Every stat is suffering from that, yes. But indirect stats (like +/- and similar stats that go by team performance with and without the individual) require larger sample sizes to yield meaningful results than direct stats (like PER or other stats that work directly from the individual's numbers). Of course, one could argue that once you have that larger sample size, something like Adjusted +/- or win% is more powerful...but you need that sample size. Taking this to an extreme, on a single play basis, direct stats tell you something while +/- tells you literally nothing. If Blake makes a steal and hits a three-pointer on one play, his shooting, steal and scoring numbers reflect it. If Blake makes a great pass to Bayless, but Bayless misses the shot and then on the ensuring rebound and transition, Cunningham gives up an easy basket, Blake's +/- or win% for that segment will be negative. For quite a long time, direct stats have the decisive edge on indirect stats in terms of non-noise content. I've read statisticians say that you need several seasons worth of data for +/- types of statistics to provide meaningful conclusions.
In the blizzard of numbers on 82games.com's profile of Blake you found one that is amazing. There are others that are damning. What to believe? We don't have any statistics in basketball that really hold up in all cases. That's a sad fact. The team does at times play much better with Blake at the helm. Other times it plays it's worst. Without Synergy sports giving us a feed of every single play Blake was involved in it's pretty hard to tell what his overall impact is. I wish I had a couple hundred thousand dollars to get a subscription to that service. It would be amazing.
This seems to be the common refrain pertaining to any statistic that makes BLANKY look like anything other than a horrendous player. It's the board version of a poster closing their eyes, plugging their ears, and humming to make it all go away.
How about we point to the specific damning ones and discuss them? I have said, pretty clearly in this thread, that I do not know how to explain this one, but it is damn unusual because it's correlation to his individual stats is so out of whack. What other stats are there that jump out at you? The clutch stats are bad, individually, but actually darn good in the context of the team members. Again, if you have interesting ones you think we should look at - I would be interested in looking at them. Is it possible that Blake's amazingly high win% is a coincidence? Sure. It is a freaking small sample. But, it is interesting. Minstrel has an interesting point about looking at multiple-years of +/- like stats - which win% is a less volatile one of - so I looked at his win% last year - he was 2nd highest win% last year with LMA - at 65%, behind Roy on the team) - Went back to 2007/2008 - 3rd best in the team, 2nd best of those that played significant minutes (James Jones, who was 2nd, played only 32% of the minutes) - behind Roy and over LMA. What are we doing with it now? Honestly, I do not know. But we have more than 2 years of data now that paint Blake as one of our best win% guys. Why? I don't know. I am not a coach. His individual stats sure don't explain it. Maybe it is the combination of low-risk play, ability to handle the ball and space the floor? Something that brings up the best in better players around him? I don't know. I am just looking at the data and wondering. I have been working in the analysis industry for about 20 years now, while I am not an analyst - I have developed many tools that analysts work with - so I suspect I have a bit of experience looking at data and noticing patterns. Making the jump from seeing the patterns to having reliable interpretation of them usually requires someone with better understanding of the material subject - which I am not. But, the pattern seems to be there. Blake has consistently been one of our best win% guys on this team. Now, let's look at the 2006-2007 Denver Nuggets. You want to guess who their 2 guys in win% are? 1. Andre Miller at 60% 2. Steve Blake at 59.2% (They did not play together, if memory serves, Miller was traded for Iverson and they traded for Blake after that). There is something that Blake does, that make him a high win% guy when playing next to good players. What it is? You tell me. So in the context of multiple years, it does not look that bad and strange that Blake is a high win% guy, sure his win% this year is just ridiculously high - which is why it jumped at me when I looked at it - but Blake is consistently a high win% guy when he plays next to good players.
Well, that's the truth for every kind of stat - you take a lot of individual occurrences and you try to notice patterns and outliers. I am not buying this one bit. If someone is a gifted passer but is saddled with lousy shooters and finishers as team-mates, their "noise" taints the value of the direct assists to really judge his ability as passer. If someone is a great rebounder and scorer but plays god-damn awful defense - his individual "noise" is less reliable than his win% "volatile" stat (See Randolph, Zach). Well, we have multiple years of win% data for Blake next to good players (2+ years in Portland, half a year in Denver) - and the numbers are still much higher than his individual stats will lead you to expect. What now?
