CORP formula's

[trelos6]

For those who have a CORP system for all time rankings, how do you weigh seasons?

Let's call an All-Star level 1 point, how much more do you each value the following types of seasons?

All time
MVP
Weak MVP
All NBA
All Defense

Currently, my system is working off semi-arbitrary numbers which were generated using a recursive formula. I'm working on refining it based of some more information.

[eminence]

I don't give bonuses for All-Defense, just try to more fairly grade defenders than traditional awards.

Anywho.
Low end bench - 1 pt
Good bench/Lower Starters - 2
Good Starter - 3
Allstar - 5 (more players than there are actual allstars, top 40ish guys modernly)
AllNBA - 8
1st Team/Weak MVP - 12
MVP - 17

As stated in the other thread I'm not currently giving out All-Time grades (they exist, but I felt too biased when assigning them), but it would be a 23 on that scale.

It's a lazy caterer's sequence +1. At one point I was using fibonacci, but felt it took off at higher levels. Same points at bench - through AllNBA. 13 for 1st, 21 for MVP, 34 for Alltime. Considering going back to it with the removal of Alltime level.

Obviously for the most detail one would have a continuous scale, but the discrete ones function pretty well for estimates imo.

[trex_8063]

For those who have a CORP system for all time rankings, how do you weigh seasons?

Let's call an All-Star level 1 point, how much more do you each value the following types of seasons?

All time
MVP
Weak MVP
All NBA
All Defense

Currently, my system is working off semi-arbitrary numbers which were generated using a recursive formula. I'm working on refining it based of some more information.

Firstly, I don't think you can really do it as "1 point", "2 points", etc.
CORP stands for something: "Championship Odds above Replacement-Level Player".........it's an estimation on how much a certain calibre of player [in a vacuum, or across a comprehensive spectrum of settings] RAISES a team's odds of winning a title, if you replace a true 'replacement level' roster spot with the player in question.

Thus, the "points" awarded should be in terms of what % improvement to the team's chances of winning a title. EDIT: though I suppose using 1 pt as a baseline is okay if you're just trying to gauge the difference between levels.


Secondly, I don't start at "All-Star" as the baseline of awarding a score. Again, it's there in the acronym: we're talking about any improvement above a 'replacement level player'. An "All-Star" is not the lowest baseline that is a significant improvement over a 'replacement level' guy.

Further, starting there would completely omit guys like Danny Green or Bruce Bowen or MPJ, etc.......guys who were CLEARLY very valuable (essential, even) to their teams' success.
Even lesser players can be of significant value (for instance, solid bench players makes it possible to rest your starters with some degree of comfort, knowing the whole ship isn't going to sink in the 3-4 minutes you rest some guys). That could be the difference between winning a title and not winning the title in rare circumstances (I think of some of those late 90s Jazz teams, where their bench was SO weak, they just fell off a cliff if at least 1-2 of Malone/Stockton/Hornacek wasn't on the court.....like absurdly bad -30 net ratings and such; I honestly believe if they'd had a GOOD bench, they'd have at least one title [and likely more]).


For my own CORP (haven't played with them in awhile), I started at roughly "league average" players (basically anyone who is ABOVE 'replacement level'); to my philosophy, 'replacement level' players are the only ones who are truly, well.......replaceable. And I had the "Average Player" adding 1.5% (0.015) CORP value.

My next category of player was "Sub-All Star" [or "Decent Starter-Level", if you will]: somebody who is clearly above league average, though also not up to the level of an All-Star. I awarded 4% (0.04) CORP value for these.

"All-Stars" were next, and earned 6.5% (0.065).
Then All-NBA [2nd/(3rd)-ish Team level], earning 10% (0.1).
Next was "Weak MVP" [this might be considered roughly "lower All-NBA 1st Team level"], earning 16% (0.16).
Then "MVP" (this is like top 1-2 player in the league, but not an "all-timer"), earning 21% (0.21).
Then "All-Time" season (this is generally clear best player in the league, and likely would be one of the top-5 seasons of an entire DECADE); it earns 29% (.29).

And last is a "GOAT-tier" season; these were rare (as of '23, I had awarded only 10 such seasons [total], which were achieved by only six guys at that point: LeBron, Jordan, Kareem, Wilt, Shaq, and Mikan). These were worth 35% (0.35).


Adding each season up yielded a player's base score. Though I next applied an longevity calibration (to avoid the "Sub All-Star for 50 years would be the GOAT" dilemma) to yield my **final score.

**However, I also had an additional version which takes that^^ score and applies an era-adjustment (based on some subjective(ish) assessments of the strength of each season a player played in).


So anyway, that's how I did it.
Oh, and I allowed half-measures between categories (for seasons that felt right in the middle of one category or another). A player might, for example, get credit for 0.5 "All-Star" seasons and 0.5 "Sub-All Star" seasons for a particular year.

[trelos6]

I start at all star level because I don’t have the time to rank the Danny Green’s of the world. I recognise and appreciate their value, but it’s a lot of work.

I agree with your weights for each category. They’re around about where mine sit, give or take 1-2%. I like the idea of the 50/50 weighting split for seasons on the borderline. I’ll look to implement that.

Thanks for the feedback T-Rex.