FREE hit counter and Internet traffic statistics from

Thursday, October 11, 2007

Chatting and Probabilities

Just a quick reminder that I'll be chatting on Baseball Prospectus at 11am mountain time tomorrow morning. As always you can submit questions and topics to discuss ahead of time...

Also, my column this week on BP took another look at the atmosphereic effects on pitched balls using PITCHf/x as well as a short rant on probability versus predictions and models versus reality.

In the latter vein, I ran my simulator for the National League Championship series with the following results:

  • D'Backs win 54.6%, Rockies with 45.4%

  • D'Backs sweep 5.9%, Rockies sweep 5.0%

  • Series goes seven games 33% of the time with the D'Backs winning those contests 61% of the time

  • Series goes six games 31% of the time with the D'Backs winning those 60% of the time

  • The D'Backs do well in the simulator because of their excellent home winning percentage (.617) and the fact they have home field advantage while neither team did well on the road (.491 for the Rockies and .494 for the Snakes). But echoing today's column probabilities like this are not prediction and any model won't factor in all the variables that play into a result in the physical world. In this case the model is simple and takes into account only home and road record along with the series format. It could of course include pitching matchups, health, and head to head matchups, all of which would improve the model. That said, even with a reasonable model seemingly improbable events occur all the time and so don't be surprised if something "improbable" occurs and instead just enjoy the ride.

    No comments: