NO-KG-AI wrote:Wow, KG's rebounding is shocking. His scoring isn't so bad that it makes a huge difference, he really doesn't drop off much from his regular play, but I'm shocked at his rebounding.
Small sample size, but still, that one is weird, because even when he's passing the ball, he's rebounding fiercely.
I just posted this on the actual blog, but I'll put it here too.
Again, the sample size is just way too small to make even any inferences off of. For example, in 2004 KG had only 4 "crunchtime" rebounds in 22 clutch minutes as defined by 82games. But out of the 7 close games this data came from, in 2 of them KG had a total of 6 more boards in the last 5 minutes of those close games that didn't count as "clutch" because the lead had gone up to 6 or 7 points". One of those games was his "legendary" game 7 against the Kings that came down to a last-second shot for OT...he grabbed 4 late boards, but only 1 of them counted as "clutch". If you actually count those 6 boards and those minutes in his clutch total, all of a sudden KG averages 12.9 rebounds/36 minutes in 2004 crunchtime instead of 6.5 and we're not having this conversation.
The whole reason we can use stats as somewhat reflective/predictive is because the law of large numbers suggests that noise like this will drop over a large enough sample. In a 2000 minute sample you can be fairly confident that the flukes will tend to work themselves out. In a 20 minute sample like 2004 Garnett's crunchtime, the noise can literally be bigger than the signal. You can't just caveat that as "small sampled"...it makes any conclusion meaningless.
