Embiid/Jokic career's RS/PO on/off netrtg

[SpreeS]

One simple question. Why does Jokic impact disappear in PO comparing to RS netrtg? And wow How better looks Embiid than Jokic!!!

RS career

Jokic ON 16017min +5.81 netrtg 116.57 ortg 110.76 drtg
Embiid ON 10241min +8.40 netrtg 114.92 ortg 106.51 drtg

Off numbers looks very similar

Jokic OFF 10828min -4.31 netrtg 107.99 ortg 112.30 drtg
Embiid OFF 12608min -3.06 netrtg 107.98 ortg 111.04 drtg

PO career

Jokic ON 1767min -2.46 netrtg 115.64 ortg 118.10 drtg
Embiid ON 1499min +9.41 netrtg 115.54 ortg 106.39 drtg

Jokic OFF 572min -2.91 netrtg 113.29 ortg 116.20 drtg
Embiid OFF 911min -8.31 netrtg 106.39 ortg 115.15 drtg

Jokic PO on/off looks w/o significant impact comparing with Embiid. Embiid's on/off is crazy good, like Jokic last 3 RS seasons. PHI w/o Embiid is -9.15 on offence and +8.76 on defence (+17.91 on/off). The data covers half of these players career's (27y/28y) so that must mean something.

[falcolombardi]

I have always thought that the narratives against playoff embiid were overstated and is nice that data back it up

[jalengreen]

not surprised, it felt like that was the story of the 2019 sixers. great starting lineup but no backup big, team was atrocious without him on the floor.

[scrabbarista]

Raw on/off can be skewed greatly depending on the quality of the other 19 players in the equation (the four with ON, the five against ON, the five with OFF, and the five against OFF), and the playoff numbers here are a sample size of about half a regular season.

Still, this is a nice little building block if you want to compare Embiid favorably to Jokic. I'd say the numbers here are stark enough that it's definitely not nothing.

[homecourtloss]

One simple question. Why does Jokic impact disappear in PO comparing to RS netrtg? And wow How better looks Embiid than Jokic!!!

RS career

Jokic ON 16017min +5.81 netrtg 116.57 ortg 110.76 drtg
Embiid ON 10241min +8.40 netrtg 114.92 ortg 106.51 drtg

Off numbers looks very similar

Jokic OFF 10828min -4.31 netrtg 107.99 ortg 112.30 drtg
Embiid OFF 12608min -3.06 netrtg 107.98 ortg 111.04 drtg

PO career

Jokic ON 1767min -2.46 netrtg 115.64 ortg 118.10 drtg
Embiid ON 1499min +9.41 netrtg 115.54 ortg 106.39 drtg

Jokic OFF 572min -2.91 netrtg 113.29 ortg 116.20 drtg
Embiid OFF 911min -8.31 netrtg 106.39 ortg 115.15 drtg

Jokic PO on/off looks w/o significant impact comparing with Embiid. Embiid's on/off is crazy good, like Jokic last 3 RS seasons. PHI w/o Embiid is -9.15 on offence and +8.76 on defence (+17.91 on/off). The data covers half of these players career's (27y/28y) so that must mean something.

Embiid has some of the most impressive on-off numbers for any player in the data age, i.e., 1994–current. In each season, he’s been a solid plus on court and even more impressive on-off.

not surprised, it felt like that was the story of the 2019 sixers. great starting lineup but no backup big, team was atrocious without him on the floor.

In the 2019 playoffs, the Sixers were only outscored twice in the entire playoffs while Embiid was on court. Embiid was +20.8!! on court per 100 possessions and +42.4 on-off. Against the eventual champions Raptors, Embiid was:

+10 in 45 minutes in game 7, rest of Sixers -12 in other 3 minutes
+40 in 35 minutes in game 6, rest of Sixers -29 in other 13 minutes
+17 in 35 minutes in game 5, rest of Sixers -22 in other 13 minutes

[Peregrine01]

Jokic's on/off looked horrible against the Warriors this past playoffs despite being the best player in the series. This was because the Warriors purposely played Steph/Draymond whenever he was on the floor. Boogie, his back-up, had a much better looking on/off but shared very few minutes against Steph and Draymond.

In short, match-ups are everything in the playoffs and raw on/off isn't as informative in a condensed sample size. That said, Jokic only emerged recently as a top-5 player in the world and hasn't had a long playoff run since that happened.

[Owly]

Why does Jokic impact disappear

This is (as presently phrased) a leading question, assuming a premise not established.

The data covers half of these players career's (27y/28y) so that must mean something.

The playoff part covers, per your numbers off amount to 572 minutes for Jokic. This typically comes across not across evenly distributed samples (teams don't play the same number of games in each series nor retain the same ratio of minutes on and off). In short the noise that would already be there with a single season's plus-minus - and we don't even have that much data yet - is concentrated by a less balanced schedule (and we're now trying to measure multiple versions of the player at once). In short, I'd be wary of drawing strong conclusions off playoff on-off (without extraordinary results suggestive of signal - and even more so when the conclusions don't align with RS or box performance) I'd be more so doing so on partial careers ...

Thus I'm unclear as to why this "must mean something." at least on the Jokic side (Embiid's degree of positivity is enough to suggest some signal though I don't have the expertise to properly contextualize, quantify, give margins of error etc - nor to dive deep into 76ers and opponent lineup data).

None of which is to say playoff plus/minus can't be useful (if someone says "Team X won because of player Y" and they're negative [or less positive than without] over the series and most games - that argument loses traction - if someone says "player Y is great because they won" and the above circumstances ... again it really harms that argument). But I don't like the certainty that a noisy metric, with likely unbalanced samples and small samples (going in the opposite direction to box composites, and RS impact [and RS box composites] "must" be signal rather than noise.

[capfan33]

While raw on/off in a big enough sample definitely means something, I don't think it means a whole lot in this case. 1st, 570 minutes is not that much. Especially when presumably a large portion of those minutes came during the Warriors series as Peregrine stated. Definitely need some adjustments for lineups and a bigger sample size.