there is also the issue of opposition. Wilt in the regular season faced 7-8 different opponents in his big scoring years most of whom were ill-equipped to deal with him. However the Celtics had Bill Russell who could limit Wilt to an extent and that's the guy Wilt faced in the playoffs more often than not.

Even with other players that can be the case at least to some extent. Kareem faced Wilt twice, Thurmond thrice and Unseld once in six playoff series from 1971-1973. Based on that, some decline in production is expected.

Djoker wrote:Good thread. Although with the Wilt example, there is also the issue of opposition. Wilt in the regular season faced 7-8 different opponents in his big scoring years most of whom were ill-equipped to deal with him. However the Celtics had Bill Russell who could limit Wilt to an extent and that's the guy Wilt faced in the playoffs more often than not.

Even with other players that can be the case at least to some extent. Kareem faced Wilt twice, Thurmond thrice and Unseld once in six playoff series from 1971-1973. Based on that, some decline in production is expected.

By my count, Wilt faced opposing HOF Centers in 126 of his 160 career playoff games. 49 head to head with the greatest defensive center of all time. And, as someone noted, he played the vast majority of them after he stopped shooting and scoring a ton.

As a frame of reference, Russell played 104 out of 165 playoff games against HOF opposing Centers.

So, yes, absolutely, we should expect a significant PS drop off in offensive production for someone like Wilt.

70sFan and I showed/proved Wilt faced the toughest defenses, on average, of any ATG player in NBA history.

viewtopic.php?f=64&t=1836300

Wilt, Entire Career, Playoffs:

Vs. Bad Defenses (15 games, 9.4% of total games):
47.6 mpg 26.5 ppg 22.2 rbg 7.5 ast/g 54.6 TS% 5.9 rTS%

Vs. Average Defenses (47 games, 29.4% of total games):
47.4 mpg 22.8 ppg 23.5 rbg 3.8 ast/g 54.5% TS% 5.2 rTS%

Vs. Good Defenses (26 games, 16.3% of total games):
47.1 mpg 18.6 ppg 24.7 rbg 4 ast/g 53.1 TS%. 4.9 rTS%

Vs. Elite Defenses (53 games, 33.1% of total games):
46.9 mpg 18.9 ppg 24.6 rbg 4.1 ast/g 50.7 TS% 2.1 tTS%

Vs. All Time Great Defenses (19 games, 11.9% of total games):
47.5 mpg 31 ppg 28.6 rbg 2.9 ast/g 53.3 TS 5.5 rTS%

Kareem, Entire Career, Playoffs:

Vs. Bad Defenses (24 games, 9.7% of total games):
32 mpg. 20.8 ppg. 6.5 rbg 2.7 ast. 63 TS% 8.7 rTS%

Vs. Average Defenses (114 games, 48.1% of total games):
35.8 mpg. 23.9 ppg 10.4 rbg 3.1 ast. 59.6 TS% 6.9 rTS%

Vs. Good Defenses (67 games, 28.3% of total games):
38.4 mpg 22 ppg 10.2 rbg 3.6 ast 52.6 TS% 0.7 rTS%

Vs. Elite Defenses (33 games, 13.9% of total games):
42.1 mpg. 29.4 ppg 14.2 rbg 3.8 ast 56 TS% 4.8 tTS%

Vs. All Time Great Defenses (0 games, 0% of total games):
------

Wilt vs. Elite + All Time Great Defenses:
45% of total playoff games played
47.2 MPG
25.0 PPG
26.6 RBG
3.5 AST/G
rTS%: +3.8%

Lebron vs. Elite + All Time Great Defenses:
22.1% of total playoff games played
42.3 MPG
26.3 PGG
7.8 RBG
rTS%: -1.3%

Kareem vs. Elite + All Time Great Defenses:
13.9% of total playoff games played
42.1 MPG
29.4 PPG
14.2 RBG
3.8 AST/G
rTS%: +4.8%

The PS is so matchup dependent and has so many variables (teammates, synergy, coaching, etc.) that career averages are useless. Because of the limited sample size problem, the best way to understand how someone performs in the PS is through analytics + watching the tape.

The only time career averages should be used is when there are no other options, i.e., there is little tape to watch.