Sunday, May 15, 2005

This Is Why Statheads Drive Me Crazy

I love stats. I use stats. They're fun to tinker with. They're fun to analyze with.

But, at a certain point, you need to drop the abacus and look at the game to see if the results your numbers are giving you are realistic.

The problem with most amateur statisticians is that they come to believe that their models are infallible. And anyone who's ever worked with modeling (the theoretical kind, not the Heidi Klum kind) can tell you that they have their limits and that you need to constantly be aware of any sorts of biases or misjugements you made when setting up that model.

And, even if your model is close to perfect, there are always going to be things that fall through the cracks.

For an example of this, see this BTF thread about our dearly beloved former centerfielder.

Pay attention to comment fifteen, especially. Read it. Then reread the first clause of the first sentence.

I'm just sayin'.

3 Comments:

  • but the regression in a defensive progression algorithm adjusts for its inherent inaccuracy.

    As Shaq would say, "Excuse me?"

    By Blogger Basil, at 5/15/2005 6:33 PM  

  • Hey, Bodes found a nut!

    By Blogger Basil, at 5/15/2005 7:25 PM  

  • Excellent post. I would have penned a clever explanation of why I agree, but Aaron Haspel has said everything I would have said, and much better, and that's why hyperlinks were invented:

    http://www.godofthemachine.com/archives/00000410.html

    Aaron's blog is worth rummaging through, if you have a spare moment or two.

    By Blogger DM, at 5/15/2005 11:16 PM  

Post a Comment

<< Home