• voracitude@lemmy.world
    link
    fedilink
    arrow-up
    16
    ·
    1 month ago

    Fun fact: aggregated crowdsourced predictions for sports games have ~55-60% accuracy. That means if you get enough of them and play the odds you get, you’ll win more often than not; the challenge is getting enough people to participate. I knew a guy who started a company doing exactly that; we lost touch but they were doing alright last I heard from him.

    I dunno if that translates to other areas, but I’d sure be interested to find out. Valve certainly has the audience size to get enough participants, and that audience are from all walks of life so it’s a more accurate cross-section of the populace as well.

    … now that I put it like that, I’m a little jelly. That sounds like really interesting data.

    • Dave@lemmy.nz
      link
      fedilink
      arrow-up
      2
      ·
      1 month ago

      But sports games don’t pay even odds. You can’t just bet $10 on the favourite in every game, win 60% of the time, and come out ahead. You might only get paid $1.10 per dollar for a win. I’m not sure I understand how it’s helpful. Maybe only apply to games that pay enough?

      Wisdom of crowds is a thing, but that isn’t how this ex-Valve dude’s example was structured. You can’t reward someone in particular for getting it right if you are averaging the crowd.