RealGM Greatest Peaks Project (2022)
Posted: Mon Jun 6, 2022 5:01 am
It has been 3 years since the last peaks project so it is time for another one. We will be ranking single season peaks including both regular season and playoffs and each player will be represented on the list with only one season. The scope of this project will be every season from 1947 to 2022 in the NBA/ABA/BAA. For reference, here are the results of the peak project from previous years:
The purpose of this thread is to first gauge if there is sufficient interest for this project to go ahead. Posters who participated in the peaks projects before have been automatically included in the quote list at the end of this post but anyone not on the list is also welcome to sign up and join in the project. If we don't get 10~20 people, the project will be postponed to another year. Otherwise, my current plan is for the project to start a few days after the Finals and to end before the start of the 2022-23 season in mid October. The voting window will be 72 hours for the first 20 rounds and it will be reduced to 48 hours from round 21 onwards as the amount of discussion usually declines at that point.
In terms of the voting system, past peaks project have always had problems with some players potentially having multiple peaks. Ideally, there should be no advantage or disadvantage to players who had many peak season candidates compared to others with a clear unanimous peak. The best solution in my opinion is to vote for individual player seasons and choose the winner through head to head comparisons via the Condorcet method. To show how this works, you will write your 3 highest player peaks and at least one line of reasoning for each of them. But in addition, you can also list other peak season candidates from those three players. As an example, this is what a vote could potentially look like:
1. 1991 Jordan: Explanation
2. (1990 Jordan)
3. 2013 LeBron: Explanation
4. (2012 LeBron)
5. (1992 Jordan)
6. (2009 LeBron)
7. 1967 Wilt: Explanation
8. (1964 Wilt)
The reason for also including seasons that you don't consider to be peaks in your vote is so that players with multiple viable peak seasons can be evaluated more accurately when the votes are added up at the end of the round. If a player season is not listed in your vote, I would treat it as you ranking it behind every other season present in your vote. At the end of each round, every season is compared pairwise and the winner is the season that beats every other season head to head. For example, if these are the votes (in descending order of preference) in a round for players A, B, C and their seasons A1, A2, A3, B1, B2, C1:
A1, A2, A3
A1, B1, A2
A1, B1, A2
B1, A1, A2
B1, B2, A1
B2, A1, B1
B2, B1, A1
C1, B1, B2
C1, B2, B1
The winner is the season B1 because a majority of the voters preferred it over each of A1 (5>4), A2 (8>1), B2 (5>3), and C1 (6>2).
I chose the Condorcet method instead of the voting methods used in previous projects because I think it is the most accurate and least susceptible to manipulation. A simple plurality in seasons (2012 project) disadvantages players with multiple peaks while a simple plurality with combined peak seasons (2015 project) doesn't account for how differently individual seasons by the same player can be ranked. The Borda count method (2019 project) gives an artificial advantage to players with multiple peak level seasons and can easily be gamed to inflate the distant between your preferred candidate and their closest rival. The Condorcet method requires more work when adding up the votes but it should give the best results.
In the event of a tie at the end of the voting period, there will be an additional 24 hour runoff between the top seasons in contention. If the tie is between multiple seasons from the same player, the runoff will run concurrently with the next round because the winning player has already been determined. If there is still a tie at the end of the 24 hour runoff, my idea is to discard the last vote by timestamp until there is no tie. This might seem too basic and arbitrary of a tiebreak rule but
1. It keeps the project moving. Later rounds with already low participation often die when we wait indefinitely for tiebreak votes.
2. It guarantees a winner without adding too many complicated tiebreaking mechanisms (see top 100 project).
3. It incentivizes people to get their votes in as early as possible so their votes aren't the last one and discarded.
4. If the vote was really close, the runner up will likely win the next round anyway so I don't think a one spot difference is that big of a deal in the overall scheme of things.
That's everything I have for the rules. If there is anything people don't like or want to add, I am always open to new ideas.
In any case, I am looking forward to start round #1 sometime between June 16-22.
Note: If you find yourself in the list of names below and don't wish to receive further notifications about this project, simply withdraw and your name will be removed.
Spoiler:
Spoiler:
Spoiler:
In terms of the voting system, past peaks project have always had problems with some players potentially having multiple peaks. Ideally, there should be no advantage or disadvantage to players who had many peak season candidates compared to others with a clear unanimous peak. The best solution in my opinion is to vote for individual player seasons and choose the winner through head to head comparisons via the Condorcet method. To show how this works, you will write your 3 highest player peaks and at least one line of reasoning for each of them. But in addition, you can also list other peak season candidates from those three players. As an example, this is what a vote could potentially look like:
1. 1991 Jordan: Explanation
2. (1990 Jordan)
3. 2013 LeBron: Explanation
4. (2012 LeBron)
5. (1992 Jordan)
6. (2009 LeBron)
7. 1967 Wilt: Explanation
8. (1964 Wilt)
The reason for also including seasons that you don't consider to be peaks in your vote is so that players with multiple viable peak seasons can be evaluated more accurately when the votes are added up at the end of the round. If a player season is not listed in your vote, I would treat it as you ranking it behind every other season present in your vote. At the end of each round, every season is compared pairwise and the winner is the season that beats every other season head to head. For example, if these are the votes (in descending order of preference) in a round for players A, B, C and their seasons A1, A2, A3, B1, B2, C1:
A1, A2, A3
A1, B1, A2
A1, B1, A2
B1, A1, A2
B1, B2, A1
B2, A1, B1
B2, B1, A1
C1, B1, B2
C1, B2, B1
The winner is the season B1 because a majority of the voters preferred it over each of A1 (5>4), A2 (8>1), B2 (5>3), and C1 (6>2).
I chose the Condorcet method instead of the voting methods used in previous projects because I think it is the most accurate and least susceptible to manipulation. A simple plurality in seasons (2012 project) disadvantages players with multiple peaks while a simple plurality with combined peak seasons (2015 project) doesn't account for how differently individual seasons by the same player can be ranked. The Borda count method (2019 project) gives an artificial advantage to players with multiple peak level seasons and can easily be gamed to inflate the distant between your preferred candidate and their closest rival. The Condorcet method requires more work when adding up the votes but it should give the best results.
In the event of a tie at the end of the voting period, there will be an additional 24 hour runoff between the top seasons in contention. If the tie is between multiple seasons from the same player, the runoff will run concurrently with the next round because the winning player has already been determined. If there is still a tie at the end of the 24 hour runoff, my idea is to discard the last vote by timestamp until there is no tie. This might seem too basic and arbitrary of a tiebreak rule but
1. It keeps the project moving. Later rounds with already low participation often die when we wait indefinitely for tiebreak votes.
2. It guarantees a winner without adding too many complicated tiebreaking mechanisms (see top 100 project).
3. It incentivizes people to get their votes in as early as possible so their votes aren't the last one and discarded.
4. If the vote was really close, the runner up will likely win the next round anyway so I don't think a one spot difference is that big of a deal in the overall scheme of things.
That's everything I have for the rules. If there is anything people don't like or want to add, I am always open to new ideas.
In any case, I am looking forward to start round #1 sometime between June 16-22.
Note: If you find yourself in the list of names below and don't wish to receive further notifications about this project, simply withdraw and your name will be removed.
_Game7_ wrote:.
70sFan wrote:.
Amares wrote:.
Ambrose wrote:.
ardee wrote:.
BallerHogger wrote:.
Bel wrote:.
Blackmill wrote:.
cecilthesheep wrote:.
ceiling raiser wrote:.
Clyde Frazier wrote:.
Colbinii wrote:.
DatAsh wrote:.
Doctor MJ wrote:.
dontcalltimeout wrote:.
Dr Positivity wrote:.
Dr Spaceman wrote:.
drza wrote:.
E-Balla wrote:.
eminence wrote:.
freethedevil wrote:.
GoldenFrieza21 wrote:.
Gregoire wrote:.
HBK_Kliq_33 wrote:.
HHera187 wrote:.
Homer38 wrote:.
Jaivl wrote:.
Joao Saraiva wrote:.
JoeMalburg wrote:.
Joey Wheeler wrote:.
JordansBulls wrote:.
LA Bird wrote:.
liamliam1234 wrote:.
Lou Fan wrote:.
Mavericksfan wrote:.
michievous wrote:.
Moonbeam wrote:.
Mutnt wrote:.
MyUniBroDavis wrote:.
Narigo wrote:.
No-more-rings wrote:.
Owly wrote:.
pandrade83 wrote:.
PaulieWal wrote:.
PCProductions wrote:.
penbeast0 wrote:.
Point-Forward wrote:.
Quotatious wrote:.
RebelWithoutACause wrote:.
RSCD_3 wrote:.
Samurai wrote:.
SideshowBob wrote:.
SKF_85 wrote:.
Sublime187 wrote:.
theonlyclutch wrote:.
The-Power wrote:.
thizznation wrote:.
Timmyyy wrote:.
trex_8063 wrote:.
Vladimir777 wrote:.
yoyoboy wrote:.
