The Oden Factor
by lee3022 on Nov 23, 2008 11:38 PM PST
Email
PrintThe Question:
The recent excellent post by Dave responding to Henry Abbott has started all of us thinking about how to measure Greg Oden's becoming a significant factor on behalf of the team.
In my analysis of the games through 11/19/08 I found several measures that say he has definitely boosted the team. His effect is not seen in how many points or rebounds he gets or even his plus/minus. It is seen in improvement in overall team defense and overall team production and from production from the Center position.
Joel Przybilla is playing his best basketball. Why? Joel tells us why: He can finally be aggressive and risk some fouls because Greg is his backup. But even more so these two Centers are (one of) the best pair in the league on defense. They are intimidating. Teams that want to beat the Blazers have to hit outside shots. Perversely the Blazer perimeter defense has improved as well forcing them inside.
Data derived from NBA.com and ESPN.com box scores of each game.
Here is the analysis:
The interior defense is being measured by points in the paint, percentage of shots in the paint and percentage makes from 2-point range. The Oden Factor considers the 6 games Oden was out of the lineup or not playing significant minutes against the 5 games following his return to the rotation (Miami through Sacramento). I do realize that the Phoenix game is an anomaly to this study but the work was done before that game and should continue to play out with anyone except a highly motivated Shaq and Amare.
Opponent average points in the paint w/o Oden = 40.0
Opponent average points in the paint with Oden = 35.6
Opponent 2-point FG% w/o Oden = 51.7%
Opponent 2-point FG% with Oden = 45.5%
Opponent ratio of 2-point shots w/o Oden = 74.5%
Opponent ratio of 2-point shots with Oden = 78.5% (see perimeter defense)
Team Defensive blocks from Center w/o Oden = 0.7
Team Defensive blocks from Center with Oden = 3.8
The perimeter defense has also taken a significant leap forward as measured by opponent 3-point shooting percentages. This reason for this might partly be the intimidating presence of the shot blocker in the interior allowing the perimeter defenders to play tighter.
Opponent average 3-point FG% w/o Oden = 41.2%
Opponent average 3-point FG% with Oden = 31.7%
These comparisons show a positive change with Oden in every defensive category except opponent offensive rebounds and the 17 OR by Chicago skews this significantly.
The interior offense has gained a boost from Oden as well as measured by points in the paint, shots from 2-point range & accuracy and rebounds of our shots taken. I also present the scoring improvement from Center
Team average points in the paint w/o Oden = 33.3
Team average points in the paint with Oden = 40.0
Team 2-point FG% w/o Oden = 46.7%
Team 2-point FG% with Oden = 47.9%
Team ratio of 2-point shots w/o Oden = 74.5%
Team ratio of 2-point shots with Oden = 78.5%
Team average offensive rebounds from Center w/o Oden = 3.0
Team average offensive rebounds from Center with Oden = 4.0
Team average points from Center w/o Oden = 4.2
Team average points from Center with Oden = 21.6
The perimeter offense has benefited also as measured by 3-point percentage.
Team 3-point FG% w/o Oden = 41.2%
Team 3-point FG% with Oden = 42.9%
The total offense has been steady at 99 points per game with and without Oden.
The total defense has been dramatically improved:
Opponent average points w/o Oden =100.0
Opponent average points with Oden = 90.2
Conclusion:
In my opinion Greg Oden has already made the team significantly better. He has improved nearly every aspect of the team while admittedly has more improvement to come. We are not sacrificing anything to play Greg as many minutes as his stamina can endure, while keeping Joel highly productive as well. Expect the coming months to add to this meager sample size and validate with even more improvement.
(With apology to the previous posters on this subject. This has taken me some time and presents a bit more data)
20 comments
|
3 recs |
Do you like this story?
Comments
Henry has lost some of his shine
It happens to the best journalists when they reach the peak of their profession. Look at Colin Cowherd and Katie Couric for example.
just b/c you disagree with him
doesn’t mean he has lost his shine. Henry is an excellent journalist and you shouldn’t knock on him just because he said something negative about the Blazers. We don’t need fans with blinders on
Woof
by Charles Barkley McLovin on Nov 24, 2008 1:04 AM PST up reply actions
great post
thanks for putting actual figures to what everyone is trying to explain. GO has a huge impact on this team and he still has a ways to go. Sit back, relax, and enjoy the annual playoff berths that this Blazer team will achieve for years to come.
Rip City REVIVAL
We never ever blew teams out in the Nate era
We have twice with Greg already. Good enough for me.
"Brandon Roy is your favorite restaurant" - Dave
Good post
But….
Doesn’t the Chicago game skew these numbers somewhat? Since I don’t think we can claim to be consistently 42 points better than them, I think we have to view that game as an aberration. So 20% of your sample size on the “with Oden” numbers is significantly skewed (we might be 15 points better, but not 42). In light of that, a lot of these comparisons are not really trustworthy.
The facts are right, but they perhaps don’t mean very much. If the numbers hold when the sample size is double what it is now, then we’ve got something to discuss.
One other factor is that the with Oden schedule has not been as difficult as the without Oden schedule. There’s an element of comparing apples and oranges here.
I think Greg makes us a lot better, too. I’m just not sure we can really argue the case statistically, yet. Right now, the blowouts are statistical out-liers. If they become habitual (please!) then that would bolster your arguments here. Right now, I think it deserves the good old Scottish jury verdict, “Unproven” (which means we know you probably did it, but we can’t prove it, so you’ll get off this time, but the cops are watching you from now on, mate).
The most amazing thing about my amazing ego is I have amazingly little about which to be egotistical.
The pick and roll this year will emphasize "roll" followed by "dunk", followed by the wailings and lamentations of your women.
With that logic
Doesn’t the LA game skew the numbers as well? (I am assuming the OP included the LA game) We might not be as good as the Lakers, but we are better than what we played on opening day.
Yep
But not as significantly as the Chicago win.
Blowouts are very nice (or very bad, when you are on the receiving end), but they significantly skew any kind of statistical analysis when you’ve got a small sample size. A 40 point win is really not twice as good as a 20 point win, it only means your 8-12 guys put a thumping on their 8-12 guys. But statistics don’t care about that aspect of it.
The most amazing thing about my amazing ego is I have amazingly little about which to be egotistical.
The pick and roll this year will emphasize "roll" followed by "dunk", followed by the wailings and lamentations of your women.
Actually the chicago game skewed the rebounding numbers as noted
But there are always anomalies to every study. As noted the sample size is small but comperable on both sides. With a greater sample size should come also greater team production (Miami skewed the numbers as his first game back) as Greg and Joel force defensive strategies and as Greg improves.
when you’ve been in a dry land even a little dew looks inviting —Dave
"When you want to win a game, you have to teach. When you lose a game, you have to learn." - Tom Landry
Skewed against the Blazers that is
when you’ve been in a dry land even a little dew looks inviting —Dave
"When you want to win a game, you have to teach. When you lose a game, you have to learn." - Tom Landry
Sorry the LA game was not included
Since Greg played only a part of the game it fit neither criteria and was the one game omitted. I do think from studying that game that he was doing very well against Bynum. Head-to-head Oden had 3 offensive rebounds, 1 defensive rebound and 1 block of Bynum. Bynum had 1 defensive rebound. However Greg missed all four of his FGA.
when you’ve been in a dry land even a little dew looks inviting —Dave
"When you want to win a game, you have to teach. When you lose a game, you have to learn." - Tom Landry
For what it's worth...
There was evidence of Oden’s effect on the defensive end before the Chicago game.
How about this for a buzz-kill? TOO SMALL A SAMPLE SIZE TO DRAW MEANINGFUL INFERENCES EITHER WAY.
"Sergio is clearly the MJ of 3rd string point guards!" —Mortimer Pritchard
Could be...
But if you regress the opponents offensive efficiency on whether or not oden plays, homecourt advantage, and the opponent’s average offensive efficiency, the “oden effect” comes up as statistically significant (with “White” standard errors). When Oden plays, the opponent’s offensive efficiency is -13.9 points lower, controlling for the opponents average offensive efficiency and home court advantage. The 95% confidence interval for the Oden effect is -25.7 to -2.1, which suggests that there is a 95% chance that Oden has had a negative effect on the opponent’s offensive efficiency.
The regression that I described might not correctly identify the “Oden effect”, but the sample size itself is large enough to identify an effect. In addition, as simplistic as that mode might sound, it describes 62% of the variance in opponents offensive efficiency.
So, why might regression analysis produce the wrong answer in this case? There are two main reasons: 1) an omitted variable, 2) non-normal or non-independent distribution of errors. Regarding the first problem, there could be some other factor that affects opponents offensive efficiency that is not included in the model, like opponent injuries, a time trend for the Blazers defensive strength, an interactive effect between the “type” (like perimeter vs. interior oriented offenses) of team the Blazers play, the opponents strength of schedule, or a non-linearity in the relationship between opponents average efficiency and their performance against the Blazers. For what it’s worth, if you include a variable for opponent injuries and count the Miami (Marion), Chicago (Deng), Sacramento (Martin) games where opponents were missing a star player and the San Antonio (Parker) as one where they had star player that has not played most of the season, the Oden effect is smaller (9.3 instead of 13.9 just misses statistical significance.
Regarding the second problem, it’s possible that the errors in the model or more variable than one would expect using normally distributed errors. Secondly, statistical evidence that I have referred to rests on the assumption that the errors in each observation are independent—-if teams go on “streaks”, then each game is not an independent observation, and the number of observations is effectively overstated. Both of these problems could be addressed, in principle.
I would argue that we do not have to worry about a problem that typically plagues observational analysis: endogeneity or reverse causality. That is, unless you believe in conspiracy theories, whether or not Oden has played has not been caused by how well the opponent is expected to play against Portland. Similarly, whether or not Oden played was not caused by any factors that affect opponent scoring efficiency. Injuries provide a relatively nice natural experiment for assessing the impact of a player on a given team. By contrast, plus/minus ratings suffer severly from the problem of reverse-causality (or endogeneity or selection bias) become coaches put players in the game depending on who is playing for the other team.
In plain English
PoliSam just said that he thinks the Oden effect that we are all seeing is showing up in the statistics, even when you consider strength of opponents and home court advantage. And he thinks that even with the small sample size, it is probably statistically significant.
And I would say that there are too many variables to be sure with such a small sample size, but he may be right.
The most amazing thing about my amazing ego is I have amazingly little about which to be egotistical.
The pick and roll this year will emphasize "roll" followed by "dunk", followed by the wailings and lamentations of your women.
Well said and authoritive beyond my simple calculations
I am working on expanding the model comparing the opponent’s production up to 11/19 without the Blazers against the Team’s game with those opponents. I have captured the data and simply need time to finish the study.
Thanks for the assist.
when you’ve been in a dry land even a little dew looks inviting —Dave
"When you want to win a game, you have to teach. When you lose a game, you have to learn." - Tom Landry
I will continue to track this progress but hope that the sample size for the "without Oden" remains small.
One thing from the study not deliniated above is that Minnesota is included in both samples. The game without Oden was at home and with Oden away from home. In those two games we improved significantly in fast break points (12) and points in the paint (8) net of Minnesota’s production. But also improved in net: turnovers, offensive and defensive rebounds, 3-point ,2-point FG attempted, 3-point FG attempted, and assists. We declined in net: points given up off turnovers, FT attempted and made and 2-point FG, Given the normal home vs away decline in shooting % these two games support the larger (but still small) sample size above.
when you’ve been in a dry land even a little dew looks inviting —Dave
"When you want to win a game, you have to teach. When you lose a game, you have to learn." - Tom Landry
Mike Barrett posts the numbers that make this analysis even more compelling
His reasoning , in part, is that we are holding down opponent rebounding.
he also notes:
• The Trail Blazers also crack the Top 10 (in fact, 10th to be exact) in center PER with a 19.3 rating. Greg Oden is 6th in the league in PER among centers with a 21.27 rating with Joel Przybilla coming in 11th among centers with a 18.40 rating.
Good article go read.
when you’ve been in a dry land even a little dew looks inviting —Dave
"When you want to win a game, you have to teach. When you lose a game, you have to learn." - Tom Landry
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
FanShots
Quick hits of video, photos, quotes, chats, links and lists that you find around the web.
Recommended FanShots
Recent FanShots
In Association With
Ball Don't Lie - NBA - Yahoo! Sport
- Days of NBA Lives: Wherein Roy Hibbert tap-dances to Maine
- Grizzlies Valentine’s Day cards are the key to love
- Kevin Love would like the NBA to take the MLB’s approach on All-Star games
- Kobe Bryant fires a few hilarious salvos at Jeremy Lin
- Brandon Jennings states the obvious regarding his 2014 free agency, everyone goes nuts
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
