I’ve read Superforecasting, but I find that actually applying the "10 commandments" is difficult in isolation. The feedback loops in the real world are too slow, and it’s too easy to skip post-mortems when no one is watching.
My goal for this year is to put in substantial work to become a superforecaster (or at least get much closer).
To do this, I am starting a dedicated online community for peer accountability and high-frequency practice. I’m looking for a small cohort of people who want to actually improve their forecasting skills.
The Plan:
Regular Meetups: We will hold regular video calls (ideally weekly, depending on interest).
Post-mortems: We will present post-mortems of our misses during meetups or publish them as joint posts on LessWrong.
Expert Insight: I plan to arrange calls with a few (super)forecasters in my network to discuss their workflows.
Pastcasting: We will use Sage to forecast on historical events (where the answer is hidden) and immediately discuss our results and process.
Training various relevant rationalist techniques/tools - calibration, CFAR techniques, AI tools useful for forecasting etc.
Community: We will coordinate via a public Discord group.
Commitment: There are no hard requirements to join, but I am looking for people willing to:
Make several forecasts per week.
Actually show up to meetups (or Discord) and discuss their post-mortems.
I’d like to hold the first meetup in the coming weeks, during which we will do short calibration excersize + pastcasting. Please indicate your interest in this form and join Discord. Date of the first meetup will be also announced on lesswrong event section.
(Open to other ideas on how to structure this—let me know in the comments).
A little info about me
In the past, I helped organize a forecasting tournament for the Czech Priorities, which had almost 200 participants. I am board member or Confido institute. Until last year, I was vice-president of Effective Altruism Czechia on CBG grant. I made a few dozen forecasts, but mostly to build the habit rather than to rigorously invest time in improving my skills—consequently, my actual score is abysmal right now.
EDIT: The first meetup will be Sunday, March 1, 11:00–12:00 CET.
I’ve read Superforecasting, but I find that actually applying the "10 commandments" is difficult in isolation. The feedback loops in the real world are too slow, and it’s too easy to skip post-mortems when no one is watching.
My goal for this year is to put in substantial work to become a superforecaster (or at least get much closer).
To do this, I am starting a dedicated online community for peer accountability and high-frequency practice. I’m looking for a small cohort of people who want to actually improve their forecasting skills.
The Plan:
Commitment: There are no hard requirements to join, but I am looking for people willing to:
I’d like to hold the first meetup in the coming weeks, during which we will do short calibration excersize + pastcasting. Please indicate your interest in this form and join Discord. Date of the first meetup will be also announced on lesswrong event section.
(Open to other ideas on how to structure this—let me know in the comments).
A little info about me
In the past, I helped organize a forecasting tournament for the Czech Priorities, which had almost 200 participants. I am board member or Confido institute. Until last year, I was vice-president of Effective Altruism Czechia on CBG grant. I made a few dozen forecasts, but mostly to build the habit rather than to rigorously invest time in improving my skills—consequently, my actual score is abysmal right now.