What will Jordan & Cho do?

[Crow]

Continue being really bad for 2-3 more years or try a big splash in free agency or spend big on a pretty good coach? I assume they will do the former. Might kill fan support of the franchise off in Charlotte for good though.

Maybe there is a middle ground. Wait a year then hire D'Antoni to bring the offense back towards average, then maybe start spending later and perhaps eventually get a different coach, maybe a promising younger one? That would be one plan.

Not good with draft picks. Walker and Biyombo were both bad picks last season.

[YaoPau]

The Bobcats are fine. I'd rather be in their position than the Bucks'/Rockets'/Magic's/Blazers'/Suns', probably a handful of other teams too.

Kemba started distributing more in the second half. I don't think he's a particularly high upside prospect, but he can be a serviceable starter for awhile. They have two nice talents inside with Biyombo/Mullens. They'll have a top-4 pick this year, and likely a top-10 pick next year. If I'm a GM, I either want to be top-4 in my conference or last in my conference, and the Bobcats are in one of those spots. No reason to rush out of it imo.

[Mike G]

Walker and Biyombo were both bad picks last season...

They were the 9 and the 7 pick, respectively.
I have Kemba 2nd in the league in rookie eWins, and close to that in eW/Min (among those >1000 min). Biyombo 17th in eW, and 14th per min., among >1000 min; he's just 19, and he made significant strides in the season.
Neither player was likely in an ideal development situation, and both did OK.

[bbstats]

I have Kemba as the 11th-worst player in the league, with -3.98 Wins Above Average...although I have a relatively large team adjustment for Charlotte.
His raw stats have him at a much more respectable 83rd-worst in the NBA (-0.95).

But honestly, RAPM likes Walker. I have him at: -3.1 (boxscore) and 0 (Jerry's NPI 2012 RAPM), which agrees only 25% in ranking each player-season.

[Mike G]

... the 11th-worst player in the league, with -3.98 Wins Above Average....

And this is why we use replacement level to make absolute statements.

[J.E.]

But honestly, RAPM likes Walker.

Hm? He's bottom 10 in both versions

[bbstats]

... the 11th-worst player in the league, with -3.98 Wins Above Average....

And this is why we use replacement level to make absolute statements.

Don't follow the logic. Looks like just a disagreement with my statement. EDIT: Kemba harms the Bobcats' point margin strongly and are therefore is very damaging to his team's win total. Doesn't matter if he's "better than replacement"...I'm just talking about versus a 0.500 team, one with an efficiency margin of 0. Seems pretty straightforward to me.

But honestly, RAPM likes Walker.

Hm? He's bottom 10 in both versions

Not sure why I had him at 0! Woops.

[Crow]

On the lowest winning percentage team in NBA history and playing mostly with other starters, Biyombo and Walker had the 2nd and 4th worst team offensive ratings while on the court and only 2 of those players had worse team defensive ratings while on the court than Biyombo.

I really doubt Biyombo will ever be an average starter and I think it is more likely than Walker will stay below average for his position than exceed it.

[bchaikin]

I really doubt Biyombo will ever be an average starter...

biyombo is (supposedly) just 19 years of age. he was a 4.7% BS shot blocker (4.7 blocks per 100 opposing team FGAs) and a 12.1 reb/48min rebounder. in combination this is quite good for a player so young, and numbers comparable at a similar age (19/20) to those of players like tyson chandler and serge ibaka, yet with a lower rate of fouls committed...

...and I think it is more likely than Walker will stay below average for his position than exceed it.

walker is just 21 years of age. his stats are similar to (touches/min on offense, what they did per touch) but actually better (lower turnovers per touch) than those of both terrell brandon and kenny anderson at a similar age...

[Crow]

Kemba Walker had the worst eFG% of any player in the league who played 1500+ minutes and had the 4th worst offensive rating. Offense is supposed to be his strength.

Biyombo was in the bottom 20% on eFG% for guys 6-9+ who played 1200+ minutes and had the 2nd worst offensive rating in that group. That isn't his strength obviously but that will probably always hurt his overall impact.

Neither showed much improvement as the season went on.

Maybe I am being too harsh and hasty about these young guys in a bad place, but my impression is that they really had very bad seasons for high draft picks or anyone. Both in the bottom dozen in the league on non-prior informed overall RAPM and both estimated at least -1 on both sides of the court.

They could potentially be ok or better on impact if they were on a good team where they were role players- Walker, the 3rd or 4th option on offense; and Biyombo, the 2nd big off the bench.

Henderson is the relatively bright spot on the team.

[Mike G]

Kemba Walker had the worst eFG% of any player in the league who played 1500+ minutes

One of 2 Bobcats to top 1500 minutes, his TS% (.464) was lower than that of Prince (.471), Turner, World Peace, Jamison, Rondo (.483)... lowest among the top 100 minutes.

Of course, on a better team, he wouldn't have gotten those minutes or shots.
Over here -- http://basketballvalue.com/teamplayers. ... 2&team=CHA

... his team was -17.2 pts/100 while he was off the court but -15.7 while he was on, for a net difference of +1.5
His adjusted on/off, though, is a team worst -7.8. Implying his courtmates + opponents averaged some 9.3 pts/100 poss in his favor?
Somehow he and Biyombo both have 2-year adjusted rates shown.
And the numbers don't seem to be for quite the whole season.

[mystic]

Somehow he and Biyombo both have 2-year adjusted rates shown.
And the numbers don't seem to be for quite the whole season.

2yr APM on basketballvalue.com means that a dataset is used which consists of the current season and the previous season games. The playoff games are weighted twice as much. So, for this season the 2yr APM has all games from the regular season 2010/11, playoffs 2011 and regular season 2011/12 included.

Saying that Walker and Biyombo were "bad picks" is really premature. As bchaikin pointed out, both are really young and they have a lot of time ahead of them before reaching their respective peak level. Additional to that should the expecations for 7th and 9th picks be lower anyway. Compare the Bobcats to the 2007/08 Sonics. Durant was the 2nd and Green the 5th pick in the 2007 draft. The Sonics were close to be the worst offensive team and were a bad defensive team that season. And all that while having also more expierenced players on their roster.

The Bobcats should evaluate their options and have to look also into the financial situation. They can't just suck for the next couple of years, they have to offer some success in order to make money. But success coupled with young players seems to be a good way to attrack people. It seems to be a good mixture. So, it depends on who they are able to draft, but I really think they should try to get some high impact role players additional to that via free agency, while keeping their options open for a run at James Harden in 2013. Having a bigger SG next to Kemba Walker seems crucial. I also think that a frontcourt with Biyombo and Davis (assuming they get the 1st pick) can work. I like Gerald Henderson on the wing, he is a keeper. I'm not convinced that DJ Augustin can help in the long run. Maybe Maggette (expiring) + Augustin can net something in a trade. If the Bobcats aren't winning the lottery, they may as well go for Jason Thompson (Kings). He is restricted FA, but maybe paying him a bit more can be beneficial in the long run. A scoring wing would be a good idea. Or Chris Kaman, a pretty good lowpost scorer. As long as they still keep a max spot for 2013 open, they should be fine with adding veterans, who can help them. As I suggested earlier, throwing the max at James Harden in 2013 should be a priority.

From the financial standpoint, it may also be helpful to get the name "Hornets" back. The new owner of the Hornets seems to be willing to change the name, Jordan should push in that direction as well. I guess more fans can be attracked by having the name Charlotte Hornets back on the map.

[Mike G]

...the 2yr APM has all games from the regular season 2010/11, playoffs 2011 and regular season 2011/12 included.

Understood. So, how do rookies have 2-yr rates which are substantially different from 1-yr?
Are all those minutes (and the teammate and opponent 2-yr rates) calculated as part of the time said rookies were "off court"?

Biyombo's 2yr is 3.5 pts worse than his 1yr. Walker's is 1.9 pts better.
In Cleveland, Kyrie Irving's 2-yr is 4.1 pts better than his 1-yr, while TThompson's is 1.4 worse.
Brandon Knight's is 6.1 better.

[mystic]

Understood. So, how do rookies have 2-yr rates which are substantially different from 1-yr?
Are all those minutes (and the teammate and opponent 2-yr rates) calculated as part of the time said rookies were "off court"?

No, that is not how that works. Each time segment has the informations about the players on the court and the respective scoring margin. That would be one equation. Over a given season we have about 30k of such equation. The regression now finds the set of coefficient for which the error becomes the smallest (basically, the best approximation of the scoring margins for any given time segment). That means that a coefficient for a specific player is also effected by the coefficients of the other players on the court. Due to the bigger sample the teammates and opponents of the rookies can just have different coefficients and that will change the coefficients for the rookies as well in order to keep the error smaller.

With normal APM we have the problem of overfitting. To compensate that we can increase the sample. That should increase the predictive power and thus lower the error in an out-of-sample test. While the one year APM values might produce a substantial smaller error within that sample than the 2yr value, the 2yr value will likely be the better predictor.

A better way to stabilize the values is ridge regression. We have a ill-posed problem and in order to decrease the error in out of sample tests (to make it a better predictive value) we can include a bias (lambda) which restricts the range in which a coefficient can change. That prevents overfitting issues.

Anyway, the point is that the regression is always directly based on the minutes (possessions) while the player was on the court, indirectly it covers the minutes the player was not on the court. For rookies the latter includes all minutes from last season (2010/11) obviously. We could argue that the players aren't exactly the same in both seasons (let us use Lamar Odom as an extreme example here). Thus, without a proper aging or development curve, we might as well over- or underrate certain players with increased sample size. But in the end the 2yr APM is to prefer over the 1yr APM.

For Biyombo it seems as if the teammates he was on the floor with were in average better last season than this season, thus his 2yr APM will be seen as worse than the 1yr APM. For Walker it seems to be the opposite. The RAPM values (prior informed and non-prior informed) are suggesting something similar. For Jerry's prior informed values we have to keep in mind that rookies starting with a negative prior (-1.3, if I'm not mistaken). Both, Biyombo and Walker had the same prior (again, if I'm not mistaken), but Biyombo has a worse prior informed value than Walker, while Walker finishes worse than Biyombo in non-prior informed. That indicates as well that the Bobcats' players Walker was on the court with were worse in average last season than those Biyombo was on the court with.

[Crow]

I only use pure traditional APM to look at possible trend. For mainline use I prefer RAPM of some variety.

I am premature in judging Biyombo and Walker by fair and reasonably complete standards for making such a call. But I am intentionally trying to be early to make this call, in contrast to the hype they got as draft picks. Being early on an evaluation of a guy, if you are right, can have lots more value than being in step with conventional evaluation of a guy. And it opens up some discussion which is not very easy or running at a very high pace around here. I could be wrong about one or both of these guys but I don't think I will turn out to be. Check back on this next spring and will see how it looks. Maybe I should have waited til then but I didn't for better or worse.



Bobcats with Walker instead of Augustin but with the same other four were 2 points worse. Augustin is a -3 RAPM player and Walker was worse in the same situation.

Biyombo was worse than Mullens and White with the same 4 other guys and -28 compared to Diaw. Biyombo was not only the single guy Diaw was better than on team results, he was the other guy he wasn't greatly worse than.

It was debatable whether Mullens and White were draft mistakes of Presti or mistakes not to use more. With the additional data in Charlotte it appears that they were draft mistakes or draft non-wins. Their RAPM in Charlotte is still replacement level, not good for 1st round picks given good opportunity. If one views them as routine draft gambles and low 1st rounders, then at least the decision that they were non-wins was fairly quick and they got used as very marginal trading chips.