| View previous topic :: View next topic |
| Author |
Message |
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
|
 Posted: Wed Apr 08, 2009 10:34 am Post subject: How Do Euroleague Statistics Translate to the NBA? |
 |
|
A few weeks ago I published an article in which I looked at how well NCAA stats translate to the NBA. I remember that at the Sloan Conference Mike Zarren talked about how one thing teams haven't figured out yet is translating European stats to the NBA. Today is the first step of my attempt to do that. You can find the article here:
http://basketball-statistics.com/howdoeuroleaguestatisticstranslatetothenba.html
-Jon Nichols |
|
| Back to top |
|
 |
Carlos
Joined: 21 Jan 2005
Posts: 64
Location: Montevideo, Uruguay
|
| Posted: Thu Apr 09, 2009 9:54 am Post subject: |
|
|
| Hollinger has done some work on translating Euroleague stats too. He has been working on it for quite some time, if I remember well. |
|
| Back to top |
|
|
Carlos
Joined: 21 Jan 2005
Posts: 64
Location: Montevideo, Uruguay
|
|
| Back to top |
|
|
billfromBoston
Joined: 07 Apr 2009
Posts: 1
|
| Posted: Thu Apr 09, 2009 10:09 am Post subject: Question regarding both NCAA and Euro... |
|
|
I was just curious as to whether or not you simulated the predicted statistics for any NBA players of the past to gauge how accurate or proximate they were?
I've seen your stuff for Hansborough, Blair, and Lawson...can you determine how likely these values are to be based off historical outcomes?
Thanks... |
|
| Back to top |
|
|
mikez
Joined: 14 Mar 2005
Posts: 75
|
| Posted: Thu Apr 09, 2009 10:31 am Post subject: |
|
|
Just to be clear - at Sloan I never said anything about "translating" European stats. The question was about data sources, and I believe I said that sometimes it's difficult to get accurate/reliable data on some foreign (including European) leagues. I did not (and cannot) comment on how we might use this data.
I also definitely didn't say anything about what other teams have or haven't figured out, since they don't tell me what they're doing.
-MZ |
|
| Back to top |
|
|
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
|
| Posted: Thu Apr 09, 2009 10:40 am Post subject: |
|
|
| mikez wrote: | Just to be clear - at Sloan I never said anything about "translating" European stats. The question was about data sources, and I believe I said that sometimes it's difficult to get accurate/reliable data on some foreign (including European) leagues. I did not (and cannot) comment on how we might use this data.
I also definitely didn't say anything about what other teams have or haven't figured out, since they don't tell me what they're doing.
-MZ |
I apologize...didn't mean to misquote you. I threw that in more as an interesting sidebar, but I'll be sure to take it out of my article.
I also should mention that I got the statistics from www.keyhoops.com , a site I found on this board.
-Jon |
|
| Back to top |
|
|
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
|
| Posted: Thu Apr 09, 2009 10:45 am Post subject: |
|
|
I do slightly remember him doing something similar (just like he did with college players), so thanks for finding the link. His system may be more accurate than mine since he's been working on it for a little while, but I'd be interested to see if we get similar projections.
-Jon |
|
| Back to top |
|
|
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
|
| Posted: Thu Apr 09, 2009 10:50 am Post subject: Re: Question regarding both NCAA and Euro... |
|
|
| billfromBoston wrote: | I was just curious as to whether or not you simulated the predicted statistics for any NBA players of the past to gauge how accurate or proximate they were?
I've seen your stuff for Hansborough, Blair, and Lawson...can you determine how likely these values are to be based off historical outcomes?
Thanks... |
If you're talking about my projections using NCAA stats, I have done historical projections, and they are quite accurate. The thing is, those projections were run on players who were accurately scouted. Once they found a place in the league, their numbers were similar to what BSPS predicted. With new rookies, the question is whether or not they'll have a place in the NBA. For example, if Blair becomes an NBA rotation player, I'd think he would perform similarly to my projected numbers.
I know that makes it sound like my system is useless, but I just want to make sure I'm not touting it as something better than it is.
-Jon |
|
| Back to top |
|
|
Kevin Pelton
Site Admin
Joined: 30 Dec 2004
Posts: 978
Location: Seattle
|
| Posted: Thu Apr 09, 2009 10:51 am Post subject: |
|
|
Jon, are you using career stats here as with the college players?
JohnH has argued in the past that Euroleague stats are more accurate predictors of NBA success than college success, a notion I generally agree with in an anecdotal fashion. It was a little surprising to see you get lower correlations. Using career stats might have something to do with that.
There are also a handful of players each year who come over from EuroCup, which I don't think anyone else has ever looked at.
The problem, as MikeZ suggests, is guys who come without having played in either of the major European competitions (Andris Biedrins comes to mind). The national leagues, naturally, differ widely in their level of play. |
|
| Back to top |
|
|
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
|
| Posted: Thu Apr 09, 2009 11:11 am Post subject: |
|
|
| Kevin Pelton wrote: | Jon, are you using career stats here as with the college players?
JohnH has argued in the past that Euroleague stats are more accurate predictors of NBA success than college success, a notion I generally agree with in an anecdotal fashion. It was a little surprising to see you get lower correlations. Using career stats might have something to do with that.
There are also a handful of players each year who come over from EuroCup, which I don't think anyone else has ever looked at.
The problem, as MikeZ suggests, is guys who come without having played in either of the major European competitions (Andris Biedrins comes to mind). The national leagues, naturally, differ widely in their level of play. |
Yes, I did use career stats. My correlations are slightly lower than with the NCAA, although I haven't done regressions using multiple variables yet. Perhaps when I add that the R^2's will be higher.
Like John, I picked the Euroleague because it is the second highest level of play and most (but definitely not all) international prospects have played in that league at some point. Biedrins and Tony Parker never played in the Euroleague, if I recall correctly.
I thought about getting data on each league from around Europe and translating those numbers to the NBA, but the sample size is too small for now. Even with the Euroleague, my sample was only slightly over 40 players.
-Jon |
|
| Back to top |
|
|
mikez
Joined: 14 Mar 2005
Posts: 75
|
| Posted: Thu Apr 09, 2009 1:18 pm Post subject: |
|
|
| JNichols42887 wrote: | | I apologize...didn't mean to misquote you. I threw that in more as an interesting sidebar, but I'll be sure to take it out of my article. |
No offense taken at all; confidentiality rules here mean I can't be talking about what our team is or isn't doing with any particular type of stats, so I just figured I'd make sure it was clear I wasn't doing that.
Data availability and (perhaps more importantly, long term) reliability in other leagues can be an issue (actually, it can be an issue in the NBA too, though perhaps not as much), and it's one reason I'm happy not to be dealing much with high school stats anymore. If a guy averaged 25 assists/game in high school, who's to say it wasn't his younger brother at the scorer's table giving the guy an assist for every team scoring possession in which the guy touched the ball? And this ignores the varying-league-quality issue Kevin notes -- if Al Jefferson played one-on-one against me, he might not score any more points than Rafael Araujo would if he played against me. But one guy is a top NBA scorer, and the other is not in the NBA.
-MZ (yes, both of them would probably score on every possession) |
|
| Back to top |
|
|
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
|
| Posted: Thu Apr 09, 2009 1:48 pm Post subject: |
|
|
| mikez wrote: | | JNichols42887 wrote: | | I apologize...didn't mean to misquote you. I threw that in more as an interesting sidebar, but I'll be sure to take it out of my article. |
No offense taken at all; confidentiality rules here mean I can't be talking about what our team is or isn't doing with any particular type of stats, so I just figured I'd make sure it was clear I wasn't doing that.
Data availability and (perhaps more importantly, long term) reliability in other leagues can be an issue (actually, it can be an issue in the NBA too, though perhaps not as much), and it's one reason I'm happy not to be dealing much with high school stats anymore. If a guy averaged 25 assists/game in high school, who's to say it wasn't his younger brother at the scorer's table giving the guy an assist for every team scoring possession in which the guy touched the ball? And this ignores the varying-league-quality issue Kevin notes -- if Al Jefferson played one-on-one against me, he might not score any more points than Rafael Araujo would if he played against me. But one guy is a top NBA scorer, and the other is not in the NBA.
-MZ (yes, both of them would probably score on every possession) |
Agreed...that's why doing this stuff is so scary. Who knows what crazy numbers my system is going to spit out?
-Jon |
|
| Back to top |
|
|
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
|
| Posted: Thu Apr 09, 2009 1:51 pm Post subject: |
|
|
Here are the adjusted R^2's from the multiple linear regression that I will be using for my Box Score Prediction System:
FGA: 0.2971
FG%: 0.3004
3PA: 0.6834
3P%: 0.7729
FTA: 0.4111
FT%: 0.6887
REB: 0.8609
AST: 0.7202
STL: 0.5606
BLK: 0.7964
TO: 0.4538
PF: 0.5432
-Jon |
|
| Back to top |
|
|
trk
Joined: 26 Feb 2009
Posts: 12
|
| Posted: Fri Apr 10, 2009 7:09 pm Post subject: |
|
|
| The problem with predicting a player's NBA stats based on their Euroleague stats is that Euroleague is more of a tournament than a real league. The Euroleague "regular season" only lasts 10 games, so if a player's team is eliminated early the predictive power of their stats suffers due to small sample size. Using Euroleague stats to try to predict the career of a player like Ricky Rubio, who only played 5 Euroleague games this year and was playing through a significant injury, isn't going to work very well. The average level of competition that a team faces will also vary depending on how far the team gets in the tournament. Using the stats that a player gets in a strong national league like the Spanish ACB league might end up giving a more accurate assessment of the player's ability than Euroleague stats. |
|
| Back to top |
|
|
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
|
| Posted: Sat Apr 11, 2009 10:49 am Post subject: |
|
|
| trk wrote: | | The problem with predicting a player's NBA stats based on their Euroleague stats is that Euroleague is more of a tournament than a real league. The Euroleague "regular season" only lasts 10 games, so if a player's team is eliminated early the predictive power of their stats suffers due to small sample size. Using Euroleague stats to try to predict the career of a player like Ricky Rubio, who only played 5 Euroleague games this year and was playing through a significant injury, isn't going to work very well. The average level of competition that a team faces will also vary depending on how far the team gets in the tournament. Using the stats that a player gets in a strong national league like the Spanish ACB league might end up giving a more accurate assessment of the player's ability than Euroleague stats. |
Those are valid points. However, I only included players in my sample that had played in at least 20 Euroleague games. That's still not a ton, but it's something.
-Jon |
|
| Back to top |
|
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
