Saturday, April 2, 2011

The Regend and the Ninja: A Response (part II)

New readers, please check out the basics before leaving any comments.

On Saturday, Tom Liston of Raptors Republic - TrueHoop's Raptors' affiliate - wrote a column about Reggie Evans and James Johnson. In his column, Tom mentions that he has some problems with Wins Produced and my analysis of Evans and Johnson in the past. Specifically, Tom's main issues seem to be:
  • Evans' WP48 is near the top of the league (currently at 0.356), but Reggie doesn't seem to belong with the other players (Kevin Love, Dwight Howard, Chris Paul, LeBron James, Dwayne Wade) at the top.
  • If Wins Produced doesn't overvalue rebounds, and the only thing Evans' does well is rebound, why is Evans' WP48 so high?
  • James Johnson is playing well in Toronto - "he’s pretty much exactly the same player" as he was during his rookie season. So why did I say that he probably wouldn't be a good player when Toronto traded for him?
  • Wins Produced is one measure of player productivity among several, and one should "never rely on one metric to draw conclusions".

The other day I responded to these first two points. Today I'm going to tackle the last two.

Why is James Johnson better than I said he'd be?

To answer this, first we have to know how well he played before he was traded to Toronto. Even though I don't see any point in using the non-WP metrics, here's a table that uses Win Score, WP, Win Shares/48, PER, and Adj+/- to examine Johnson's productivity over his career to date:

So did James Johnson play well before he got to Toronto? No. According to WP and Win Score, Johnson was was barely positive in 2009-10 if we consider him a pure small forward - which he wasn't while he played in Chicago. PER, WS/48 and Adj+/- all say that he was below-average. His time in Chicago this season was also clearly terrible - according to WP and Win Score, Johnson actually cost his team wins. PER, WS/48, and Adj+/- all tell the same story. So it's pretty well established: according to every major productivity metric (and the eyes over in Chicago), James Johnson did not play well before he arrived in Toronto.

However, after he arrived in Toronto, Johnson has played well. Or has he? A good first step would be to take a look at his specific statistics:

He has, and we can also see why he was so terrible in Chicago. For one, his foul rate was disgusting, and that has almost dropped down to the average rate for SFs. His turnovers were not very good, but have improved in Toronto. His blocks have always been excellent, so not much has changed there. Compared to last season, Johnson has improved with respect to rebounding, assists, and steals (in both Toronto and Chicago). While his field goal shooting has slightly improved in Toronto, once we take into account free throw attempts and percentage, Johnson is shooting about as well as he did during his rookie season. We'll have to assume that his poor shooting in Chicago this year was a result of a small sample size. Put it all together, and Johnson is playing at an above average rate in Toronto - somewhere around a WP48 of 0.160.

So, short answer as to why he's playing better in Toronto: less fouls, more extra possessions, and more assists.

But wait - let's take a look at his first table again. For your convenience, I'll repost it below:

According to Win Shares per 48 minutes - the measure used over at - Johnson hasn't improved, even though he's fouling less, adding more possessions, and getting more assists. PER, on the other hand, says that Johnson has improved, but is still below average...even though he's clearly above average in most categories. And Adjusted +/-...well, ADJ+/- says that Johnson is playing his best basketball right now, but it also says that his best season in Chicago was this season - not last season - which is pretty odd. Perhaps that's because ADJ+/- is, as Andres has said (and Arturo and Alex Konkel have shown), about as good as Tarot readings or Horoscopes. On the other hand, Win Score and Wins Produced seem to follow the numbers (odd, innit?).

Borat meme
At least I have the right actor
So what does that say about the various methods of measuring basketball productivity?
  • WS/48 is suspect because it can't even tell that Johnson has played better in Toronto.
  • ADJ+/- is suspect because it thinks that Johnson's best season in Chicago was this season.
  • PER is suspect because it overvalues taking shots and still thinks Johnson is a below-average player.
And that is why we should not consider all the different metrics - not all of them are useful. ADJ+/- has huge errors and needs multiple years to find out anything (and even that has little meaning). PER is obsessed with players who take lots of shots and undervalues those who make contributions without making any mistakes. Win Shares is better than ADJ+/- and PER, but basically it is a watered down Wins Produced. None of these metrics improve on Wins Produced, nor do they add any additional information that Wins Produced doesn't consider.

So, to recap:
  • James Johnson was bad in Chicago.
  • Johnson is playing much better in Toronto.
  • ADJ+/-, PER, and WS/48 are all suspect and don't help to inform us beyond Wins Produced (they are actually less informative).

Answer the Question!

But wait! I still haven't answered an important question: I said that James Johnson wasn't going to be good in Toronto, but he has been. Why was I wrong?

It all has to do with my method of predicting future performance and the nature of young NBA players. When I said that Johnson probably wasn't going to be productive in Toronto, my reasoning was this: he hadn't been good and he's relatively old. Most players are who they are once they hit 25, and Johnson had just turned 24. But young players can always see unpredictable peaks or valleys in their productivity. Johnson was a young player who improved significantly in a new situation, and such things are not only unpredictable now, but are likely to remain unpredictable forever. How was I (or anyone else) to know that Johnson would improve his rebounding by more than a rebound per 48 minutes? Or reduce his foul rate from 8.0 per 48 minutes to 4.8 per 48 minutes? The only remotely reliable way we can predict the future is to extrapolate trends from the past. Young players have fewer data points and fluctuate more than veteran players. Similarly, veteran players who suffer (or recover) from significant injuries are unpredictable. How were we supposed to know that Udonis Haslem would miss the entire season due to injury? How were we supposed to know that Elton Brand would finally - after three years - start returning to his old standard of play? How were we to know that Wilson Chandler would leave David Lee such a wonderful parting gift?

My inability to predict the future says nothing about the validity of Wins Produced. Wins Produced is a descriptive number, not a predictive number; it can only describe performance that has already happened and provide clues towards future performance. I would also note that the same thing is true for ADJ+/-, PER, and WS/48.

Even though I've been quite impressed with Johnson's play so far, I'm still not totally okay with the trade. Why? Johnson, although better than I expected, still isn't much better than Julian Wright. Who's Julian Wright? He's that guy who has been sitting on the bench (occasionally of his own accord) ever since Johnson arrived - and who is unlikely to return next season. Here are Wright's specific stats:

Wright's had a better career than Johnson, and even though he's played two more seasons than Johnson, Wright's actually younger by a couple of months. Both are similar players - both can't shoot, although Johnson seems to be better in that department overall - and both can play SF and PF. Both are athletic and good at blocking shots, although Johnson is better there as well. Wright is better at steals, fouls, turnovers, and Johnson has been better at rebounding and assists this season. Overall, Johnson has been better while in Toronto, but only time will tell if this will continue. In any case, I don't feel that the possibly slight upgrade from Wright to Johnson was worth a late first round draft pick. As I've pointed out many times before, draft picks are very valuable. There is almost always a very good player waiting for you in the late first round, and every additional draft pick you have is another chance at hitting it big. The Raptors could have retained the draft pick and re-signed Wright on the cheap during the off-season - at minimal cost. That would've given the Raps another chance at drafting a good player...although the Raptors seem to have trouble identifying talent in the draft.

If Johnson continues to improve - and there are no Rodmans, Ginobilis, Boozers, or Fields available to Chicago with that late first round pick - I will admit that I was wrong. Luckily, it will only take a season or two to find out!

 - Devin


  1. Devin,
    Win Shares is a "worse" metric than WP48 because it is more complicated and less predictable. However it is still a pretty good metric. The one thing to consider is that it has a much larger team influence than Wins Produced. So ironically a player on a bad team that Win Shares says is bad is probably better than they appear and a player on a good team that Win Shares says is good is probably worse than they appear.

    Also I give very little credit to the Raptors trade if they just "knew" it would work out as Barbosa and Kleiza have fizzled and earlier they tried to trade Calderon. In short, I don't trust their eyes at all.

  2. Yeah, I know about WS/48. Johnson was worse in Chicago, but his team was better, and better in Toronto, but his team was worse. Strangely, the WS/48 come out to be the same.

    To play devil's advocate...they were trying to acquire Tyson Chandler in the failed Calderon deal. And Boris Diaw, but that's another story....

  3. Nerd numbers - understand why the Barbosa trade happened. You may have missed economics class I believe. Kleiza was a poor signing.

    Watching the Orlando - Toronto game currently and it clearly points out the Evans/WP issue. Yes, he's effective as he doesn't take many shots. But he's had the ball 5 times and he drained the shot clock because he cannot pass nor shoot effective. Yet WP loves that he makes others force a last second shot and doesn't make a "mistake".

    See the problem with his MVP like WP48 numbers?