1968 Wilt passing issue (estimated ORtg and DRtg for 1967/68 Sixers)

[70sFan]

Hi, we had a long discussion about Sixers estimated ORtg and DRtg across the 1967/68 season in Wilt vs Jokic thread. I decided to create a new thread, because I calculated pace estimates and ratings adjusted for opponents.

I took my pace estimation formula from this article (used by Basketball Reference):

https://fansided.com/2015/12/21/nylon-calculus-101-possessions/

Here is the formula:

Code: Select all

poss=0.5 * ((Tm FGA + 0.4 * Tm FTA – 1.07 * (Tm ORB / (Tm ORB + Opp DRB)) * (Tm FGA – Tm FG) + Tm TOV) + (Opp FGA + 0.4 * Opp FTA – 1.07 * (Opp ORB / (Opp ORB + Tm DRB)) * (Opp FGA – Opp FG) + Opp TOV))

Pace=poss/(2*TM/5)*100


I used estimates for TOV% and ORB% suggested by Ben Taylor in one of his old articles (30.2 ORB% and 16.1 TOV%), so it doesn't give us a perfect value of course. Using this formula, I got Sixers pace at 124.785, which is slightly different than 124.3 value from bball ref (could be related to my TOV% and ORB% estimations). I think it shows that my calculations works well enough to give an error not much bigger than 2 possessions at most.

When I looked at gamelogs, I realized that Sixers scored far less in the first 30 games compared to the rest of the season. Here are the numbers:

First 30 games: 117.7 Tm, 111.1 Opp
Last 52 games: 125.4 Tm, 115.6 Opp

I suggested that Sixers slowdown in 1967/68 season was strongly related to Wilt's poor beginning of the season, but it had little to do with his assist hunting or the rest of the league realizing how to guard Sixers offense. Let's take a look at Wilt's numbers in the first 30 games and the last 52 games of the season:

First 30 games: 19.0 ppg, 24.1 rpg, 7.0 apg on 52.4 FG%, 34.7 FT% and 49.1 TS%
Last 52 games: 27.4 ppg, 23.6 rpg, 9.5 apg on 62.8 FG%, 39.7 FT% and 59.0 TS%

Individual production usually isn't strong enough evidence. Besides, there are often pace differences across different parts of the season. I decided to run pace calculations for these two periods, here are results:

First 30 games: 124.0
Last 52 games: 125.1

Ignoring the error, it gives us basically no difference in pace between these two periods. Using these pace estimates, we can get Sixers ratings in both periods:

Full season: 98.3 ORtg, 91.4 DRtg
First 30 games: 94.9 ORtg, 89.6 DRtg
Last 52 games: 100.2 ORtg, 92.4 DRtg

Note that my ratings are slightly different than official ones, because of minor difference in my calculated pace for full season. Another important thing is adjusting your ratings to opponents faced. I also did that:

Full season: 96.7 ORtg, 97.3 DRtg
First 30 games opponent ratings: 96.5 ORtg, 97.2 DRtg
Last 50 games opponent ratings: 96.8 ORtg, 97.3 DRtg

As you can see, differences aren't big, but I included them in the next step - calculating relative ratings:

Full season: +1.0 rORtg, -5.3 rDRtg
First 30 games opponent ratings: -2.3 rORtg, -7.0 DRtg
Last 50 games opponent ratings: +2.9 ORtg, -4.4 DRtg

The difference in full seasons numbers are caused by different methodology - basketball reference use league average to get relative numbers. I used the difference in average opponent ratings and Sixers ratings. As you can see, Sixers were excellent both offensively and defensively in the majority of the season. Now, if I actually use basketball reference definition of relative ratings and not the one I used, the numbers would look this way:

Full season: +1.5 rORtg, -5.4 rDRtg
First 30 games opponent ratings: -1.9 rORtg, -7.2 DRtg
Last 50 games opponent ratings: +3.4 ORtg, -4.4 DRtg

Bonus!

I also did the same thing to the last 2 months of the season for the Sixers. It was the time when Wilt averaged staggering 11 apg. Some people accused him of being a statspadder and they implied that Sixers offense didn't reach their potential because of that. Let's see what my calculations show:

Last 29 games:

Pace: 125.03
ORtg: 101.7, +4.3
DRtg: 92.0, -4.3

As you can see, no visible reason to believe that Wilt high assist numbers hurt Sixers offense, quite the contrary in fact.

[70sFan]

My question - how do you interpret these numbers? Do you think that they show something that is the opposite from well established notion on the PC Board that Wilt's assist obsession hurt Sixers offense? Do you think that there is no correlation between Wilt's production and Sixers offense? Do you think that we shouldn't look at such small sample of size and it's just a noise?

[Owly]

68 76ers offense - Wilts assists don’t correlate with early season offensive dip

So my gut reactions are, without fully knowing the earlier theories of others (nor diving that deep into the numbers nor thinking too deeply – very otoh):

It substantially harms any idea that Wilt was “solved” at this time.
The early season slump did still happen and is rightly part of those full season numbers.
Cause and effect are complex.
All being equal, the Sixers scoring more, Wilt’s assists should go up.
So we’re trading off between that “Wilt assist hunting is harming the offense” is decidedly not holding all else equal and “of course in the spell the 76ers scored more he got more assists. That doesn’t mean that the reduction versus the previous year isn’t him seeking the quick shot pass rather than the best one.”
I don’t know what the assists increase should be all else equal, though maybe that’s easily do-able?
I tend to think it harms the argument that he was assist hunting in a manner detrimental to the team (I have, fwiw, iirc, in the past, noted that the assist increase just isn’t that big, doesn’t seem like a substantial enough increase to indicate a substantial change in approach, nor were the biggest assist totals racked up early on – when Robertson’s absence might have motivated him to try to rack up an insurmountable lead; official leaders being recognized by totals back then). But again how much etc … you’d want to look closer.

Stuff that might be interesting
How other players shot splits.
4 factors (don’t know how possible this is) splits
Schedule – e.g. home-road proportions. Strength of schedule. Proportion of Boston games (maybe SF too if it’s Wilt’s shooting that went down early on [semi-tangent: as I thought Thurmond missed a fair chunk of that season, his final game occurring January 19th]).
How much the fall versus previous year is him/his free throw shooting. His TS add is down -228.8, of which -146.2 is from the field and -82.6 is from the line.
Are any circumstances documented as different. They were better with Costello the previous year (iirc) and his numbers this year might be better, but he plays 28 games up to early December, though it's a fringe role. Is this anything more than noise?
Tangential stuff that may affect things (team offensive numbers) at the margins that I sometimes think about with earlier eras
Missing turnover data (from a super-modern perspective regarding player evaluation, though I don’t tend to see these numbers, we're missing time on ball data).
Differences in free throw rules mess with average FT possession value and with the specific player possession consumption (when were 3-to-make-2 [was 1+1ever NBA – I can’t see that I was] implemented, which situations were they applied to and when were they junked - and getting a second shot at all, was that early/mid-50s, in part after stallball?).

[ty 4191]

This is (slightly) off topic, but I'm just curious...(and I'm a total novice when it comes to basketball, stats, NBA history)....

Why are 17 of the top 20 team dRTG seasons of all time come from the 1970's?

[70sFan]

It substantially harms any idea that Wilt was “solved” at this time.

That's my thought as well. It's just hard to intepret these numbers that way.

The early season slump did still happen and is rightly part of those full season numbers.
Cause and effect are complex.

I agree and I want to point out - I don't want to create excuses for Wilt here. We should criticize him for a weak start, but as you know - the interpretations vary a lot and I want to eliminate the least probable ones.

All being equal, the Sixers scoring more, Wilt’s assists should go up.
So we’re trading off between that “Wilt assist hunting is harming the offense” is decidedly not holding all else equal and “of course in the spell the 76ers scored more he got more assists. That doesn’t mean that the reduction versus the previous year isn’t him seeking the quick shot pass rather than the best one.”
I don’t know what the assists increase should be all else equal, though maybe that’s easily do-able?

If we give Sixers FG% of the last 52 games (48.4 FG% vs 44.9 FG% in reality), then Wilt's assist numbers would go up to 7.5 apg. Of course we should also exclude Wilt's FGs from the equation, but the difference would be even lesser then.

I tend to think it harms the argument that he was assist hunting in a manner detrimental to the team (I have, fwiw, iirc, in the past, noted that the assist increase just isn’t that big, doesn’t seem like a substantial enough increase to indicate a substantial change in approach, nor were the biggest assist totals racked up early on – when Robertson’s absence might have motivated him to try to rack up an insurmountable lead; official leaders being recognized by totals back then). But again how much etc … you’d want to look closer.

I agree, we don't have any evidences of this small increase hurting Sixers offense.

Stuff that might be interesting
How other players shot splits.
4 factors (don’t know how possible this is) splits
Schedule – e.g. home-road proportions. Strength of schedule. Proportion of Boston games (maybe SF too if it’s Wilt’s shooting that went down early on [semi-tangent: as I thought Thurmond missed a fair chunk of that season, his final game occurring January 19th]).
How much the fall versus previous year is him/his free throw shooting. His TS add is down -228.8, of which -146.2 is from the field and -82.6 is from the line.
Are any circumstances documented as different. They were better with Costello the previous year (iirc) and his numbers this year might be better, but he plays 28 games up to early December, though it's a fringe role. Is this anything more than noise?
Tangential stuff that may affect things (team offensive numbers) at the margins that I sometimes think about with earlier eras
Missing turnover data (from a super-modern perspective regarding player evaluation, though I don’t tend to see these numbers, we're missing time on ball data).
Differences in free throw rules mess with average FT possession value and with the specific player possession consumption (when were 3-to-make-2 [was 1+1ever NBA – I can’t see that I was] implemented, which situations were they applied to and when were they junked - and getting a second shot at all, was that early/mid-50s, in part after stallball?).

I'm afraid that I won't be able to adjust for all of these things. I agree that there are a lot of uncertainties here, as always with 1960s debates.

Thank you for very detalied reply!

[ShotCreator]

It substantially harms any idea that Wilt was “solved” at this time.

That's my thought as well. It's just hard to intepret these numbers that way.

The early season slump did still happen and is rightly part of those full season numbers.
Cause and effect are complex.

I agree and I want to point out - I don't want to create excuses for Wilt here. We should criticize him for a weak start, but as you know - the interpretations vary a lot and I want to eliminate the least probable ones.

All being equal, the Sixers scoring more, Wilt’s assists should go up.
So we’re trading off between that “Wilt assist hunting is harming the offense” is decidedly not holding all else equal and “of course in the spell the 76ers scored more he got more assists. That doesn’t mean that the reduction versus the previous year isn’t him seeking the quick shot pass rather than the best one.”
I don’t know what the assists increase should be all else equal, though maybe that’s easily do-able?

If we give Sixers FG% of the last 52 games (48.4 FG% vs 44.9 FG% in reality), then Wilt's assist numbers would go up to 7.5 apg. Of course we should also exclude Wilt's FGs from the equation, but the difference would be even lesser then.

Where does passing quality and shot quality come in on something like this? Why do you think you can assume every possession is static to the degree of assuming the same kind of passes would be made to the same players, even though the actual results are much different?


Why is Wilt’s much better scoring in the cherry picked sample not the talk of this thread? That is inherently connected to any passes he would make because of how the defense would react to a below average scorer vs a decent one.

You think Wilt was created the same looks for his teammates at 19 points on 49 TS% as 27 on 59 TS%?

[feyki]

My question - how do you interpret these numbers? Do you think that they show something that is the opposite from well established notion on the PC Board that Wilt's assist obsession hurt Sixers offense? Do you think that there is no correlation between Wilt's production and Sixers offense? Do you think that we shouldn't look at such small sample of size and it's just a noise?

Don't think hurt, but also don't think added much impact. With just improving PPG by 8 and TS by %10 adding more than 5 TmORTG, wouldn't say more assists hurt but at least, seems not much added anything.

[70sFan]

Where does passing quality and shot quality come in on something like this?

That's very important point and I have no answer for that. Without extensive tracking database or big collection of footage from that season, it's impossible to answer.

Why do you think you can assume every possession is static to the degree of assuming the same kind of passes would be made to the same players, even though the actual results are much different?

I can't do that and I don't assume it. Maybe I missed something, but I said that it's impossible to estimate the value of these passes and the difference in Wilt's playmaking value without footage.

Why is Wilt’s much better scoring in the cherry picked sample not the talk of this thread?

Why do you think it's cherry picked sample? It's basically two-third of the season and I didn't try to pick the best sample possible for Wilt - I just realized that Sixers offense was much worse in the beginning of the season than usually and I found out that Wilt had a bad scoring streak.

Also, Wilt's bad scoring streak is literally the talk of this thread. A lot of people assumed that Sixers offense got worse because of Wilt's statpadding his assist numbers, but I propose different explaination. I think that Sixers offense regressed because of Wilt's poor beginning of the season and it came back to their 1967 level after Wilt came back from this shooting slump.

That is inherently connected to any passes he would make because of how the defense would react to a below average scorer vs a decent one.

Again, I don't think it's possible to prove without footage, but I don't think teams treated Wilt like below average scorer in sampled size. A good analogy would be Curry's poor scoring stats from last few months - nobody treats Curry like below average scorer simply because he has been missing his shots. That's not how basketball works.

You think Wilt was created the same looks for his teammates at 19 points on 49 TS% as 27 on 59 TS%?

Probably not, which again - proves my point that the league didn't "solve" Sixers offense, but they struggled simply because Wilt had a down moment. Once he came back to his usual self, Sixers dominated just like the last season.

[Mazter]

Wilt's improvement in scoring itself was a +4.0 boost for the Ortg. His teammates did the rest. Could be due to Wilt's passing, but the next year they shot about the same from the field without him.

[70sFan]

Wilt's improvement in scoring itself was a +4.0 boost for the Ortg. His teammates did the rest. Could be due to Wilt's passing, but the next year they shot about the same from the field without him.

So you agree that Wilt's passing wasn't the problem? When he scored on his usual level, Sixers played on the same level they did in 1967.

[DQuinn1575]

In 67, (and up until the early 70s) 1 free throw was shot on every foul - you didn't wait until the 5th foul of the quarter.
Pretty sure the NBA never had a 1 and 1, but they did have 2 to make 1 and 3 to make 2 - I didn't check and 67 is before my time, so I dont know what the rule was then.
Bottom line is that possessions are misstated a little, as the factor for free throws is not correct. It won't matter much to what you are doing here - comparing within the season - as your numbers are consistent.

[70sFan]

In 67, (and up until the early 70s) 1 free throw was shot on every foul - you didn't wait until the 5th foul of the quarter.
Pretty sure the NBA never had a 1 and 1, but they did have 2 to make 1 and 3 to make 2 - I didn't check and 67 is before my time, so I dont know what the rule was then.
Bottom line is that possessions are misstated a little, as the factor for free throws is not correct. It won't matter much to what you are doing here - comparing within the season - as your numbers are consistent.

That's true and I can't adjust for these small, but important changes in rules. As you noted, it shouldn't change the relative differences though.