Why are peaks so much harder to rank than careers?

[ardee]

It's weird to me. I'm reasonably confident about my top 25 career wise, then it gets a bit murkier but if I was to really think about it I think I could make decisions over who I think had more career value, player by player, without an issue. Also have ranges and tiers I'm even more confident in.

With peaks I'm already lost after my top 11. Feels like there's 25 guys who might have a case for the 12-20 spots.

Anyone else feel this way?

[Doctor MJ]

I think it's one of the more telling phenomenon frankly. To me the essence of it is this:

It's easier for us to rank careers rather than peaks, because ranking careers requires less understanding of basketball.

More generally: The longer the duration we look at, the less our rankings are depending on the combination of eyeball read & n-dimensional player shape, and the more it is based on objective data and the on-record subjective opinions of others.

Of course, this isn't to say that no one finds it easier to judge players in the moment than career. imho, this is largely what players do when they evaluate each other, and what they have to say about each other is typically very insightful...but when you listen to enough player opinions, it becomes clear that they typically don't have a great sense of how to translate their scouting of other players into an accurate & precise measure of a player's impact. There doesn't seem to be any substitute for analytical work to have a real sense of impact and cumulative achievement.

[jalengreen]

i think there's more variance career to career than there is season to season.

compare the careers of certain players and there's a good chance one player clearly had a shorter prime / clearly had less longevity.

but when you're comparing the absolute *best* of certain players, the factors that make it easier to rank careers are eliminated.

[70sFan]

There are way more players that peaked very high than the ones with excellent all-around careers. On top of that, it's tougher to differentiate the noise from the signal in smaller samples.

Personally, the more I think about it, the more I prefer looking at 2-3 years peaks instead of individual seasons. It's just much more reliable to get a bigger sample.

[No-more-rings]

I disagree, I think careers are much harder to rank. You have to evaluate every season and not just one or two.

[Texas Chuck]

There are way more players that peaked very high than the ones with excellent all-around careers. On top of that, it's tougher to differentiate the noise from the signal in smaller samples.

Personally, the more I think about it, the more I prefer looking at 2-3 years peaks instead of individual seasons. It's just much more reliable to get a bigger sample.

Yep samples sizes have to matter. I ignore one year peaks completely with the exception of a guy like Walton where injury robbed him. TMac in 2003? He never had another season close to this, so this isn't his real level of play. He was on the right side of variance is all.


The other thing is different people are measuring different things. I'm not one trying to figure out who the "best" player was. I'm trying to figure out the value of their career. For those interested in some esoteric goodness, they are going to be heavily invested in peak years and go into granular detail I'm just not going to.

Which is handy since as Doc points out, it allows me to have a lessor understanding of basketball than some of you geniuses.

[Owly]

Why are peaks so much harder to rank than careers?

It's weird to me. I'm reasonably confident about my top 25 career wise, then it gets a bit murkier but if I was to really think about it I think I could make decisions over who I think had more career value, player by player, without an issue. Also have ranges and tiers I'm even more confident in.

With peaks I'm already lost after my top 11. Feels like there's 25 guys who might have a case for the 12-20 spots.

Anyone else feel this way?

First glance responses some are related and this is going to be messy, rather than coherent

1) Small samples. This could perhaps have many more offshoots but the first ones to mind are:
1a) Small sample luck: Small sample means greater chance of luck being influential. Greater chance of luck being influential means differences in how we interpret or account for it factor in more. Was T-Mac's shooting lucky? Does it matter (are we rating our imagined interpretation of a holistic player or that season in bottle)? (Related: how much do we allow other years to influence?).
1b) Small sample greater outliers: With more extreme years that are often smoothed off a bit over a career methodological differences give less varied results. In a single year different tools and measures will vary more.

2) Laugh test/courage. Most will be within a vague mostly not absolutely terrible consensus about the rough range of great careers. Now if you are going against that on a career list ... yeah that would be the hardest. But I think that list will shape peak lists. If it's not a notorious absolute apex or key narrative peak ... it just might require bravery to say I trust this measure and it says … say … Isaiah Thomas had a top X level peak.
2a) Worse “anchors”: As implied above I think career ratings might be a somewhat restrictive anchor. If you’ve focussed in on peaks you’re unlikely to peak at the exact same time as multiple others. West vs Robertson career is an easy comparison (same time) but peaks is harder.

3) Greater complexity. Versus careers you first have to decide which year to pick. It’s just one more step.

4) Smaller gaps. One could certainly argue comparing more or less like with like (season with season - though length may change, conventions, talent pool etc) makes things simpler. But here longevity can’t break any ties. There’s limitations to how much … say, CORP? … one can add in one season.

5) Playoffs: People view them differently. Single season means much more variance in what sample we have. Many will have none.

6) Impact (stats): These are where evaluations have been heading. These work better with larger samples.

[eminence]

basketball goodness · time played = career value (for most people)

We've introduced a 2nd variable (time played) that is (somewhat) independent of the first. Just more knowledge to work with.

[70sFan]

There are way more players that peaked very high than the ones with excellent all-around careers. On top of that, it's tougher to differentiate the noise from the signal in smaller samples.

Personally, the more I think about it, the more I prefer looking at 2-3 years peaks instead of individual seasons. It's just much more reliable to get a bigger sample.

Yep samples sizes have to matter. I ignore one year peaks completely with the exception of a guy like Walton where injury robbed him. TMac in 2003? He never had another season close to this, so this isn't his real level of play. He was on the right side of variance is all.

I do the same thing and that's why I don't even consider Tracy inside my top 25 peaks.

[Texas Chuck]

basketball goodness · time played = career value (for most people)

We've introduced a 2nd variable (time played) that is (somewhat) independent of the first. Just more knowledge to work with.

I don't think that formula works for me because how do you determine basketball goodness? Take Tim Duncan who was some form of positive his entire career. Do we take 03 and multiple it by 20 years? That can't be right.

Now I realize for some they only care about what they have somewhat arbitrarily defined as prime years, and they simply cut off the rest completely. So for them I guess your formula works?

I'd be curious actually if most people cut off non-prime years or feel like me, that if a season benefited your real-life team then we need to account for it when assessing your career. And again that goes back to we all measure different things. I'm interested in the actual career rather than something more esoteric. Others are definitely after finding out who the best player was independent of who had the best career. So for them 70's Kareem is basically all that matters and Showtime career outside of the first handful of years may as well have not happened.

[eminence]

basketball goodness · time played = career value (for most people)

We've introduced a 2nd variable (time played) that is (somewhat) independent of the first. Just more knowledge to work with.

I don't think that formula works for me because how do you determine basketball goodness? Take Tim Duncan who was some form of positive his entire career. Do we take 03 and multiple it by 20 years? That can't be right.

Now I realize for some they only care about what they have somewhat arbitrarily defined as prime years, and they simply cut off the rest completely. So for them I guess your formula works?

I'd be curious actually if most people cut off non-prime years or feel like me, that if a season benefited your real-life team then we need to account for it when assessing your career. And again that goes back to we all measure different things. I'm interested in the actual career rather than something more esoteric. Others are definitely after finding out who the best player was independent of who had the best career. So for them 70's Kareem is basically all that matters and Showtime career outside of the first handful of years may as well have not happened.

I should've more clearly indicated that my variables were vectors

The longform would look something like this:

Season1 Goodness · Season1 Time + Season2 Goodness · Season2 Time + ... + FinalSeason Goodness · FinalSeason Time

Or whatever timeframe you'd like to break it down to instead of seasons: games, series, broad strokes career sections (Duncan w/Robinson, Duncan without Robinson, post-prime Duncan, etc).

[DQuinn1575]

I think your careers are easier because of the lifetime factor.

If you compare Stockton to Price, KJ, Gus Williams you could argue who has better peak, but the amount of seasons gives it to Stockton.
Similar to Havlicek vs Frazier, Cowens, Reed
In both cases I think you would get a variety of answers as to who has best peak of the group, but the career answer would be fairly lobsided to 1 guy.

[Blame Rasho]

It is in my estimation more subjective to rank a season than a career.

A good example would be T-mac. I think he had one single year at his apex that I can say with absolutely certainty he was better than Kobe, but someone can see it differently and say that Kobe’s season was better than his and got held back because he had ummm Shaq. Am I right or wrong? Does it matter in the end? Most people don’t talk about single seasons but careers.

Another example… who was the best swing player after MJ retired?

As per all nba voting it was…..

Latrell Sprewell.

I bet most people just think of him of the guy in the Knicks with the overpaid Allen Houston and one of the two guys that helped KG get to the WCF. Oh and he needed a family to be fed

No one really talks of him as the all nba 1st teamer.

[HeartBreakKid]

They're not but it requires people to put names that aren't very sexy or prestigious in front of guys who won many nice accolades. It makes people uncomfortable.