modeling calibrated estimations aggregating predictive forecasts formulating intelligent wisdom mapping probable estimations aggregating accurate estimations mapping the future forecasting quantitative insights computing precise predictions generating contingent wisdom computing accurate contingencies delivering calibrated understanding delivering probable wisdom forecasting probable futures exploring definitive futures


Cryptocurrency competition results!

By Anthony on Jan. 8, 2018, 4:10 p.m. GMT

After some internal deliberating and data crunching, we're finally ready to announce the results of the 2017 Cryptocurrency Prediction Competition. First we'll say a few words on the reason for delay, then the results.

As those following the competition know, we got stalled on four questions that were based on the highest or total amount amount of initial coin offerings (ICOs) in October and November. It turns out that the data for ICOs is terribly incomplete and unreliable, with numbers dramatically varying from one source to another. On top of this, it seems fairly clear that there is an order-of-magnitude uncertainty in the ICO funds actually raised by Paragon. In the end, we decided to resolve these questions per their letter, keeping with Metaculus's general policy. However, in recognition that these questions could be considered flawed because based on a misleading resource, we've decided to also issue prizes to the two predictors who would have won had we resolved these four questions as ambiguous.

So, without further ado, the winners are:

  • First place prize of 1 ETH jzima with a total score of 748

  • Second place prize of 1 LTC: Barbarossa with a total score of 678

  • Alternative first prize of 1 LTC: AndrewMcKnight with a total score of 716 (but four less questions included)

  • Alternative second prize of 1 LTC: ElliotOlds with a total score of 623

We'll be in touch with these individuals to deliver coins to their wallets, which are worth about 3x what they were at question launch. Of course, bragging rights are even bigger!

Now without taking anything away from the accomplishments of these four, it should be said that the cryptocurrency question results were somewhat out-of-line with Metaculus as a whole: performance was somewhat worse, with a bloodbath on some questions. So some generally excellent predictors got creamed by being too high-confidence. We'll post an in-depth analysis of the results, trends, and some lessons learned.

We hope the contest was fun and interesting, and we look forward to running more – ideas for topics, or for tweaks in the way they are run, are welcome in comments.