In terms of average prime level in terms of offensive value generated, I might go:
Magic
MJ
Lebron
However, if we do in terms of average prime level in terms of "goodness," (so we see them in different situations), I might go in the exact opposite order.
Lebron
MJ
Magic
The reason being, I think Lebron's offensive value was notably muted in Miami versus some of his roles in Cleveland where he functioned more like a PG. This doesn't mean, that Lebron wasn't still extremely good offensively in Miami...he was, however, I believe his overall creation and playmaking value took a dip when playing next to Wade.
Jordan and especially Magic are often cited for being among the GOATs offensively because of his offenses, but I think Lebron has a strong case even when looking at things from that lense.
Looking at PS rORTG since the 60s per Backpicks, Lebron's teams have 1 spot in the top 5 and 2 spots in the top 10. Magic's Lakers (86-87) have 1 entry in the top 20.
PS offensive ratings can be noisy because of small sample sizes but depending on the range you use, it still shows Lebron leading better offenses. For example, Lebron's Cavs-Heat years topped the Magic's Lakers and Jordan's Bulls in 8-year PS offense.
https://thinkingbasketball.net/2018/04/08/backpicks-goat-2-michael-jordan/To me this is noteworthy, because when we are talking greatest offensive players of all-time, a 8 year stretch is good enough to capture someone's prime while not being a long enough span to veer into the territory of longevity (which Magic lacks). Lebron could be argued to have led better offenses throughout his career.
If you look at shorter peak PS stretches such as single year or 3-year PS stretches, I think once again, Lebron comes out looking stronger. According Backpicks, the 2015-17 Cavs have the 3rd best unique offensive PS stretch for relative offensive rating, and keep in mind Lebron did not have a healthy Kyrie or KLove for much of the 2015 PS. This once again surpasses the Bulls and Lakers' offensive performance, and reinforces the idea that Lebron at the PG is probably best served for exerting his full impact.
If you don't like using relative offensive rating to judge playoff offense, there is another method called common offensive rating.
Common offensive rating is comparing a team’s postseason play to other teams against that same given opponent (for that particular PS). The rORTG is also listed on the side too for those who, where a team’s playoff offensive rating is compared to it’s opponent’s regular season defensive ratings. The Cavs have the best common offensive rating of the time period.
The best 3-year offenses and defense (minimum of 20 games played across three postseason trips), we see the following unique team peaks in playoff offense per common offensive rating (cORTG) via Backpicks since 1984 (but only other potential contenders would be if you go back to Mikan days).
Team Year cORTG rORTG
CLE 2015-17 13.0 9.5
MIA 2012-14 9.7 8.7
LAL 1987-89 9.4 9
CHI 1991-93 8.8 8.4
CHI 1994-96 8.3 6.9
Lebron's offenses come out looking better under this approach as well. The 2016 Cavs (+15.3 cORTG) and 2017 Cavs (+14.6) have the two highest single-season offensive marks using this approach. Under the 3-year guise, the Cavs would be at least #1 going back to 1984.
Now, the question might be, "how do we know Lebron had the capacity in Miami to elevate offenses to better heights in Miami then he showed?"
While, YOU DO NOT want to take the numbers at face value, and want to keep in mind we don't always know what players Lebron was facing with Wade off, we see a drastic spike in Lebron's floor-raising performance without Wade on the court.
Right. And in mind my, whose to say that Lebron was stiff in adapting his game. Lebron certainly had more aggression as a scorer when Wade was out.
LeBron in the playoffs with Wade off the court from 2012-14:
▫️ 36.5 PTS/75 on 65.2 TS%
▫️ 7.7 REB/75 and 7.8 AST/75
▫️ Led a +18.1 NRTG outside of garbage time
(stats opponent and inflation adjusted)
And if you want a bigger sample size that has the RS:
12-'14 Lebron without Wade on the floor:
34.4 IA PTS/75 (4th Ever)
+9.5 rTS%
7.2 IA AST/75
39.5% from 3
67.7 Points Generated (2nd Ever)
(3700 Minutes Played Sample)
If you look at Lebron's offensive box-score metrics at face-value, on average he comes out ahead of Magic but BEHIND MJ. Though, I would once again argue the metrics show his value was lesser than Jordan but I'm not arguing Lebron's average offensive prime year was as valuable as Jordan, but rather shifting Lebron and Jordan to a variety of other teams (specifically where Lebron handles the ball more), would show Lebron to be better than his Miami numbers.
I would argue Lebron meshed decently well with Chris Bosh, someone who was a floor-spacer and could be a role man, opening up driving lanes for Lebron. I think Lebron and Wade's styles clashed a bit, to the point where even though they both were good, their offensive impact isn't what it would be in many other situations.
Of course, I could be gauging things incorrectly, and not penalizing Lebron enough for his box-score production not being as great with Wade on. However, I just feel like more times than not, Lebron can elevate great enough variety of teams around the league, that his offensive impact would overall be greater.
My final piece of evidence for my belief that Lebron at the PG position being able to have more value than perhaps Magic and MJ is by just looking at his peak in all-in-one metrics. As I mentioned, MJ on average grades out better offensively than Lebron in an average prime year.
However, if you look at single season peaks (and PS peaks for the metrics that have PS only component) in the following metrics:
O-BPM
Backpicks O-BPM
O-PIPM
O-RAPTOR
O-TWPR
Crafted OPM
Lebron actually surpasses Jordan in all of them, which is perhaps suggestive of the idea that Lebron can elevate teams to special heights with the ball in his hands.
Finally, I would probably give the edge to MJ over Magic in terms of offensive goodness due to being less reliant on his teammates to finish offensive possessions to accrue value. This could be the wrong thinking, but that is where my head is at.