2016 predictions: I will end up being underconfident on my 2016 predictions.

Created by EloiseRosen on 2016-01-01; known on 2016-12-31; judged right by EloiseRosen on 2016-12-31.

  • EloiseRosen estimated 50% on 2016-01-01
  • Tapetum-Lucidum said “Oh, come on, this is easily a 60%!on 2016-01-01
  • JoshuaZ said “Is this measured by being underconfident in every 10 percent segment, or most of them, or what? on 2016-01-01
  • EloiseRosen said “Measured as having more underconfident predictions than overconfident ones.on 2016-01-01
  • EloiseRosen said “I like this prediction because it seems like I have to put 50%. If I was expecting some other probability, then why wouldn’t I just adjust my predictions to account for the expected over/under confidence?on 2016-01-01
  • EloiseRosen said “(The set I’ll be looking at to judge this is just the ones I submitted at the start of the year, that are all labeled “2016 predictions” at the start, to make this easier to judge.)on 2016-01-01
  • JoshuaZ said “And you intend to measure under/over confidence by what metric? on 2016-01-01
  • EloiseRosen said “Sorry, meant to say “measured as having more underconfident segments than overconfident ones.” As in look at each 10% bucket, and count it as an underconfident bucket if my accuracy is higher than the confidence level of that bucket.on 2016-01-01
  • EloiseRosen said “I suppose if I had 50 underconfident predictions in 1 bucket, and 1 overconfident one in the other buckets, I would be considered overconfident by my method, which seems silly. Unlikely to have case like that tho, so probs not worth weighting.on 2016-01-01
  • JoshuaZ estimated 47% on 2016-01-01
  • splorridge estimated 45% on 2016-01-02
  • EloiseRosen   judged this prediction right on 2016-12-31.

Please log in to respond to or judge prediction