The Disappointment of the 2007-08 season has been the Chicago Bulls. Chicago was expected to contend with the Boston Celtics in the Eastern Conference. Yes, I thought the Celtics would be better, but I thought the Bulls were clearly the second best team in the conference before the season started. Instead, this team is the worst team in the conference. You can see it in the won-loss record. And you can see it in the team’s efficiency differential (offensive efficiency minus defensive efficiency) which currently stands at -8.5.
When we look at defensive efficiency, we see a team allowing 97.4 points per 100 possessions. This is virtually identical to what we saw in 2006-07 (97.0 points per 100 possessions). On the offensive end, though, we see a team that’s only scoring 88.9 points per 100 possessions. Last year, with virtually the same roster, this team scored 102.2 points per 100 possessions. So what the hell happened?
John Hollinger wrote a very good column detailing this team’s woes. In case you didn’t read it (it is an Insider Column), I will give you the basic argument in one sentence. Chicago has forgotten how to get the ball to go through the hoop.
To see this, consider points-per-shot {or points-per-field goal attempt, or [(PTS-FTM)/FGA, or Adjusted Field Goal percentage*2, these are all the same thing}. Last year the Bulls scored 0.99 points per field goal attempt. This year the Bulls are only scoring 0.83 points per field goal attempt.
To see the impact all these errant shots have had on the Bulls, I made three projections for this team.
The first is how many wins the Bulls could expect this year if each player (except the rookies) performed on a per-minute basis as they did in 2006-07. The second projection takes only the shooting efficiency (points-per-shot again) from 2006-07, but leaves everything else in a player’s performance exactly as it is this season. And the last projection is based solely on what the players are doing this season. All three projections are reported in Table One.
Table One: Projecting the Chicago Bulls
As Table One reports, had the each Chicago player maintained what they did last year, the Bulls would be on pace to win 57 games. This would not be good enough to catch Boston or Orlando, but it would put this team in the thick of the Eastern Conference title chase.
What happens if all we only utilize shooting efficiency from 2006-07, but keep everything else as it is in 2007-08? Now we see the Bulls would be on pace to win 55 games. Yes, virtually the entire problem this team has is tied to shooting efficiency. If this problem were solved, this team would start winning again.
Without this problem being solved, we can see that this is going to be a very sad season. If the Bulls maintain their current shooting problems, we can expect this team to win between 18 and 19 games.
Let’s summarize what we’ve learned. Although people have argued that scoring doesn’t matter much in calculating Wins Produced, we see evidence in this story that a team can dramatically alter Wins Produced simply by changing shooting efficiency. Rebounds, steals, turnovers, etc… can all stay the same, but if the shots don’t go in the team will have trouble winning. Of course, if a team keeps shooting efficiency the same but forgets about rebounds, steals, turnovers, etc…, the team will also have problems. But that’s a story for another day.
Let me make four more observations.
1. At least half of this team’s decline is tied to the performance of Kirk Hinrich and Ben Gordon. That being said, the shooting woes have inflicted the entire roster. So this is not an issue with just one or two players.
2. I don’t know why the shots aren’t going on. We can look at the numbers and see where a team’s problems are (or are not). But the numbers alone are not going to tell us why. In sum, the numbers are not going to do your thinking for you.
3. I would note that this could be entirely a small sample problem. After all, this team has only played 13 games. Because this stretch happened at the beginning of the season it’s very noticeable. A similar streak in mid-season would not be quite the same story.
4. Ben Wallace has played somewhat worse, but I think his decline has been somewhat over-stated. Overall his WP48 has declined from 0.210 to 0.165. But I think part of this is due to how badly he played the first week of the season when he was playing hurt. My sense is a healthy Big Ben will still produce wins this year.
Of course, if this team doesn’t hit its shots, all of Big Ben’s production isn’t going to get this team into the playoffs.
- DJ
Our research on the NBA was summarized HERE.
The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Introducing PAWSmin — and a Defense of Box Score Statistics
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.
33 responses so far ↓
Owen // December 1, 2007 at 8:16 am
Excellent post. Not sure why it bears mentioning, but I noticed yesterday that Luol Deng was born in Wow, Sudan,
Brian // December 1, 2007 at 9:15 am
I agree part of it could be just a sample size issue. But I’d be pretty confident part of it is true decline in performance levels.
The Wizards started off 0-5 or so, and then climbed back to .500. So maybe the Bulls can claw their way back.
Pete // December 1, 2007 at 9:39 am
Durant scored 35 points last night on 12-20 shooting. How do you like them apples?
joe // December 1, 2007 at 9:42 am
Pete: 1 game is a great sample!
Owen // December 1, 2007 at 10:09 am
Pete - I like my apples without worms in them. Durant had seven turnovers. He had a good game, but not a truly great game….
Pete // December 1, 2007 at 10:17 am
Owen,
He had 7 turnovers but that’s because he has the ball most of the time, much more than the average of a player at his position. Wins scored does not adjust for that.
Owen // December 1, 2007 at 10:52 am
Wins Scored? What are those?
Pete, clearly you have an ax to grind. I think most reasonable people agree though that seven turnovers is a lot to commit, no matter how much you are handling the ball.
I will be very happy if Durant turns out to be a star. I loved watching him at Texas, and I certainly am not rooting against him. But I think’s its fair to say he hasn’t been that good yet, and that there is a possibility that he won’t turn out to be a superstar. We shall see though, small sample size so far…
Rashad // December 1, 2007 at 10:53 am
Of course he had 7 turnovers because he had the ball most of the time, but there are lots of players that have the ball most of the time and average 2-3 turnovers per game. I think the point of the comments is that he had a good game from an true shooting percentage perspective, but it was marred by his turnovers.
7 turnovers is really horrendous from a non-point guard. I mean Nash will ocassionally have a ton of turnovers, but there are usually 12-14 assists to go with them.
givensna // December 1, 2007 at 11:47 am
He was also only 8-12 from the line.
Pete // December 1, 2007 at 12:34 pm
Owen,
Win score is explained here–
http://dberri.wordpress.com/2006/05/26/wins-produced-vs-win-score
Here is the calculation: Points + Rebounds + Steals + ½Assists + ½Blocked Shots – Field Goal Attempts – Turnovers - ½Free Throw Attempts - ½Personal Fouls
For last night Durant had an AWESOME game.
35 + 5 + 2 + 1/2 * 1 + 1/2 * 3 - 20 - 7 - 1/2 * 12 - 1/2 * 4 = 9
9 win units in 39 minutes = 0.23 wins scored per minute
Pete // December 1, 2007 at 12:37 pm
The 0.23 figure is his PAWSmin. Sorry if I misused terminology.
Pete // December 1, 2007 at 12:50 pm
Owen,
I hope you don’t construe this as “an ax to grind,” but why do blocks only get 1/2 credit in this win score model. is this b/c only 1/2 of blocks culminate in a change of possession?
And, if so, what if blocks proxy good defense. Dberri shoves most of defense (except d rebounds, blocks + steals) in his residual and then assigns to the players equally. But what if (and this would be an empirical issue) blocks proxy good defense. So dberri’s possession intuition which seems to be embedded in his model wouldn’t be giving us the best assignment of credit to players. Anyone?
dberri // December 1, 2007 at 1:02 pm
Pete,
Quick note…I do not assign a residual to everyone on the team. I assign factors tracked for the team to everyone on the team. These factors are independent variables in the model, not the residual.
Pete // December 1, 2007 at 1:49 pm
Residual Shmidual. I bet your independent variable is just defensive efficiency so your result would be the same as a calculation that ignored defensive efficiency and had that spill out into a residual that could be split up among the team.
Mike H // December 1, 2007 at 2:40 pm
Pete, RTFB.
Pete // December 1, 2007 at 3:01 pm
Mike,
The book does not fully go into the mathematics. If blocks, defensive rebounds, and steals proxy for defensive ability than they should get a higher weight than just giving them a weighting commensurate with what they do for the team with respect to possessions gained/lost. that’s all I’m saying.
Pete // December 1, 2007 at 3:02 pm
It’s like a cult here. The slightest criticism and everyone goes bonkers.
Owen // December 1, 2007 at 3:04 pm
Pete - I am familiar with Win Score, and I wouldn’t call that an “awesome game.” Andrei Kirilenko had an awesome game last night. Dwight Howard had an awesome game last night. Durant had the kind of game that star players have practically every day in the NBA. Manu, another shooting guard, had a much better game than Durant last night.
I am encouraged by Durant’s ability to get to the free throw line and convert there. He is already 80-98 from the line this year. He is in the top twenty in fta, which is a crucial statistic for offensive efficiency. His ts% is inching towards average also. I think he may very well figure it out, but it hasn’t quite happened yet, and you can’t be sure it will happen. And that’s the simple point DB has been making about Durant, a point which you seem anxious to take exception to for some obscure reason.
I can picture Durant playing well. If he works on his shot selection, continues to get to the line a lot, if he focuses a little bit more on rebounding, while cutting back on the turnovers as so many young players seem to manage to do, he could be extremely productive. And when that actually happens, when Durant is actually playing well, I will be the first to say you were right about him. How is that Pete?
Owen // December 1, 2007 at 3:07 pm
And also, no one minds disagreement here, but you do seem a tad confrontational, starting with the “how about them apples” and following up with the residual schmidual.” Just try to keep it civil…
Pete // December 1, 2007 at 3:12 pm
Owen, very good analysis. I looked up how 0.23 / min compared to the top players. You’re right, some of them do even better than that and across the whole season.
I do kind of like how wins scored reveals what area (e.g. rebounding, shooting %) is hurting the player’s productivity. I’m just trying to get a handle on the construction right now. That’s why I’m curious if there’s a way to proxy defense using box score statistics.
Pete // December 1, 2007 at 3:14 pm
My problem with the Durant comments following one of the earlier posts on this site were mainly that his performance through 10 games was being compared with other player’s rookie seasons. The more apt comparison would have been a comparison with their first 10 games.
dberri // December 1, 2007 at 3:15 pm
Pete,
You lose the bet. The team defensive adjustment, which is explained here
http://www.wagesofwins.com/CalculatingWinsProduced.html
incorporates the independent variables in the wins model that are not tracked for individual players. These include made field goals by the opponent, opponent turnovers that are not steals, and team rebounds that change possession. It is not the residual.
I think the problem you are having in this discussion is that people who read this forum on a regular basis already know this stuff. You are not taking the time to find out how things work before commenting.
You are free to comment without understanding, but you are going to find that others are going to react as they have today.
Owen // December 1, 2007 at 3:46 pm
Pete - To see how DB used +/- data to proxy “non box score defense”, see:
http://dberri.wordpress.com/2007/03/26/incorporating-defense/
dbg // December 1, 2007 at 4:08 pm
And speaking of the Bulls, that 8th playoff spot is still within reach. That is if they get back to their former form. Knicks on the other hand…guaranteed lottery season.
Mike H // December 1, 2007 at 4:10 pm
Pete, I posted the snarky RTFB primarily due to the tone of your post. If you had simply posited your concerns without mockery then I doubt anyone would have bonked out.
Pete // December 1, 2007 at 9:18 pm
Thank you everyone for the informative replies. I especially like the idea of combining +/- with win score to better attribute defense at the individual level. (In this good post that Owen pointed out to me– http://dberri.wordpress.com/2007/03/26/incorporating-defense)
dberri // December 1, 2007 at 10:31 pm
Hopefully that is something that can be done better in the not too distant future. It would be nice to combine what plus-minus tells us about defense into Wins Produced. I am not entirely satisfied with my approach. But it was a start.
GV // December 1, 2007 at 11:00 pm
The Rockets have also forgotten how to shoot. Shane, Rafer, James, and Bonzi couldn’t hit an open shot to save their life.
The Franchise // December 2, 2007 at 9:17 am
WS/WP is a metric completely derived from the box score, right? Which makes it a very good measure. It may not be quite as accurate as using some +/- data with it to measure defensive contribution, but simplicity should definitely be a factor in evaluating the usefulness of a model.
Of course, the argument of WOW is that the very simple model most often used—ppg and, to a lesser extent, rpg, is very poor in its ability to explain results. If a +/- refinement adds greatly to explanatory power, that would be great, but it seems like it would be a model unsuited for casual use, whereas I can look at a box score and figure WS on my own.
Pete // December 2, 2007 at 9:29 am
Franchise, RTFB.
Just kidding.
I agree with you. If win score is the best available combination of box score statistics then that is a very useful metric. That’s not to say a combination of box score statistics captures every aspect of player performance and ability, but it is useful to have a summary measure that translates a bunch of box score stats.
For increased precision it might be good to add the +/- refinement. Perhaps the incremental precision isn’t necessary for casual use but for purposes of GMs predicting performance the incremental precision could be sufficiently worthwhile.
Pete // December 2, 2007 at 9:33 am
Franchise, Win score is derived completely from the box score. I was a skeptic at first until I read this– http://dberri.wordpress.com/2006/05/26/wins-produced-vs-win-score/ It convincingly explains how wins scored is composed.
The Franchise // December 2, 2007 at 8:53 pm
Sorry about the unclear tone–I was asking that rhetorically. It’s just an easy and effective way to actually measure player performance in a game I didn’t watch. A great example was the Jazz-Lakers game a few days ago. The Jazz won handily, and the headline was about Deron Williams having a career-high in points. However, he also had nine turnovers, so his contribution was solid, with a Win Score of 14, which is a good mark for a point guard, but not heroic. The hero of the game was Andrei Kirilenko, who had double digit assists and rebounds, efficient scoring, no turnovers, and several steals and blocks, which created a WS of 29–extremely good for a small forward.
Adding +/- may allow even better analysis of that game, but I couldn’t have managed it on my own with newspaper in hand. Thus, a model like Win Score is quite useful, even when it’s not as accurate as WP, or WP with a +/- adjustment.
Better Experts Needed: More on an article in the Philadelphia City Paper « The Wages of Wins Journal // April 19, 2008 at 3:17 pm
[...] for other teams - Chicago and Dallas immediately come to mind– past performance doesn’t match perfectly with the [...]
Leave a Comment