Professional and College Basketball Forums banner
1 - 20 of 37 Posts

·
Registered
Joined
·
2,725 Posts
Discussion Starter · #1 · (Edited)
Put Up or Shut Up

Long thread. Tough Concept. Read carefully. But a new idea that has never been posted. This is where people can show some new details and creativity for a change. I challenge you to show something instead of debating who is underrated or who is better. This is the HEART and SOUL of almost EVERY debate in the forum.

This is a hot topic. I hate comparing players with their team's accomplishments. It isnt technically correct. Here's an example:

Analyzing a player's SPECIFIC contributions (what that player does, separate from the rest of the team) you can see what players are the best separate from the TEAM'S OVERALL performance. There are so many things that come into play during a win. You are taking the easy way out if you look at the team's record. Put it into numbers. Say a team needs to reach 100 'win points' to win a game. These 'win points' are just a hypothetical remember. Define them as everything an INDIVIDUAL does to contribute to winning.

Player A contributes 30 win points to the win (best individual on the team)

BUT the rest of the team, coaching staff, team chemistry of players other than Player A only contributes 60 win points to a win

30 + 60 = 90, thus, the team loses on average = NOT GREAT TEAM RECORD


Player B contributes 20 win points to the win (best individual on the team)

The rest of his team, coaching staff, team chemistry of players other than Player B contribute 85 win points to the win

20 + 85 = 105, thus, the team wins on average = BETTER TEAM RECORD

Player A is the better player (30 win points > 20 win points). So, in theory, but not likely, the best player of all time could never make the playoffs (or win a game, but i didnt think you guys could take that extreme). But in your mind, you say, "i just cant conceive the best player not making it to the playoffs." Well, that is because, usually, the best players get help, and they do contribute a lot. But in today's media, W-L is one of the TOP things considered, and people throw it out as a INDIVIDUAL stat all of the time. I'm not saying that the human basketball follower cant get a good idea of these 'win points' by watching closely. But when people compare, they do not use what they specifically watched, they use W-L. It's like W-L is the is all that matters now, and it is FOR A TEAM. All over TV, it's 'but his record is this' and 'he's on a team with 7 more wins'. They dont even consider that people like 'Player A' are giving more. Plus, many players compared by W-L are so close in wins that my example could really come into effect. So how do we measure 'win points'?

I would like your ideas of ANY individual stat we could take that would 'paint' a picture of what an individual player contributes. Think outside the box, picture what players do in games that isnt documeted but could be. Now, weighting these stats to determine what matters most to a win is the hard part that Im stil thinking about.

Please dont give me the "stats dont show what a player does" thing. The idea is to try to get as close as possible. I realize we can never fully define a player by stats.

Any ideas???
 

·
Registered
Joined
·
8,045 Posts
assists resulting in layup
assists resulting in shot in paint
assists resulting in shot + foul
passes that are then passed for an assist
opponent FG%
opponent 3PT%
opponent steals
charges taken
steals caused (resulting in a steal for a teammate)
deflections
unforced turnovers
fast break turnovers
halfcourt PPG
transition PPG
PPG scored in paint
PPG scored off assists from teammates
PPG scored resulting in no assist
 

·
Registered
Joined
·
2,725 Posts
Discussion Starter · #3 ·
no thinkers? that's what I thought, too bad.

secondary assists - assists for setting up a player that makes an easy pass to another for a score

charges taken - shows good D

close assists - assists to player within a certain distance from hoop

contested rebounds vs. uncontested rebounds - some rebounds are gimmies and some are one teammate over another with the other team not close

passing lane steals vs. positioning steals

double teams drawn - tough measurement, but can be done

forced turnovers

other team's points of turnovers from a player


still thinking
 

·
Registered
Joined
·
8,925 Posts
I think we have enough stats to do a good enough job of showing a players impact. Individual stats arent that important anyways, but a couple stats could be

blocks from on the ball defense
blocks from help defense
steals from on the ball defense
steals from help defense

some kind of stat against you if you miss a steal after gambling to try and get it
 

·
Registered
Joined
·
8,045 Posts
Good stuff tpb2 and John.

close assists - assists to player within a certain distance from hoop
I was going to list this, but felt it might be a bit too subjective. Then again, some stats are (ie, some assists).

contested rebounds vs. uncontested rebounds - some rebounds are gimmies and some are one teammate over another with the other team not close
Nice.

passing lane steals vs. positioning steals
Very nice.

double teams drawn - tough measurement, but can be done

forced turnovers

other team's points of turnovers from a player
Good stuff as well.
 

·
Registered
Joined
·
2,725 Posts
Discussion Starter · #8 ·
Originally posted by <b>John The Cool Kid</b>!
Individual stats arent that important anyways
when comparing INDIVIDUALS, they are

Like I said, if you get an individual stat that comes really close to measuring his impact on WINS, you can argue individual stats without ever caring about the team they play for.

You must clear your mind of the team. a plyer should only be measured for the things in his control - thus, the stat we're going for
 

·
Registered
Joined
·
2,004 Posts
It seems to me some statistics should deal more with the qualitative aspect. Most stats now only look at the quantitative side, like how many points are scored, how many assists, steals, blocks, etc. They do not rate the quality of each of these categories. It would be hard to set up a hierarchy of quality though.

Some other possible statistics:

1)Critical turnovers vs. Regular turnovers
2)Amount of times pass set up open man but he missed
3)Number of times a player was able to get open by moving without the ball
4)Number of blockouts for rebounds per game
5)Quality of screens set
6)Assists from double teams/triple teams
7)Number of shots made with the shot clock winding down
8)Open shots made/missed vs. contested shots made/missed

I'm sure I could think of more, but it's late.
 

·
Registered
Joined
·
2,725 Posts
Discussion Starter · #10 ·
It would be hard to set up a hierarchy of quality though.
This is the key.

I was thinking you could study a large sample of games over every year. You could see how all of our stats matter in each year, and try to detect a trend of how they are changing by comparing years. If there is a STRONG trend, then you could have the value of each stat change by year. This seems a bit too complicated though. I dont really like extrapolation unless the trend was EXTREMELY strong. I doubt it would be.

or


you could standardize each year and just assume the values of these stats stay the same. Then you could compare eras by percentiles. I have no idea why percentiles arent used to compare eras in the stats we have now.
 

·
Top Of The Pops
Joined
·
27,473 Posts
These are all subjective stats, up to the scorer.

Pass forced out of the post - if a main scoring threat is being defended in the low post and the ball is entered into that post player, but the defender forces a pass back out, it's a sign that he played good defense

Travels forced - a player's defense caused the travel

Loose balls saved - diving into the crowd to save a ball from going out of bounds, diving to get to a ball that the opposition seemed to have, etc

Ball denial - preventing a pass from being attempted when a player is clearly trying to make it

Unsubjective:

Post position forced - The average distance from the hoop the opponent a player is defending receives the ball, in the post. Obviously, you want to prevent deep post position, as the defender.

I'll stop thinking about these for tonight.
 

·
Nobody
Joined
·
6,373 Posts
Ultimately, I think people use stats to justify their subjective impressions... I don't think looking at the stats really changes anyone's mind about something they decided while watching a game earlier. Though the stats can reinforce what you already think if they agree with you. Stats of any kind are not going to overrule the spectator experience... I would be very skeptical of the opinions of anybody for whom they did.
 

·
Registered
Joined
·
2,725 Posts
Discussion Starter · #13 ·
Stats of any kind are not going to overrule the spectator experience... I would be very skeptical of the opinions of anybody for whom they did.
Im not sure if natual human bias is greater than statistical bias or not. I do know that there is a human bias to believe this isnt the case.


But people dont argue with the spectator experience or else the absurd frequency of W-L arguments with individuals would not occur. They retreat to the COMFORT ZONE of easily comparing team records because the human spectator cant (or DOESNT REALLY TRY TO) distinguish between players that are somewhat similar in true impact(or 'win points' in my example). Scouts may be able to do this. I know I could if I could watch a ton of games live.
 

·
Registered
Joined
·
2,004 Posts
Originally posted by <b>tpb2</b>!

you could standardize each year and just assume the values of these stats stay the same. Then you could compare eras by percentiles. I have no idea why percentiles arent used to compare eras in the stats we have now.
So kind of like they use the consumer price index to estimate changes in the rates in inflation? There is a base year and that year the value of a statistic is 1. If there is an increase let's say of the impact of that statistic on the outcome of games during that season by let's 4% then the statistic in the present year would have a value of 1.04.

I'm not sure if this is what you meant but the problem with this is that it would be hard to assign qualitative increases in impact of a certain statistic. Also, deciding what the base year should be, would seem random. This once again would be subjective. Although I am probably misinterpreting what you meant.
 

·
Registered
Joined
·
2,725 Posts
Discussion Starter · #16 ·
Originally posted by <b>RapsFan</b>!


So kind of like they use the consumer price index to estimate changes in the rates in inflation? There is a base year and that year the value of a statistic is 1. If there is an increase let's say of the impact of that statistic on the outcome of games during that season by let's 4% then the statistic in the present year would have a value of 1.04.

I'm not sure if this is what you meant but the problem with this is that it would be hard to assign qualitative increases in impact of a certain statistic. Also, deciding what the base year should be, would seem random. This once again would be subjective. Although I am probably misinterpreting what you meant.
go back to my post

For option #1:
say for each year, we gave a multiplier (in order to weight) to some stat based on how important it is for a win- this would require advanced research to get this number for each year of the NBA.

1971 - .56
1974 - .60
1980 - .66
1984 - . 667
1998 - .668
2003 - .669
2004 - year my stat is started; extrapolate multiplier

Seeing this trend, maybe we can just calculate what the weight should be in 2004. The trend is used to take into account that the game is CHANGING. So we could do research in the past in order to get the value for the future by following the curve (since we cant update the value of a stat in the season)

For option #2:

Instead of saying Jordan shot 49% and Kobe is only shooting 42%. Say Jordan was in the 49th percentile of FG% in 1988 and Kobe was in the 50th percentile of FG% in 2003. Statistically, Kobe is shooting better. Ask any statistician and they'd agree (i mean they put it on the AP Stats test but it wasnt the NBA). So do percentiles for all stats. You could do the research from the previous example and if you determine the value of a stat doesnt change significantly by year, you could just make every multiplier constant. Thus, you could technically have one number saying who is better. Without the weight, even today with our simple stats, you should compare eras with percentiles!
 

·
Registered
Joined
·
2,725 Posts
Discussion Starter · #18 ·
the win share is the same concept.

I think the difference is the idea we are discussing, which "strength of stat"

There has to be a way to deterine the strength of a stat. What about..........

Technically, with a lot of work, you could measure how important the stat is by measuring over alomst every game the relationship between the numerical value of the stat and win percentage. I think the variability would be really high, and the data would be pretty reliable if you did this for every game. Doesnt that seem logical? The only hump then, would be the work of watching all the games and recording. Then, like I said, you would assume a constant weight or extropolate the weight.

Every good stat vs. win %

anything im missing?
 

·
Registered
Joined
·
8,925 Posts
Individual stats dont matter because basketball is a team game. Also, like Nevus said, no matter how many stats there are it wont change peoples mind from what they see on the court.
 

·
Top Of The Pops
Joined
·
27,473 Posts
Originally posted by <b>John The Cool Kid</b>!
Individual stats dont matter because basketball is a team game.
Individual stats matter in evaluation. If you're a GM, do you say, "I think I'll pick up Malik Rose instead of Kevin Garnett, because he has more winning experience?"

Similarly, fans want to discuss which players are the best, and better, smarter tools for individual analysis lead to more informed opinions.
 
1 - 20 of 37 Posts
Top