Tuesday, October 13, 2015

Week 6 Results

First of all, both dashboards have been updated through week 6:

Results
Last week I was asked to share the results of how the model performed at predicting each game. I do a lot of looking at the results myself, trying to figure out where there are holes in the model and how to patch them, but I've not done a lot of posting about them for a few reasons: 1) I'm not sure the best way to present the data. 2) Looking at a single week's worth of games isn't super insightful.

That said, I'm willing to give it a try, so I made this table. As a start, I'm focusing on which team the model thought would win, and how much confidence it had in that team.

The table below shows each game, my model's predicted winner, the actual winner, and whether the model got it right or wrong. It also shows (and is sorted by) the model's confidence in the pick.

Interested to hear feedback. Is this interesting/valuable? Is the data presented in an understandable way?








4 comments:

  1. This is great! Is it my imagination, or is your model consistently more wrong on the Saturday evening games? At least when I looked back at week 2 out week 3 I remember noticing that.

    ReplyDelete
  2. This is great! Is it my imagination, or is your model consistently more wrong on the Saturday evening games? At least when I looked back at week 2 out week 3 I remember noticing that.

    ReplyDelete
    Replies
    1. Glad you like it. To your question, I'm not sure. I don't know that I've studied how effective my model is by game time. I do think there are specific teams it has trouble with (Baylor) though.

      Delete
    2. Glad you like it. To your question, I'm not sure. I don't know that I've studied how effective my model is by game time. I do think there are specific teams it has trouble with (Baylor) though.

      Delete