Re-Evaluating FT Rate as the Fourth Factor (Chilltown, 2011)
Posted: Fri Apr 15, 2011 5:32 am
page 1 of 1
Author Message
Chilltown
Joined: 16 Apr 2010
Posts: 15
Location: Boston
PostPosted: Mon Feb 21, 2011 12:48 pm Post subject: Re-Evaluating FT Rate as the Fourth Factor Reply with quote
Recently, I've become very interested in how we measure production from the Free Throw line. I've read Dean Oliver's justifications for why FT Rate is the Fourth Factor (because getting to the line is more important in the long run than simply making shots), but that thought always struck me as strange.
I decided to test FT Rate against Free Throws Made per 100 Possessions, a stat which more explicitly measures FT line production in points, and found that by the criteria I could think of, FTM/Poss is a better Fourth Factor.
The full analysis/conclusions are here. I have only done this for offense, because the effects of "free throw line defense" seem more subtle and worthy of a more in depth study.
I would welcome any discussion on this. Hopefully this post gets a good discussion of the Fourth Factor going because I really do believe that FT Rate could be improved.
Back to top
View user's profile Send private message Visit poster's website
Crow
Joined: 20 Jan 2009
Posts: 825
PostPosted: Mon Feb 21, 2011 7:27 pm Post subject: Reply with quote
As a stand alone stat I'd probably prefer Free Throws Made per 100 Possessions over FTM/FGA for some uses but in a set of 4 Factors FTM/FGA is independent of TO rate (avoiding double counting of a trait) while Free Throws Made per 100 Possessions isn't, if it is based on total possessions and not scoring possessions.
Hasn't the traditional Factor alway been Free Throws Made (not "attempted" as stated in your first paragraph of the post at HSAC) divided by Field Goals Attempted?
Have you calculated "FT+" for the NBA? I'd be interested in seeing that.
Last edited by Crow on Mon Feb 21, 2011 11:19 pm; edited 2 times in total
Back to top
View user's profile Send private message
Chilltown
Joined: 16 Apr 2010
Posts: 15
Location: Boston
PostPosted: Mon Feb 21, 2011 9:09 pm Post subject: Reply with quote
In some places I have seen it as FTM/FGA, but most of the college stats guys (Pomeroy, Statsheet, Team Rankings) calculate it as FTA/FGA. That's principally what I'm arguing against.
I understand that theoretically the Four Factors as presently constituted are independent, but in the data? The correlation between eFG% and FTM/FGA (.23) is larger (as an absolute value) than the correlation between FTM/100 Poss and Turnover Rate.
As for NBA FT+, I have calculated it, but I avoid posting any NBA stuff on the internet because of my job (probably nothing wrong, but I'm trying to be overly cautious). I think DSMok1 has calculated it, so he could probably post it.
Back to top
View user's profile Send private message Visit poster's website
DSMok1
Joined: 05 Aug 2009
Posts: 611
Location: Where the wind comes sweeping down the plains
PostPosted: Mon Feb 21, 2011 11:10 pm Post subject: Reply with quote
Chilltown wrote:
In some places I have seen it as FTM/FGA, but most of the college stats guys (Pomeroy, Statsheet, Team Rankings) calculate it as FTA/FGA. That's principally what I'm arguing against.
I understand that theoretically the Four Factors as presently constituted are independent, but in the data? The correlation between eFG% and FTM/FGA (.23) is larger (as an absolute value) than the correlation between FTM/100 Poss and Turnover Rate.
As for NBA FT+, I have calculated it, but I avoid posting any NBA stuff on the internet because of my job (probably nothing wrong, but I'm trying to be overly cautious). I think DSMok1 has calculated it, so he could probably post it.
Yeah, it's really easy to calculate. I didn't realize you were employed by the Suns! Congrats.
This was a few days back:
Code:
Team FT+ (per 100 Pos) FTORtg NonFTORtg %PosFT
Oklahoma City Thunder 11.6 174.4 98.5 15%
Denver Nuggets 9.5 165.9 103.1 15%
Miami Heat 8.8 161.3 102.2 15%
Utah Jazz 8.6 164.7 100.1 13%
New Jersey Nets 8.6 162.4 94.0 13%
Houston Rockets 8.5 166.2 101.6 13%
New York Knickerbockers 8.3 168.2 101.5 12%
Charlotte Bobcats 8.3 156.8 94.5 13%
Milwaukee Bucks 8.3 156.4 93.0 13%
Cleveland Cavaliers 8.2 156.5 91.7 13%
New Orleans Hornets 8.1 161.3 97.3 13%
Portland Trail Blazers 8.1 167.2 98.7 12%
Washington Wizards 8.1 161.4 94.4 12%
Minnesota Timberwolves 7.7 161.8 97.3 12%
Dallas Mavericks 7.7 165.2 100.4 12%
Indiana Pacers 7.6 164.4 96.1 11%
Los Angeles Lakers 7.6 165.5 104.8 13%
San Antonio Spurs 7.6 163.0 103.5 13%
Philadelphia 76ers 7.6 160.7 98.5 12%
Atlanta Hawks 7.5 165.4 100.0 12%
Phoenix Suns 7.2 161.5 102.7 12%
Memphis Grizzlies 7.2 157.5 99.1 12%
Toronto Raptors 7.0 156.6 98.4 12%
Chicago Bulls 6.8 152.9 99.4 13%
Detroit Pistons 6.7 153.6 98.7 12%
Sacramento Kings 6.7 151.8 96.0 12%
Boston Celtics 6.5 157.4 101.6 12%
Los Angeles Clippers 6.3 145.9 100.6 14%
Golden State Warriors 6.0 158.7 102.1 11%
Orlando Magic 5.7 146.2 102.8 13%
_________________
GodismyJudgeOK.com/DStats
Twitter.com/DSMok1
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Crow
Joined: 20 Jan 2009
Posts: 825
PostPosted: Mon Feb 21, 2011 11:23 pm Post subject: Reply with quote
Ok Chilltown. I guess I've seen FTA/FGA for college. With the NBA it is generally FTM/FGA. That is my focus but you were applying it to college in your HSAC post.
Thanks DSMok1 for the NBA FT+ data.
The average is +7.75 per 100 possessions.
So the Thunder get a net 3.85 points from FT+ offense per 100 possessions or about 3.6 per game at their pace. Given a 2.3 overall point differential, it appears they are -1.3 points below league average on the net of the other 7 factors. Since they are 1.2 points below average on Defensive Rating it works out in summary to being #1 on FTM/FGA, average on the rest of offense (weak on shooting but offsetting on TO rate and offensive rebounding) and a bit below average on defense.
The Heat is roughly +1 and the Magic are -2 on FT+. If both were equal and average on this factor the Magic would have the better remaining point differential from the other 7 factors. If Boston were also average on this instead of -1.2 they would be essentially tied with Orlando for the league lead on point differential from the other 7 factors.
If playoff foul calls go up or down compared to regular season, or become more even between teams or favor superstars more, or just swing between home and away, or if you have room & the ability to crank your FT opportunities up a notch in the playoffs, any or all of that could affect margins or possibly outcomes of certain games and match-ups. Lots can change in every part of the game but FT opportunities are probably the most influenced by officiating. Relying on that may have greater risk.
Back to top
View user's profile Send private message
bbstats
Joined: 25 Apr 2010
Posts: 46
PostPosted: Thu Feb 24, 2011 2:52 pm Post subject: Reply with quote
While I agree with your sentiment that FTM per 100 is a better measure of efficiency, this is pretty evident, since: FTM/100 +FGpts/100 = Efficiency
The cool thing about the four-factors is that each of them interact with one another. Although, to predict efficiency thoroughly, we would also need FT% and FG%, as I have outlined here.
So, in theory, the following is true:
Free Throw Percentage x Free Throw (Attempt) Rate x Field-Goal Attempts = Free Throws Made
EDIT: Sorry, shouldn't have said 'per 100'
The four factors tell us specifically how possessions ended. High turnover % means fewer Field-Goal Attempts and fewer Free Throw Attempts. Higher offensive rebound % means more field-goal (and free throw) attempts, and less need for high eFG% to produce on offense, etc.
_________________
http://thebasketballdistribution.blogspot.com
http://twitter.com/bbstats
Back to top
View user's profile Send private message Visit poster's website
bbstats
Joined: 25 Apr 2010
Posts: 46
PostPosted: Thu Feb 24, 2011 3:16 pm Post subject: Reply with quote
Quickly ran my deductive 'six-factors' on NCAA teams (via Pomeroy's site...had to lazily coddle together some numbers to get FG%)
I bet I would get an even higher R^2 if I used a modified version of Basketball-References formula for possessions (which I think includes team rebounds/shots that go out of bounds).
_________________
http://thebasketballdistribution.blogspot.com
http://twitter.com/bbstats
Author Message
Chilltown
Joined: 16 Apr 2010
Posts: 15
Location: Boston
PostPosted: Mon Feb 21, 2011 12:48 pm Post subject: Re-Evaluating FT Rate as the Fourth Factor Reply with quote
Recently, I've become very interested in how we measure production from the Free Throw line. I've read Dean Oliver's justifications for why FT Rate is the Fourth Factor (because getting to the line is more important in the long run than simply making shots), but that thought always struck me as strange.
I decided to test FT Rate against Free Throws Made per 100 Possessions, a stat which more explicitly measures FT line production in points, and found that by the criteria I could think of, FTM/Poss is a better Fourth Factor.
The full analysis/conclusions are here. I have only done this for offense, because the effects of "free throw line defense" seem more subtle and worthy of a more in depth study.
I would welcome any discussion on this. Hopefully this post gets a good discussion of the Fourth Factor going because I really do believe that FT Rate could be improved.
Back to top
View user's profile Send private message Visit poster's website
Crow
Joined: 20 Jan 2009
Posts: 825
PostPosted: Mon Feb 21, 2011 7:27 pm Post subject: Reply with quote
As a stand alone stat I'd probably prefer Free Throws Made per 100 Possessions over FTM/FGA for some uses but in a set of 4 Factors FTM/FGA is independent of TO rate (avoiding double counting of a trait) while Free Throws Made per 100 Possessions isn't, if it is based on total possessions and not scoring possessions.
Hasn't the traditional Factor alway been Free Throws Made (not "attempted" as stated in your first paragraph of the post at HSAC) divided by Field Goals Attempted?
Have you calculated "FT+" for the NBA? I'd be interested in seeing that.
Last edited by Crow on Mon Feb 21, 2011 11:19 pm; edited 2 times in total
Back to top
View user's profile Send private message
Chilltown
Joined: 16 Apr 2010
Posts: 15
Location: Boston
PostPosted: Mon Feb 21, 2011 9:09 pm Post subject: Reply with quote
In some places I have seen it as FTM/FGA, but most of the college stats guys (Pomeroy, Statsheet, Team Rankings) calculate it as FTA/FGA. That's principally what I'm arguing against.
I understand that theoretically the Four Factors as presently constituted are independent, but in the data? The correlation between eFG% and FTM/FGA (.23) is larger (as an absolute value) than the correlation between FTM/100 Poss and Turnover Rate.
As for NBA FT+, I have calculated it, but I avoid posting any NBA stuff on the internet because of my job (probably nothing wrong, but I'm trying to be overly cautious). I think DSMok1 has calculated it, so he could probably post it.
Back to top
View user's profile Send private message Visit poster's website
DSMok1
Joined: 05 Aug 2009
Posts: 611
Location: Where the wind comes sweeping down the plains
PostPosted: Mon Feb 21, 2011 11:10 pm Post subject: Reply with quote
Chilltown wrote:
In some places I have seen it as FTM/FGA, but most of the college stats guys (Pomeroy, Statsheet, Team Rankings) calculate it as FTA/FGA. That's principally what I'm arguing against.
I understand that theoretically the Four Factors as presently constituted are independent, but in the data? The correlation between eFG% and FTM/FGA (.23) is larger (as an absolute value) than the correlation between FTM/100 Poss and Turnover Rate.
As for NBA FT+, I have calculated it, but I avoid posting any NBA stuff on the internet because of my job (probably nothing wrong, but I'm trying to be overly cautious). I think DSMok1 has calculated it, so he could probably post it.
Yeah, it's really easy to calculate. I didn't realize you were employed by the Suns! Congrats.
This was a few days back:
Code:
Team FT+ (per 100 Pos) FTORtg NonFTORtg %PosFT
Oklahoma City Thunder 11.6 174.4 98.5 15%
Denver Nuggets 9.5 165.9 103.1 15%
Miami Heat 8.8 161.3 102.2 15%
Utah Jazz 8.6 164.7 100.1 13%
New Jersey Nets 8.6 162.4 94.0 13%
Houston Rockets 8.5 166.2 101.6 13%
New York Knickerbockers 8.3 168.2 101.5 12%
Charlotte Bobcats 8.3 156.8 94.5 13%
Milwaukee Bucks 8.3 156.4 93.0 13%
Cleveland Cavaliers 8.2 156.5 91.7 13%
New Orleans Hornets 8.1 161.3 97.3 13%
Portland Trail Blazers 8.1 167.2 98.7 12%
Washington Wizards 8.1 161.4 94.4 12%
Minnesota Timberwolves 7.7 161.8 97.3 12%
Dallas Mavericks 7.7 165.2 100.4 12%
Indiana Pacers 7.6 164.4 96.1 11%
Los Angeles Lakers 7.6 165.5 104.8 13%
San Antonio Spurs 7.6 163.0 103.5 13%
Philadelphia 76ers 7.6 160.7 98.5 12%
Atlanta Hawks 7.5 165.4 100.0 12%
Phoenix Suns 7.2 161.5 102.7 12%
Memphis Grizzlies 7.2 157.5 99.1 12%
Toronto Raptors 7.0 156.6 98.4 12%
Chicago Bulls 6.8 152.9 99.4 13%
Detroit Pistons 6.7 153.6 98.7 12%
Sacramento Kings 6.7 151.8 96.0 12%
Boston Celtics 6.5 157.4 101.6 12%
Los Angeles Clippers 6.3 145.9 100.6 14%
Golden State Warriors 6.0 158.7 102.1 11%
Orlando Magic 5.7 146.2 102.8 13%
_________________
GodismyJudgeOK.com/DStats
Twitter.com/DSMok1
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Crow
Joined: 20 Jan 2009
Posts: 825
PostPosted: Mon Feb 21, 2011 11:23 pm Post subject: Reply with quote
Ok Chilltown. I guess I've seen FTA/FGA for college. With the NBA it is generally FTM/FGA. That is my focus but you were applying it to college in your HSAC post.
Thanks DSMok1 for the NBA FT+ data.
The average is +7.75 per 100 possessions.
So the Thunder get a net 3.85 points from FT+ offense per 100 possessions or about 3.6 per game at their pace. Given a 2.3 overall point differential, it appears they are -1.3 points below league average on the net of the other 7 factors. Since they are 1.2 points below average on Defensive Rating it works out in summary to being #1 on FTM/FGA, average on the rest of offense (weak on shooting but offsetting on TO rate and offensive rebounding) and a bit below average on defense.
The Heat is roughly +1 and the Magic are -2 on FT+. If both were equal and average on this factor the Magic would have the better remaining point differential from the other 7 factors. If Boston were also average on this instead of -1.2 they would be essentially tied with Orlando for the league lead on point differential from the other 7 factors.
If playoff foul calls go up or down compared to regular season, or become more even between teams or favor superstars more, or just swing between home and away, or if you have room & the ability to crank your FT opportunities up a notch in the playoffs, any or all of that could affect margins or possibly outcomes of certain games and match-ups. Lots can change in every part of the game but FT opportunities are probably the most influenced by officiating. Relying on that may have greater risk.
Back to top
View user's profile Send private message
bbstats
Joined: 25 Apr 2010
Posts: 46
PostPosted: Thu Feb 24, 2011 2:52 pm Post subject: Reply with quote
While I agree with your sentiment that FTM per 100 is a better measure of efficiency, this is pretty evident, since: FTM/100 +FGpts/100 = Efficiency
The cool thing about the four-factors is that each of them interact with one another. Although, to predict efficiency thoroughly, we would also need FT% and FG%, as I have outlined here.
So, in theory, the following is true:
Free Throw Percentage x Free Throw (Attempt) Rate x Field-Goal Attempts = Free Throws Made
EDIT: Sorry, shouldn't have said 'per 100'
The four factors tell us specifically how possessions ended. High turnover % means fewer Field-Goal Attempts and fewer Free Throw Attempts. Higher offensive rebound % means more field-goal (and free throw) attempts, and less need for high eFG% to produce on offense, etc.
_________________
http://thebasketballdistribution.blogspot.com
http://twitter.com/bbstats
Back to top
View user's profile Send private message Visit poster's website
bbstats
Joined: 25 Apr 2010
Posts: 46
PostPosted: Thu Feb 24, 2011 3:16 pm Post subject: Reply with quote
Quickly ran my deductive 'six-factors' on NCAA teams (via Pomeroy's site...had to lazily coddle together some numbers to get FG%)
I bet I would get an even higher R^2 if I used a modified version of Basketball-References formula for possessions (which I think includes team rebounds/shots that go out of bounds).
_________________
http://thebasketballdistribution.blogspot.com
http://twitter.com/bbstats