Hakeem Olajuwon’s impact by numbers: peak (93 - 96) vs non-peak
Olajuwon’s famous for a huge peak from 1993 - 96, when he won both of his titles during the Jordan retirement years. But the question is, was he playing at that level all along but no one noticed it because his supporting casts were weaker? The answer to that question can help evaluate Olajuwon’s longevity as a mega impact player, which can help estimate the overall value of his career. Let’s explore.
WOWY
The With Or Without You measure (WOWY) calculated by Ben Taylor (El Gee) gives us some insight into how much impact players from the pre-databall era were having. It’s a measure for how a team plays in the games that a player plays, vs how they play when the player misses time. Olajuwon missed double-digit games in three time windows during his rough prime window of 1986 - 1996:
*1986: Olajuwon missed 14 games. ElGee’s WOWY spreadsheet accounts for 13 of those games. He measured the change in SRS of the Rockets to be +1.9 when Olajuwon was in, vs those 13 games when he was out. The team’s ultimate SRS was 2.1.
1991 & 1992: Olajuwon missed 26 games in 1991 and 12 games in 1992. ElGee’s WOWY spreadsheet measured the change in SRS of the Rockets to be +3.8 when Olajuwon was in vs those 38 games when he was out, and he calculated that the SRS of the team over those two seasons was 2.0.
1995 & 1996: Olajuwon missed 10 games in 1995 and another 10 in 1996. ElGee’s WOWY spreadsheet accounts for eight of the missed games in 95 and all 10 in 96. Unfortunately, he doesn’t have them grouped here for one number. In 95, the SRS of the Rockets was +6 higher and in 96 it was +9.4 higher with Olajuwon than in the games that he was out.
(Disclaimer: someone (I think Blackmill) performed simulations that he believes shows that WOWY is noisy for the order of 10 - 40 games missed, so take the above with whatever size grain of salt you desire)
On/off +/-
We have that data for 1994, 1995 and 1996 (nominally Olajuwon’s peak)
1994: Rockets +7 on, -7.5 off –> +14.5 net, 4th in NBA (Robinson 1st, +19.9)
1995: Rockets +5.6 on, -6.3 off –> +11.9 net, 7th in NBA (Robinson 1st, +19.8)
1996: Rockets +4.9 on, -5.4 off –> +10.3 net (unknown rank, Robinson 1st, +16.6, Jordan +15.2)
Quick thoughts on those impact results
*It is generally accepted that Olajuwon hit an absurd peak between 93 - 96, a peak that was historic level. The question that a lot of people have, then, is how different was “peak” Olajuwon from “pre-peak” Olajuwon. There is a sentiment, among his supporters, that while he may have improved a bit in 93, that he was similarly impactful before 1993 but had a team that was too poor to show it. Being a Garnett supporter, I’m of course familiar with that argument
*The WOWY data suggests (to the extent that you find WOWY on 10 - 40 games in 1 to 2 year periods useful) that
Olajuwon’s impact was not as large in his ‘86 or '91-'92 absence as it was in his '95-96 absences
.*Comparatively speaking, early Olajuwon’s impact per WOWY was more similar to 1965 Wilt (when he went from the Warriors to the 76ers mid-season with neither team changing much in SRS).
*However, later Olajuwon (1996), per WOWY, had a much larger impact (on the order of 2001-02 Shaq or 2005 Duncan, even 1978 Walton)
*Olajuwon’s 1994 - 1996 on/off +/- scores generally support his 1996 WOWY score…they also support that Olajuwon was having an impact in those mid-90s years that was among the best in the NBA in those seasons.
What may have contributed to Olajuwon’s impact pre-1993 not being as large as during his 93 - 96 peak?
I’ve got two potential reasons: 1) More complete offense at his peak and 2) More intelligent defense at his peak. Let’s start with the boxscores, and go from there.
Hakeem’s boxscore
1986 - 92: 23.2 pts (51% FG, 71% FT), 12.6 reb, 2.3 ast/3.2 TO, 2.1 stl, 3.7 blk
1993 - 96: 27 pts (52% FG, 74% FT), 11.7 reb, 3.6 ast/3.3 TO, 1.7 stl, 3.6 blk
1) More complete offense.
There aren’t huge differences in the boxscore, a bit each way. Olajuwon was scoring on a bit higher volume, about one rebound less, miniscule differences in steal/block numbers
The one area that stands out to me, though, is the seemingly small difference in assists. See, up until 1993, Olajuwon never had a single season where he had as many assists as he had turnovers. Then, from 93 - 96, he had more assists than TOs every season. Here’s why I think that’s important:
From the databall era +/- stats (1997 - current), the regressed data suggests strongly that volume scoring (even on good efficiency) is NOT sufficient for strong offense impact. No, for volume scorers, passing/assists correlates more with positive offensive impact than scoring efficiency. There’s a growing amount of research indicating this, but here’s a quote from one article from Andrew Johnson on Nylon Calculus (note: here's the link to the article. Really worth checking out https://fansided.com/2016/11/28/nylon-calculus-shooting-shot-creation-variance/ ):
“The higher COV on passing efficiency within positions indicates that passing efficiency is a measure with some real separation in talent. Further, the outliers on the upper end tend to line up with the players we think of as stars.”
“In fact, Justin Willard’s research found that there is an interactive quality between scoring and passing proficiency; being a superior passer increases the impact of a player’s scoring and vice versa.”
For a center, getting up near 4 assists per game vs the just over 2 assists pre-peak pushes Olajuwon out towards the “position outliers” that Johnson describes. In some of my own research, it seems that for bigs, getting to that assist/TO ratio over 1 is almost like a Magic number. It’s not published, but it’s very clear when you go through the RAPM studies for the last 20 years. It’s clear even when you go through the subjectively decided offensive peaks of the great big man scorers…Kareem, Shaq, Robinson, Dirk, Duncan…look at the seasons that you believe to be their best offensive seasons, and universally they’re the ones with assist/TO ratios over 1.
Olajuwon became a more intelligent player at his peak, he became a better passer, and it allowed the Rockets to play that 1 star with 4 shooters offense, which devastated the league. I believe that this helps to explain why Olajuwon’s impact at his peak was just better than previously.
2) Smarter defense. This one I’m hand-waving, which hopefully can be somewhat forgiven at the end of this marathon post. General consensus is that Hakeem’s defensive peak came just before his overall peak, and that 1993 may have been the last of his true “defensive peak” years. That’s reasonable. By his offensive peak, he was no longer at his athletic peak, and we did see a small dip in his rebounds and steals (not much in his blocks).
But, I would also say that, in general, players become smarter defenders (and players in general) as they get older. And while Hakeem was no longer at his athletic peak in the mid-90s, he was still at the upper end of the elite as far as athleticism for a center. Plus, and this shouldn’t be overlooked, but the Rockets had intelligent 4s like Otis Thorpe and young Robert Horry, which maybe allowed Olajuwon to emphasize his strengths (e.g. post-defense and rim protection) without having to cover quite as much land as he had to previously. Somewhat like Wilt, Duncan, Robinson and Garnett…they may have lost some athleticism off their defensive fast ball, but smarts seemed to allow them to either maintain or even perhaps have larger measure impacts as old guys. I’m going to postulate that maybe peak Hakeem had some of that as well.
Conclusions
*Hakeem was an athletic freak that could score in a dominant way and fill up the defensive box scores for almost his entire career
*Hakeem didn’t measure out with elite WOWY scores in his absences in 1986 or 1991-92.
*Hakeem DID measure out with elite WOWY scores, as well as elite on/off +/- scores during his 1994 - 96 peak years
*Hakeem clearly became a better passer and better overall offensive hub in his peak years than he was previously, both by the eye test and it can somewhat be seen in the boxscore. I contend that offense creation/passing is a MUCH more important factor in offensive impact than general consensus or the boxscore composite stats account for, and that this marked an important increase in Olajuwon’s offensive impact
*The consensus is that Hakeem’s defensive peak was just prior to his offensive peak, and I can see that. However, experience breeds defensive intelligence and often leads to extended (maybe even increased) defensive impact even as the athletic ability starts to wane. This seems to be especially true for the elite bigs that maintained strong physical advantages even when they lost top athletic ability, and I contend that it could be true for Hakeem as well
Thus, it DOES appear that a) peak Hakeem (1993 - 96) was a distinctly bigger impact player than pre-1993 Hakeem.
Bringing this around to the vote (2020)
Olajuwon was an amazing player, and he peaked as an all-history player. But, I think only at his peak (93-96) was Olajuwon's on-court impact comparable to what we saw from Garnett's extended prime. Olajuwon was always the better ISO scorer, and he ramped that ability up in the playoffs. But, generalizing away from Olajuwon and Garnett specifically, we've got lots of data showing that ISO/volume scoring (particularly from big men) just isn't as impactful as we used to think it was. It just isn't. Now, getting back to specifics, Olajuwon's monster ISO offense was a big part of the two rings he won. And that absolutely has to be a big factor in this comp.
But not the only factor. Garnett was having the impact Olajuwon only seemed to reach in his peak, perhaps in his peak AND the playoffs, but for around a decade. And, contrary to the narrative, Garnett's outsized impact continued through the playoffs at an extremely high, repeatable level for a similar decade-plus.
And I've seen some in this thread and the last posit that Olajuwon's peak impact should trump Garnett's unique combo of portability and scaleability, but I'd argue that those latter elements are a big reason WHY Garnett's prime- and career- impacts were larger than Hakeem's despite their peaks being comparable: Hakeem needed more of a sweet spot (e.g. improving his passing, being surrounded by shooters, Rudy T's system, perhaps the closer 3-point line) in order to reach that maximized impact, whereas Garnett was able to have a similar level of impact in a mad variety of roles on the Timberwolves and Celtics. Garnett finished top-5 in RAPM in the late 90s as a do-everything-guy and finisher next to great PGs, finished consistently near the top as the primary offensive and defensive engine for the Wolves in the first half of the 2000s, finished at the top as the dominant defensive engine and finisher for the late 00s Celtics, and maintained his spot as a top-5 RAPM finisher out into the early 2010s with the Celtics as a dominant defender and good offensive floor spacer/passer.
Thus, Vote:
1) Kevin Garnett
2) Hakeem Olajuwon
3) Larry Bird (for now, anyway. Still haven't explored much past the top two slots in this project)