1 of 1
Author Message
BadgerCane
Joined: 05 Sep 2007
Posts: 31
Location: University of Miami-Florida
PostPosted: Tue Oct 16, 2007 6:02 pm Post subject: Weight of Assists Reply with quote
There are so many great and accurate ways to measure passing abillity with statistics. However, there seems to be a tremendous divide on how to weigh passing ability against other stats when creating a complete player measure. The stat is especially hard to prove with regression analysis, because at the team level, an assist ultimately means that the basket was scored. However, I did notice that over the last 20 years there is a correlation of .28 between ast/fgm and winning. So, while that is insignificant, it is not nothing. For individuals, do we set assists equal to turnovers? To field goals made? Offensive rebounds? I don't know if there can be a definitive answer, just hoping to hear some interesting ideas.
Back to top
View user's profile Send private message Send e-mail AIM Address
THWilson
Joined: 19 Jul 2005
Posts: 164
Location: phoenix
PostPosted: Tue Oct 16, 2007 6:30 pm Post subject: Re: Weight of Assists Reply with quote
BadgerCane wrote:
There are so many great and accurate ways to measure passing abillity with statistics.
Do tell. Confused
Back to top
View user's profile Send private message
BadgerCane
Joined: 05 Sep 2007
Posts: 31
Location: University of Miami-Florida
PostPosted: Tue Oct 16, 2007 8:51 pm Post subject: Reply with quote
I like Assist/(TmFG-PlayerFG) to indicate the percentage of shots hit by other players that the player assisted on. Though, high usage players tend to do the best in this category by default. The other one I really like is Assists per 40, with adjustments for TeamFG% and Pace.
Back to top
View user's profile Send private message Send e-mail AIM Address
asimpkins
Joined: 30 Apr 2006
Posts: 245
Location: Pleasanton, CA
PostPosted: Wed Oct 17, 2007 10:54 am Post subject: Reply with quote
There's not really a good answer. Assists vary so much in their worth, and they are inconsistently awarded by scorekeepers. Shot creation is important, but assists are a feeble way of tracking it. PER assigns them a value of 0.66 points (one-third the value of a two point basket) which seems to pass the laugh test and then tries to focus on more concrete stats.
It's not a direct answer to your question, but this thread dealt with an interesting approach for recording shot creation.
Back to top
View user's profile Send private message
bchaikin
Joined: 27 Jan 2005
Posts: 690
Location: cleveland, ohio
PostPosted: Wed Oct 17, 2007 1:20 pm Post subject: Reply with quote
below is data for the past 3 decades of the nba (actually just 29 years, excluding the strike shortened 98-99 season) comparing team ratios of offensive ast/fgm over 82 games to their regular season wins:
wins----teams---fgm/g------ast/g-----ast/fgm---eFG%
65+------008------41.8-------25.7-------.614-------51.1
60-64----037------40.7-------25.4-------.623-------51.2
55-59----077------40.3-------25.0-------.620-------50.2
50-54----084------39.9-------24.4-------.612-------49.5
45-49----102------40.1-------24.4-------.607-------48.9
40-44----136------39.4-------23.6-------.598-------48.6
35-39----086------40.0-------23.4-------.585-------48.3
30-34----077------39.2-------23.0-------.588-------48.1
25-29----063------39.0-------23.0-------.590-------47.7
20-24----051------38.6-------22.4-------.581-------47.2
15-19----031------37.2-------22.0-------.591-------46.0
<15------007------37.1-------21.3-------.574-------45.5
the 1st column is simply a range for the number of games won in a season. the 2nd column is the number of teams that had that many wins in a season. the 3rd and 4th columns are simply the average FGM/g and AST/g for those teams in that win range. the 5th column is the average ast/fgm ratio for all the teams in that win range, and the last column the average effective FG% (combining 2pters and 3pters) for those teams in that win range...
you can see a general decrease in ast/fgm ratio (the 5th column) with a decreasing number of wins, from 60-64 wins down to 35-39 wins...
however you'll also notice that there is little difference in the ratio of ast/fgm for teams with wins in the 35-39 win range compared to teams with wins in just the 15-19 win range - a very large difference of 20 less wins. as a matter of fact teams with just 15-19 wins in a season have actually averaged a higher ast/fgm ratio than teams with 35-39 wins, and almost higher than teams in the 40-44 win range. plus there's a 2.6% difference in eFG% between teams in the 40-44 win range and the 15-19 wins range, but little difference in ast/fgm ratio...
so for well over 1/2 of all the team regular seasons of 82 games played since 1977-78 (sans 1998-99) there has been no direct correlation between ast/fgm ratio and wins. also the difference in ast/fgm for teams in the 45-49 wins range versus 65+ wins is very small (again a difference of 20 wins), just .007 ast/fgm or at most 1/5 of an assist per game...
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Mike G
Joined: 14 Jan 2005
Posts: 3616
Location: Hendersonville, NC
PostPosted: Wed Oct 17, 2007 2:19 pm Post subject: Reply with quote
bchaikin wrote:
wins----teams---fgm/g------ast/g-----ast/fgm---eFG%
60-64----037------40.7-------25.4-------.623-------51.2
...
15-19----031------37.2-------22.0-------.591-------46.0
...
(I deleted the upper and lower-most groups for small-sample reasons.)
Am I seeing that an increase of 3.4 Ast/G (+15%) produces 3-4 times as many wins?
The increase in eFG% is certainly profound. All other things being equal, I'd expect the A team to win 4 of 5 vs that B team.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
kjb
Joined: 03 Jan 2005
Posts: 865
Location: Washington, DC
PostPosted: Wed Oct 17, 2007 2:27 pm Post subject: Reply with quote
What's actually being said with the assist numbers here? Are efgs (and wins) higher BECAUSE of better passing (as indicated by the assists), OR are there more assists because there are better shooters and finishers?
Back to top
View user's profile Send private message AIM Address Yahoo Messenger
BadgerCane
Joined: 05 Sep 2007
Posts: 31
Location: University of Miami-Florida
PostPosted: Wed Oct 17, 2007 4:34 pm Post subject: Reply with quote
(I don't know how to paste charts from excel, any help?)- I thought it might shed some light to look at assists from the defensive side. The results are veyr interesting. Over the last 10 years, there is a .17 correlaction b/w (Ast/FG) allowed by defense and Drtg. However, there is a -.41 correlation between that same number and wins. What is even more remarkable is the consistency that teams have in this stat from year to year. I wish I could put the whole chart up, but if others would look at their team numbers, I'm sure that your eyes would open up very wide at all of the 3-year clusters that pop up where teams allow almost the exact same Ast% on defense.
Back to top
View user's profile Send private message Send e-mail AIM Address
Harold Almonte
Joined: 04 Aug 2006
Posts: 616
PostPosted: Wed Oct 17, 2007 4:52 pm Post subject: Reply with quote
What I heard is that there's a good corelation between FG% and wins (4 factors). And I read at 82games's "The value of a god pass", that a "potential assist", that's a pass with probable assist worth, improves the FG% an 8%; they even value them about 0.16 points each.
What I think is that even some decimals of difference between two teams's assists/gm doesn't say the real, or maybe says a lot about a better team passing and the influence on FG%. It's the same with blocks and defense.
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 3616
Location: Hendersonville, NC
PostPosted: Thu Oct 18, 2007 8:40 am Post subject: Reply with quote
There's little doubt that assists are not a bad thing, and the debate is about how good they are. They don't correlate as strongly to success as some other things. Teams that have great 1-on-1 scorers don't rely on assists so much. Teams that are highly reliant on assists aren't necessarily good.
Assists compensate for weak shot-creation skills, much as OReb compensate for low FG%, steals compensate for weak D, etc. These quantities also complement other strengths. You can't have too much good stuff.
The typical assist, though, is just the measured example of what may typically be 2-3 good passes (potential assists, or PA). Even good passes sometimes become turnovers. So while the occasional TO subtracts from a passer's value, the uncounted PA's are, well, uncounted. A player's Assist count is just some fraction of the value of his overall passing.
Ignoring for now that assists are counted wildly differently from place to place, we can figure that the very best passes are those that lead to the highest-% shots. The FG% on PA determines the Ast total: Ast = PA*FG%. This simplistic formula does not consider how good the receiver/shooter is, but only the effectiveness of the combo(s).
On good teams, a good passer gets lots of assists because his teammates hit shots. On bad teams, he may get assists because his mates rely on them. In both cases, the assists are an indication of that player's 'passing value', to that team.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Rasta978
Joined: 26 Mar 2007
Posts: 56
Location: Orlando, FL
PostPosted: Fri Oct 19, 2007 2:05 pm Post subject: Reply with quote
Ignoring the groups with 65+ and <15 wins (due to small sample size), the average ast/fgm ratio is 0.600. My first question is how far from that mean (in terms of ast/gm) is each subgroup.
Code:
Group Ast/Gm
60-64 -1.0
55-59 -0.8
50-54 -0.5
45-49 -0.3
40-44 0.0
35-39 0.6
30-34 0.5
25-29 0.4
20-24 0.8
15-19 0.3
These are the number of assists that would have to be added (or subtracted) in order to bring the groups' ast/fgm ratio up (or down) to the average of 0.600.
Given that every group is within 1.0 assists of the mean, I don't think this particular ratio tells us much. In fact, such small variations could be attributed to scorekeeper subjectivity.