playoff risers

[Throwawaytheone]

.

[ShaqAttac]

Why do people worry about arbitrary "Rising" or "Falling" instead of evaluating based on how good they are in the RS or PS?

isnt ps just rs+rise or fall?

[Throwawaytheone]

.

[Throwawaytheone]

.

[OhayoKD]

Why do people worry about arbitrary "Rising" or "Falling" instead of evaluating based on how good they are in the RS or PS?

isnt ps just rs+rise or fall?

No, not really. A postseason performance is just how you played within the confines of a playoff setting. Thats it. Did they play good? Did they play bad? That stuff.

Rising and falling is comparing their PS performance to their RS performance and evaluating off of that (or incorporating that into your overall PS evaluation). It's just dumb as hell, because a fringe all star could rise to being a 2nd team All NBA player in the PS and a strong MVP player could fall to simply being a weak MVP player and this "Rise/fall" talk would have you thinking the former is a legendary clutch killer and the latter is a choker. If the first dude set a higher bar for himself, it wouldn't be a convo, but people literally reward him for playing worse in the RS.

And, "How well you performed" can be seen as "rs performance+the difference between rs and playoff performance". That is an extra step, but it has the benefit of letting us use the larger regular-season sample to inform playoff assessments.

Paticularly valuable when you are trying to get to tie players to their ability to effect wins.

Players should not be rewarded for setting a lower bar no(cough kawhi cough shaq). Players should also not be penalized for setting a higher regular season bar(cough kareem cough steph). But if a player sets a similar bar and rises more, then it's fair to favor player 2(this is 15-17 lebron's case vs 15-17 steph more or less).

Of course it's preferable to cut out the middle-man and just assess the playoff goodness directly but that's not so convenient when you're trying to spin Kareem's 74 run as a negative or rank Duncan lower than Hakeem and Shaq because he won too much in the regular season...

[Bad Gatorade]

Why do people worry about arbitrary "Rising" or "Falling" instead of evaluating based on how good they are in the RS or PS?

I think simply criticising/applauding a player on rising or falling in the postseason, in itself, often creates biases where we "like" a player more for lifting their game on the biggest stage, and "dislike" chokers, so in this sense, I absolutely agree with you, especially when being a good regular season player lends itself to a higher seed, which, in turn, lends itself to worse postseason competition on average (ergo, greater chance to win a championship).

I'm a believer in impact stats. I don't believe that players drop that much in the playoffs. I believe in playoff risers slightly more (e.g. LeBron, post-2013, clearly seems to hit a new gear in the playoffs IMO). However...

Impact stats alone are flawed in the playoffs -
* Even 2 seasons of regular season data is often insufficient for calculating a reasonable RAPM, and only 35 players ever have accrued 164 playoff games
* Shorter rotations (i.e. less lineup data to use)
* Players that play a lot may have more minutes coinciding with the same players (i.e. even less lineup data to use)
* Players may fluctuate across just a few seasons, and so we lack valuable data for honing in on specific, smaller time frames

Does this mean they're useless? No.
Does this mean they're holistic? No.

Consider a player like Draymond - his box score stats in the regular season do not accurately convey his value. His box score increases in the postseason, where he is already highly successful and Golden State have historically retained a great defensive presence in the playoffs. He also has some elite postseason impact stats based on the (admittedly limited) RAPM we have for the playoffs. Now, if somebody views Draymond as, say, a +5 in the regular season (where we have ample data), looking at the box score gives valuable credence in assuming he is better than a +5 in the playoffs. IMO, that is a very reasonable (and IMO, the more correct) conclusion to reach.

Kyle Lowry, IIRC, has a similar profile of looking better in impact stats than in the box score, and looks terrific in the postseason plus minus world. However... Kyle Lowry is also known for having box score drop offs for a team that often "flakes out" of the postseason prematurely. If you look at the playoffs, Lowry's On/Off is +13.2 with Toronto, and it looks even more drastic once you consider how much time he spends playing with DeRozan (who is one of the poster boys for playoff drop offs in almost every sense of the word). Now, depending on how you assess Lowry, we can reach a bunch of conclusions -

* Looking only at box score stuff in the playoffs makes him look like a borderline starting PG
* Looking only at impact stuff in the playoffs makes him look absurdly good
* Looking at both box score/impact stuff paints a very mixed picture, and makes it hard to assess (it's almost a "pick your poison" at this point)
* Looking only at impact increases make him look even more absurdly good

Now...

Considering Lowry as an impact guy (rather than a box score guy) and seeing his impact in the postseason rise but also acknowledging a fairly clear box score decrease may lend itself to the conclusion that Lowry's offence looks far less productive in the postseason, but he still does a lot of good non-box score work that helps retain a lot of value (if not be slightly more valuable in those respects). This seems far more reasonable than anything in the above 4 dot points, and it's a conclusion that can only be reached by looking at an extensive regular vs postseason comparison.

(Note that all of this should clearly be corroborated via the eye test, but that should also go without saying, right? )

So, I do think that looking at box score drop offs as portrayed in the OP has value in terms of assessing true playoff performance. However, I do agree with you in that true playoff performance (however we choose to establish it) shouldn't be rewarded/punished based on a regular season prior.

[OhayoKD]

Why do people worry about arbitrary "Rising" or "Falling" instead of evaluating based on how good they are in the RS or PS?

I think simply criticising/applauding a player on rising or falling in the postseason, in itself, often creates biases where we "like" a player more for lifting their game on the biggest stage, and "dislike" chokers, so in this sense, I absolutely agree with you, especially when being a good regular season player lends itself to a higher seed, which, in turn, lends itself to worse postseason competition on average (ergo, greater chance to win a championship).

I'm a believer in impact stats. I don't believe that players drop that much in the playoffs. I believe in playoff risers slightly more (e.g. LeBron, post-2013, clearly seems to hit a new gear in the playoffs IMO). However...

Impact stats alone are flawed in the playoffs -
* Even 2 seasons of regular season data is often insufficient for calculating a reasonable RAPM, and only 35 players ever have accrued 164 playoff games

For "playoff impact" I think it is a lot more useful if you take the rs for your off and adjust from there. Consistent team-wide improvement is indicative I think, especially if it tracks with a jump in production, lineup-ratings, ect.

That being said, an issue here is treating RAPM as a sole impact indicator. You should always be using real-world signals, especially with shorter samples

[Bad Gatorade]

For "playoff impact" I think it is a lot more useful if you take the rs for your off and adjust from there. Consistent team-wide improvement is indicative I think, especially if it tracks with a jump in production, lineup-ratings, ect.

That being said, an issue here is treating RAPM as a sole impact indicator. You should always be using real-world signals, especially with shorter samples

As do I - I think that an adjustment is generally the "better" option when anything playoffs-related is concerned.

I think exceptions can be made depending on the situation - for example, if a LeBron has "proven" himself already, starts to coast in the regular season but maintains (if not improves) his production in the playoffs, I'd be inclined to think that his playoff performance is entirely real.

But yeah, the key takeaway is that the regular season provides a lot of useful information, and a holistic attempt at seeking "true" playoff performance is likely to use priors for the regular season.

As for using RAPM as a sole impact indicator, I do agree with you in that other signals/philosophies should come into play, although I do think quite highly of RAPM as a statistic. Although, I'd note that's probably slightly orthogonal to what the "playoff riser" discussion is getting at - not that it shouldn't be considered, but rather that the verity of RAPM as an impact measure probably deviates into an entirely different territory.

[OhayoKD]

For "playoff impact" I think it is a lot more useful if you take the rs for your off and adjust from there. Consistent team-wide improvement is indicative I think, especially if it tracks with a jump in production, lineup-ratings, ect.

That being said, an issue here is treating RAPM as a sole impact indicator. You should always be using real-world signals, especially with shorter samples

As do I - I think that an adjustment is generally the "better" option when anything playoffs-related is concerned.

I think exceptions can be made depending on the situation - for example, if a LeBron has "proven" himself already, starts to coast in the regular season but maintains (if not improves) his production in the playoffs, I'd be inclined to think that his playoff performance is entirely real.

Importantly that production increase is consistently reflected in team-wide improvement. In fact, despite narratives that would suggest the opposite, his teams getting better in the playoffs has almost always been true even going back to his first couple of runs. I think, the 2010 ecsf and the 2011 finals have an outsized influence on perception here. In general, for his career, Lebron's teams jump in a way only really rivalled by Hakeem(among mvps anyway)

Lebron has the most srs upsets and the best winning% as an srs underdog(hakeem is the other one). If you go by sans rolling srs even those miami teams were big playoff-risers. It just hits another level in his cleveland return. That it is paired with historic box-improvement increases confidence too. That mark in the OP is career-wide for a very long career. Am curious what it would be if you just looked at the 3 years he was "coasting"(and still putting up some of the most valuable regular seasons ever...)

Hakeem also sees big time-wide improvement(50 to 60+ in his "prime, big spikes in 86-88) and has the second most wins and second best win% as an srs underdog. So I'm pretty confident his increase is "real". 86 might be the biggest example of elevation ever with the Rockets going from +2(+3 net with Hakeem in the lineup tbf) to, whether you use rs srs or rolling srs to scale, a solid championship-level performer statistically:


One player who I'm curious about is Wilt who may not look great by fp4's measure, but also sees a +2 point defensive spike for his playoff teams. Those 2 points outweigh the 1-point offensive drop-off so I wonder if Wilt should be considered a playoff-riser regardless.

Russell is interesting because he's at the top of this list, he outperforms "expected championships" by a shitton despite setting a very high bar and...he also seems to get taken to 6 or 7 a bunch.

But yeah, the key takeaway is that the regular season provides a lot of useful information, and a holistic attempt at seeking "true" playoff performance is likely to use priors for the regular season.

As for using RAPM as a sole impact indicator, I do agree with you in that other signals/philosophies should come into play, although I do think quite highly of RAPM as a statistic. Although, I'd note that's probably slightly orthogonal to what the "playoff riser" discussion is getting at - not that it shouldn't be considered, but rather that the verity of RAPM as an impact measure probably deviates into an entirely different territory.

[/quote]
Fair. I also think highly of RAPM, but I think it needs to be used with itsw limitations in mind. I've seen alot of people use it like: "player a's season is = player b's season because both score at a similar spot at the top of the set" and that is not how you should be analyzing impact.

That sort of parity is artificial and while it helps mitigation rotation whatever, it is still an approximation. You should look at what lineups players are sharing minutes with, see what happens when you remove a player from a team, track granular stuff, ect, ect. I think I saw someone here argue steph>nash because steph's best 1-year orapm was higher and that's just...

[f4p]

One player who I'm curious about is Wilt who may not look great by fp4's measure, but also sees a +2 point defensive spike for his playoff teams. Those 2 points outweigh the 1-point offensive drop-off so I wonder if Wilt should be considered a playoff-riser regardless.

wilt is also really bad at actual vs expected titles at -1.04. not only is that 95th out of 103, but he's the only person with 2 titles who is even negative at all, so being 95th out of 103 is even more damning with 2 titles.

[OhayoKD]

One player who I'm curious about is Wilt who may not look great by fp4's measure, but also sees a +2 point defensive spike for his playoff teams. Those 2 points outweigh the 1-point offensive drop-off so I wonder if Wilt should be considered a playoff-riser regardless.

wilt is also really bad at actual vs expected titles at -1.04. not only is that 95th out of 103, but he's the only person with 2 titles who is even negative at all, so being 95th out of 103 is even more damning with 2 titles.

so a reductive approach places him as a faller, and a more inclusive one places him as a riser?

granted there is a trade-off there, especially in a league where you are unlikely to avoid the best team(and player).

[Throwawaytheone]

.

[Throwawaytheone]

.

[OhayoKD]

No, not really. A postseason performance is just how you played within the confines of a playoff setting. Thats it. Did they play good? Did they play bad? That stuff.

Rising and falling is comparing their PS performance to their RS performance and evaluating off of that (or incorporating that into your overall PS evaluation). It's just dumb as hell, because a fringe all star could rise to being a 2nd team All NBA player in the PS and a strong MVP player could fall to simply being a weak MVP player and this "Rise/fall" talk would have you thinking the former is a legendary clutch killer and the latter is a choker. If the first dude set a higher bar for himself, it wouldn't be a convo, but people literally reward him for playing worse in the RS.

And, "How well you performed" can be seen as "rs performance+the difference between rs and playoff performance". That is an extra step, but it has the benefit of letting us use the larger regular-season sample to inform playoff assessments.

Paticularly valuable when you are trying to get to tie players to their ability to effect wins.

Players should not be rewarded for setting a lower bar no(cough kawhi cough shaq). Players should also not be penalized for setting a higher regular season bar(cough kareem cough steph). But if a player sets a similar bar and rises more, then it's fair to favor player 2(this is 15-17 lebron's case vs 15-17 steph more or less).

Of course it's preferable to cut out the middle-man and just assess the playoff goodness directly but that's not so convenient when you're trying to spin Kareem's 74 run as a negative or rank Duncan lower than Hakeem and Shaq because he won too much in the regular season...

Informing playoff assessments is vague. If we're talking using a larger sample to indicate whether the playoff performance was just a flukey outlier or it was real, then absolutely, that's valid.

[/quote][/quote]
I'm more saying when you are trying to place a performance as "best ever" or "great" or "good", seeing what effect a player has in the regular-season and then curving based on playoff stuff(can be box, can be team-wide elevation or drop, ect) gets you a potentially better assessment of a playoff run then just going off what you think is better(especially in a comparison between completely different types of players).

Using it to inform playoff assessments like "LMAO LEBRON IS A CHOKER!!!!" because he had a PPG drop and is now a "Faller" is dumb because that playoff run is still in the 99th percentile ever.

Sure. But if a similarly valuable player sees their numbers jump, it would be fair to favor them,

I'm a little confused for your 15-17 Curry and Lebron one. Are you saying you favour Curry more here or am I misinterpreting? Because Curry is the 2nd player you listed (literally, like Lebron came first, Curry came second.)

reverse. Similarly impactful rs players, Lebron elevates while Curry shrinks or mantains.

[Throwawaytheone]

.

[OhayoKD]

I'm more saying when you are trying to place a performance as "best ever" or "great" or "good", seeing what effect a player has in the regular-season and then curving based on playoff stuff(can be box, can be team-wide elevation or drop, ect) gets you a potentially better assessment of a playoff run then just going off what you think is better(especially in a comparison between completely different types of players).

Using it to inform playoff assessments like "LMAO LEBRON IS A CHOKER!!!!" because he had a PPG drop and is now a "Faller" is dumb because that playoff run is still in the 99th percentile ever.

Sure. But if a similarly valuable player sees their numbers jump, it would be fair to favor them,

I'm a little confused for your 15-17 Curry and Lebron one. Are you saying you favour Curry more here or am I misinterpreting? Because Curry is the 2nd player you listed (literally, like Lebron came first, Curry came second.)

reverse. Similarly impactful rs players, Lebron elevates while Curry shrinks or mantains.

If you're trying to assess the greatest individual PS run ever, I don't see why you'd need to use the regular season to change your assessment in this case.

For the Curry/Lebron example, Curry has been significantly better in the RS across 15-17 due to Lebron's coasting so it's not really a fair example and falls prey to what I've been talking about before.

[/quote]
Nah, I think it was a lot closer than you are giving it credit for.
https://forums.realgm.com/boards/viewtopic.php?p=106319069#p106319069
Whether Lebron "coasts" ultimately doesn't matter if he's still generating similar value via his own advantages. Lebron also has replication across contexts Steph does not so "situation" is not a particularly strong excuse.

Notable here is that Steph does alot better relative to Lebron with conventional box-aggregation than he does in "impact on winning". Steph's 2016 being the goat regular season was always pretty flimsy, just like Lebron "sandbagging" before the playoffs