Predict CTR

Clearer Thinking Regrants: Forecasting Tournament

Clearer Thinking Regrants is a program aiming to provide grant funding to ~20 altruistic projects. Starting from over 630 applications, the Clearer Thinking team has picked out 39 finalists. Now, we're asking forecasters on Manifold to weigh in on these applications, by predicting which we'll ultimately decide to fund. Show off your forecasting skills, and win a part of the $10,000 USD prize pool!

Tournament Rules

The intent of this tournament is to reward individuals who are best at predicting which grant applications will receive any amount of funding from the Clearer Thinking project:

• The final decisions for funding will be made by Spencer Greenberg; see the below section for how these applications will be evaluated

• The plan is to distribute funding to ~20 of the 39 applications.

• Grants will be a minimum of $10,000 per project, up to a conceivable maximum of $500,000.

1. USD prizes will be awarded to the top 20 traders based on total profit in this group of markets.

   • 1st place: $3000

   • 2nd: $2000

   • 3rd: $1000

   • 4th: $750

   • 5th: $500

   • 6th-10th: $250

   • 11th-20th: $150

2. Only 1 account per person can trade in this market

3. Don't collude with other people to inflate your or their profits

   • E.g. Don't transfer money to your friends account via betting a market artificially high/low and having them correct it

4. All forms of trading are allowed (including API and limit orders)

5. Manifold and Clearer Thinking reserve the right to modify rules, exclude participants, and other changes necessary to abide by the spirit of this tournament

About the grant evaluation process

• In round 1 of this selection process, more than 630 projects were each scored independently by two reviewers (who didn’t know the applicant personally) on four criteria:

  • Reviewers were asked to give four ratings, each on a scale from 1 to 10, on four criteria (described below).

  • The four criteria were as follows (these are copied directly from the spreadsheet we used - these descriptions were visible to us at all times when reviewing):

    • VALUE TO THE WORLD IF IT SUCCEEDS: if this project was successful (at its CONCRETE goals), how much value would you expect it to add to the world?

    • CONFIDENCE IN THE SPECIFIC PLAN: how good do you think the team's plan is? (relative to how confident we could be in any plan with this little information)

    • CONFIDENCE IN THE TEAM'S ABILITY TO EXECUTE THE PLAN: how good do you think this team will be at implementing this plan? (relative to how confident we could be in any team with this little information)

    • RELEVANCE: how aligned is the goal of this project to the kinds of ideas that FTX wants to fund? See their areas of interest here

  • The scores were combined by first averaging the two conference scores (to derive an "Average confidence" score), then calculating the geometric mean of "Value," "Average confidence," and "Relevance" (i.e., the cube root of the product of the three scores).

  • Each reviewer's score was converted to a z-score using that reviewer's mean and standard deviation, so that each score became a measure of how well a particular project had done according to that reviewer, relative to the other projects rated by that reviewer. The z-scores were then averaged together for each pair of reviews to produce a final score for each project.

  • The top 58 projects according to these averaged scores were selected for progression to the next round.

  • Note: there were many strong projects that did not make it into the next round.

• In round 2, the top 58 projects from the last round were voted on by a group of reviewers. Votes were cast in such a way that we did not know how others were voting until after we had cast our own votes.

  • Based on round 2 voting, Clearer Thinking narrowed the shortlist down to 42 projects.

  • Within the 42 projects, there were five that had been submitted by just two people. One person had had three applications successfully make it into our shortlist, and another person had had two applications successfully make it into our shortlist. For both people, we asked them to choose just one of their shortlisted projects for consideration in the next round. This left a total of 39 applications to be considered in round 3.

• In round 3 (the current round), the 39 applicants have the choice to opt-in to being listed on Manifold or not.

  • Applicants who opt-in to having their project listed on Manifold still delete any information that they do not want to share publicly. We ask all applicants to delete the references section, but please note that all applicants have to supply five references, on the understanding that we will select a random number of them.

About Clearer Thinking

ClearerThinking.org conducts psychology research and creates free, interactive tools aimed at helping you make better decisions, improve your critical thinking, and better achieve your goals.