At first I thought I was going to write a Ricky Rubio rant. Recently in a podcast, I called Rubio the rookie of the year. In three games since, He's shot 8-for-33. He's had 29 assists, but also 10 turnovers, and on Monday he really killed the Wolve's chances when all 5 of his turnovers came in the 4th quarter. If he had not hit a game-tying 3 in his 1-for-11 shooting night against the Clippers, we'd probably be crucifying the poor guy right now.
But instead I decided that the Unicorn needs a break. And anyway, he's not really the issue I am trying to address. The issue is small sample sizes. Basically, Rubio has played less than 20 games. This isn't really enough to draw meaningful conclusions. We can do some speculation, and some numbers are more promising than others. For instance, over a small number of games, three point field-goal percentage is not as reliable as 2-point field goal percentage because there are far fewer threes in a game then there are two-point field goal attempts, so the sample is smaller. Or some non-shooting metrics have a lot less variance than shooting metrics; the guys with more than 200 minutes who are in the top 10 in rebounding per-48 right now will probably all be near the top in rebounding at the end of the season, but more of the current field-goal percentage leaders won't be.
So, basically, we've got to be careful about the conclusions that we draw. This isn't a rant -- I am guilty as this (perhaps more so) as any blogger/journalist/ESPN personality. It goes back to the illusion of validity, and how we want to make sense of what we see, even if what we see is just the result of variance. Rubio might just be going through a slump. Or, perhaps the first 13 games he was shooting abnormally well and now he's regressing to the mean. Or, maybe he's neither as good a shot as he looked the first 10 games, nor as bad as he's looked the last 7. We'll find out later (much later).
And I could say the same thing about any rookie. That's why I tried to make my analysis of the rookies who need more minutes sound less like, "Hey, he's really good, play him," and more like "Hey, you have this other veteran player who's been really terrible over 5 NBA seasons, you might as well play the rookie who looks good and see if it's just a fluke, right?"
And it's obviously more than just rookies. I've written columns about how everyone thinks Love is even better this year than he was last year, but other than point totals (which is purely a factor of increased shooting), he is worse in almost every category:
|Average PF(2008 - 2011)||1247||.099||2.6||20.5||7.8||3.3||11.2||2.5||2.4||1.3||1.3||4.8|
(Note: I am aware that a "bad" season for Kevin Love is one than most power forwards would kill to have. My point is that through 17 games, Love has not been as good as he was last year). But then Monday he went off on 13-19 shooting (5-5 from 3) with 12 boards. So who knows? He could have another 30/30 game soon, and that will change the per-48 numbers, won't it?
So if all these sample sizes are meaningless, why bother? Well, two reasons. First, for the most part we are trying to evaluate player performance so far. So, it's an interesting story that, so far, Kawhi Leonard has been really good (and largely unnoticed). It's a surprising story that, so far, Tim Duncan has not (even after adjusting for lower playing time). It's also interesting that the Knicks aren't very good (so far). And it's interesting to look at why, because I like to debunk analysts that say "Blah blah chemistry blah blah" by saying things like "look, it's quite simple, so far, Amare Stoudemire is turning the ball over a shitton and Landry Fields is shooting about 20% less effectively than last year while grabbing half the rebounds and turning the ball over and fouling lots more. Oh, and Chauncey Billups is gone. Chemistry my butt."
The second reason is much more simple. If I waited for big sample sizes, I wouldn't be writing any articles until after the all-star break. Or every article would just be "Well, Player X looks decent, but whatever, you can't say shizznits from a sample size this small." And even though I have embraced using the word shizznits, this would get old really fast.