April 23, 2010

The Key to Thunderous Success

Filed under: Uncategorized — wwinston @ 1:03 pm

In 50 minutes Kristic is on against Bynum and Gasol, Thunder down 35 points.  When Ibaka is in and Kristic out  against Gasol and Bynum Thunder are +2 points in 12 minutes. They need to start Ibaka  (and maybe Collison) instead of Kristic (and maybe Green) and they could win series! Also Collison in agaist twin towers Thunder are +9 points in 14 minutes.

Durant on court against Kobe Thunder are down 4 points. But with Ibaka and Durant in against Kobe Thunder are +14 in 52 minutes.

12 Comments »

  1. I agree Ibaka should start, but are these three game samples really worth anything? Doesn’t the noise essentially render this data meaningless? A few lucky makes on bad percentage shots or a few misses on high percentage ones seems like it’d skew the data quite a bit.

    Comment by Sammy — April 23, 2010 @ 6:12 pm

  2. in past playoff series first few games have done a pretty good job of predicting the rest of series. It is clear that Lakers starting lineup crushes Thunder starting lineup so why not do something about it.

    Comment by wwinston — April 23, 2010 @ 6:57 pm

  3. That’s really surprising to me. I’d love to see the data on that. Not that I don’t believe you, but it just seems so counter-intuitive to me. It seems cases like James Harden are more the norm than not in the playoffs: he looked passive and awful the first two games, then remembered to be aggressive and was a key contributor to the win in game three. To me, basing decisions off three games worth of samples would be like Ichiro going hitless and dropping three flies in a week and using that as a basis for sticking him at DH in the 8 hole.

    Comment by Sammy — April 23, 2010 @ 7:08 pm

  4. Regarding the Thunder, though, I completely agree. This starting lineup has been mediocre all season, and its weaknesses happen to play right in to the Lakers’ strengths. Brooks has been stubborn about this lineup all season, though, so I doubt it will change. On the other hand, Krstic did only play 16 minutes last night, so hopefully that’ll start a trend.

    Comment by Sammy — April 23, 2010 @ 7:13 pm

  5. I feel like your whole website is based around talk about noise. What kind of research do you have to show the sample sizes you constantly employ are meaningful in any way?

    Comment by Tom — April 23, 2010 @ 7:28 pm

  6. sigma of how well a combo plays is 12 points per 48 minutes so you can d oa tet of statistical significance. WOuld you disregard the fact that the thnder get killed with Bynum and Gasol on the court?

    Comment by wwinston — April 23, 2010 @ 7:45 pm

  7. I don’t know if I follow. Are you saying that with a ~50 minute sample size of two players past performance the prediction on how they’ll perform over 48 minutes in the future has a normal approximation with 12 point standard deviation?

    Comment by Tom — April 24, 2010 @ 2:40 am

  8. I guess it just seems awfully fishy to me to show these kind of things are more than random noise. There’s probably about 7 lakers players that get lots of playtime giving you about 20 pairings of lakers and I’ll guess 5 thunder starter/backup pairs to compare the lakers duo to.

    That’s 100 matchups to analyze and come to some sort of conclusion with. You’re bound to find plenty of extreme +/- numbers in them especially if you are considering chunks as small as 12 minutes. Is there an article or book you or someone else has done on your methods to show that you aren’t simply cherry picking exciting and random results which such small sample sizes?

    Comment by Tom — April 24, 2010 @ 2:58 am

  9. yes

    Comment by wwinston — April 24, 2010 @ 8:28 am

  10. Every game Bynum Gasol kills Thunder. Would you wait till you lose 4 games to claim a p value>=.05. Clearly the probability is high that current way of dealing with those two does not work!!!!!!!!!!!!!!!!!!!!!!!!

    Comment by wwinston — April 24, 2010 @ 8:29 am

  11. the sigma is for 5 player combos

    Comment by wwinston — April 25, 2010 @ 9:57 pm

  12. You are close on this. They actually need to start Collison for Krstic. Statisically Collison is having a great series, even better than Ibaka. Plus Ibaka off the bench has worked well for the Thunder all season. Both players bring great energy. I can also see where you are going with Green, statistically Collison or Ibaka should be on the floor instead of him, however I think team Chemistry wins out on this one. Plus I believe Green is doing a good job of wearing the lakers out physically which is probably helping Ibaka and Collison’s statistics.

    Comment by Cory — April 27, 2010 @ 5:08 pm

RSS feed for comments on this post. TrackBack URL

Leave a comment

Powered by WordPress