07/27/2008
While this is still not perfect, since both players may have played more or less minutes against a particular opponent's lineup, it is closer to an apples to apples comparison - and the story it tells is far different from their season long numbers. As expected, the offense was far stronger with Matt on the court than McInnis, but the defense was just as much in favor of Jeff, causing the two to be about equal in terms of net plus/minus. While this is by no means definitive, it would seem to change the perception of McInnis at least slightly.
How about a similar comparison between Emeka and Nazr, with the control group of Matt, Raymond, Jason, and Gerald? I have to use Matt, because Nazr came along late enough that he did not have as many opportunities with McInnis. Here are those numbers.
These results were even more suprising to me than the McInnis/Carroll comparison. I know that a game and half worth of possessions is not the basis for in-depth statistical study, but at the same time, it would seem to be enough for a sense of how players worked together. And these results are not in Emeka's favor - at all. Interesting and something else to consider as the Bobcats attempt to resign him?
Last comparison that I want to do is to see something with Jared Dudley. I will use whatever lineup he was on the floor with most and then swap him out for a player in another, highly used lineup. Or not - that resulted in a comparison of 12 and 36 possessions. Not good enough. So, the next one down is: Emeka, Gerald, Jason, and Raymond. The comparison will be to Matt Carroll - let us see if we can have a third suprise.
I am glad to see that both of these pairings was actually quite successful for the Bobcats, since both Jared and Matt are players I said the Bobcats should build around (among others). I was hoping to compare Jared to Gerald, but there just were not enough common opponents, mainly to due to Jared most of his minutes when Crash was out.
That concludes this look at apples to apples comparisons - sure, it is more like red delicious compared to granny smith, but at least it is an improvement. It should help serve as a reminder to think about all of the possible causes associated with a number - though I am still glad the Bobcats let McInnis go.
Click here to comment/view comments for this post.
If you take a look at the stats page, you will notice that the first two lineups listed as most commonly used are very similar: There is a single player difference between, a swtich of Matt Carroll for Jeff McInnis, to round out a five with Emeka Okafor, Gerald Wallace, Jason Richardson, and Raymond Felton. Oh, and the other difference is the results the lineups got: The squad with Jeff put up a net plus/minus of -3.0 points per 100 possessions, while Matt's group was +5.6. Quite an indictment of McInnis, right?
Well, maybe. But there is more to it than that. Consider the strength of the opponents played: Because McInnis started a fair amount this year, would he not have played against better overall competition? He probably would have, so I want to look at this in another way - we will only consider the numbers for common opponents. In this case, I mean both Jeff and Matt (along with the same 4) played against the same 5 players from the opposition. When you look at it that way, the numbers change a great deal.
| Bobcat Stats with Mek, Crash, J-Rich, and Felton against common opponents | ||||
|---|---|---|---|---|
| Player | Possessions | Off Eff | Def Eff | Net Eff |
| Matt Carroll | 189 | 101.1 | 123.2 | -22.2 |
| Jeff McInnis | 464 | 92.5 | 114.4 | -21.9 |
While this is still not perfect, since both players may have played more or less minutes against a particular opponent's lineup, it is closer to an apples to apples comparison - and the story it tells is far different from their season long numbers. As expected, the offense was far stronger with Matt on the court than McInnis, but the defense was just as much in favor of Jeff, causing the two to be about equal in terms of net plus/minus. While this is by no means definitive, it would seem to change the perception of McInnis at least slightly.
How about a similar comparison between Emeka and Nazr, with the control group of Matt, Raymond, Jason, and Gerald? I have to use Matt, because Nazr came along late enough that he did not have as many opportunities with McInnis. Here are those numbers.
| Bobcat Stats with Matt, Crash, J-Rich, and Felton against common opponents | ||||
|---|---|---|---|---|
| Player | Possessions | Off Eff | Def Eff | Net Eff |
| Emeka Okafor | 149 | 89.3 | 108.2 | -18.9 |
| Nazr Mohammed | 131 | 105.3 | 100.0 | +5.3 |
These results were even more suprising to me than the McInnis/Carroll comparison. I know that a game and half worth of possessions is not the basis for in-depth statistical study, but at the same time, it would seem to be enough for a sense of how players worked together. And these results are not in Emeka's favor - at all. Interesting and something else to consider as the Bobcats attempt to resign him?
Last comparison that I want to do is to see something with Jared Dudley. I will use whatever lineup he was on the floor with most and then swap him out for a player in another, highly used lineup. Or not - that resulted in a comparison of 12 and 36 possessions. Not good enough. So, the next one down is: Emeka, Gerald, Jason, and Raymond. The comparison will be to Matt Carroll - let us see if we can have a third suprise.
| Bobcat Stats with Emeka, Gerald, J-Rich, and Felton against common opponents | ||||
|---|---|---|---|---|
| Player | Possessions | Off Eff | Def Eff | Net Eff |
| Jared Dudley | 115 | 114.8 | 103.3 | +11.5 |
| Matt Carroll | 81 | 124.7 | 100.0 | +24.7 |
I am glad to see that both of these pairings was actually quite successful for the Bobcats, since both Jared and Matt are players I said the Bobcats should build around (among others). I was hoping to compare Jared to Gerald, but there just were not enough common opponents, mainly to due to Jared most of his minutes when Crash was out.
That concludes this look at apples to apples comparisons - sure, it is more like red delicious compared to granny smith, but at least it is an improvement. It should help serve as a reminder to think about all of the possible causes associated with a number - though I am still glad the Bobcats let McInnis go.
Click here to comment/view comments for this post.
| David A. Says: | Good work. Here's my trouble with this: positions. Even when McMinus was nominally the point, I'd be willing to bet that a plurality of the offensive plays actually started with another player, especially when Felton was on the floor, while on defense he pretty much exclusively covered ones. It's not particularly fair to Carroll's defense to compare his coverage of Kobe and other twos while McMinus only covered non-scorers. As for Emeka, it hurts to see that table, because his reputation is built on defense. However, position plays a role in my evaluation of his play, too. Determining he's a natural four means that you'd play Mohammed a lot more and gear your team to a specific kind of game. Determining that he's a five means building a different type of team. The Bobcats have never committed to an identity, which has to hinder their personnel decision process. |
||
| Mon Jul 28, 2008 11:17 pm |
| Brett Says: | McInnis "value" was almost entirely on the defensive end of the court. In fact, I was at the Lakers game here in Charlotte, and Touche had coverage of Kobe for a chunk of it. In general, you are correct, in that all of my numbers are based off what is gathered from a play by play, which does not detail actual positions and defensive assignments. As for Mek and Nazr - I only put so much stock in these numbers. I think the overall sample of the season, where the Cats are better with Emeka on the floor than off, and vice versa for Nazr say enough. It is interesting about this particular subset, and I suppose a deeper investigation of the specific lineups might provide some clues - though would be a separate post at some as yet undetermined time. |
||
| Fri Aug 01, 2008 7:32 pm |
