Season three is upon us and the challenge just got tougher. With 22 predictors battling it out for a spot in the top 4 the steaks have never been higher.
This market will resolve to the top 4 at 25% each after the random time period has elapsed and the scores have been finalised.
Feel free to use the comments here as a League chat to discuss/sledge.
Your competitors are in no particular order:
Chris Billington @chrisjbillington
Spongpad @LBeesley
Valery Cherepanov @qumeric
Michael @Michael
Pat Myron @PatMyron
Zach Stein-Perlman @ZachSteinPerlman
vincent pearce @vincentpearce
Heliscone @Heliscone
bessarabov @bessarabov
XComhghall @XComhghall
Lavander @Lavander
Light @light
Nikola @Nikola
Martin Randall @MartinRandall
brubsby @brubsby
Wamba Ivanhoe @ShitakiIntaki
Imbens @Imbens
power creep @astyerche
hi @DeadRhino
tinytitan @tinytitan
Ian Philips @ian
slightly @notleagues
Congratulations and the best of luck to our diamond promotions into masters. @brubsby @ian @chrisjbillington @ZachSteinPerlman. May the odds be ever in your favor.
@brubsby it is all for science! I don't know if Manifold has ever stated what sort of population distribution they want across the tiers, but I suspect they will be growing the Masters population which will lend itself to more and more cohorts and more homogenous interest neighborhoods. Season 3 Masters only had 20, but 10 are staying and 24 are incoming from diamond so Season 4 will have 34. I anticipate this growth to continue, but who knows maybe they will change the promotion/demotion thresholds to stabilize the tier populations, but Season 4 seems to be a bit early for that to happen.
The incoming Season 4 Masters
@LBeesley I don't entirely understand it myself, but I will explain what I think happens.
Manifold tracks your interests based upon what you have told Manifold and also based upon what markets you have opened up to look at.
When creating the cohorts at the beginning of each league season, they line up all the users in a shared tier, such as Diamond, sorted by the interest so the users on either side of you in that "line" have the most similar interests to you out of all the users in the same tier as you. Then the chop up the line into more or less equal parts to make the cohorts.
Theoretically everyone in a shared cohort have interests in common with their neighbors, creating a sort of interest neighborhood.
@ShitakiIntaki Here are the interest vectors that build up interest neighborhoods.
const TOPICS_TO_WORDS: { [key: string]: string } = {
'Supreme Court':
'Supreme Court, SCOTUS, Justice, Judge, Law, Constitution',
'LGBTQIA+':
'lesbian, bisexual, gay, transgender, queer culture',
'Effective Altruism':
'effective altruism, charity, donations, philanthropy, animal rights, existential risk',
AI:
'artificial intelligence, AI, machine learning, deep learning, neural networks, LLMs, large language models, GPT-4',
Science:
'science, physics, chemistry, biology, neuroscience',
Religion:
'religion, God, Christianity, LDS, church',
Education:
'education, school, college, university, student, teacher, professor',
'TV Shows':
'tv shows, television, Netflix series, HBO, Hulu, season finale',
Music:
'music, release full length album, next album, concert',
Movies:
'movies, films, cinema, movie reviews, movie ratings, Hollywood, box office, IMDB, Rotten Tomatoes, Academy Awards',
Books:
'books, novels, fiction, non-fiction, book reviews, book ratings',
'Internet Culture':
'internet culture, memes, viral videos, TikTokers, Twitch streamers, online drama',
Celebrities:
'celebrities, actors, actresses, musicians, singers, artists, athletes, sports stars, movie stars, TV stars, famous people',
Crypto:
'crypto, cryptocurrency, Bitcoin, Ethereum, blockchain, NFTs, DeFi, smart contracts, altcoins, stablecoins, Dogecoin, Shiba Inu',
'Elon Musk':
'Elon Musk, Tesla, SpaceX, Neuralink, Boring Company, Starlink, Hyperloop, Twitter, X.ai',
Climate:
'climate change, global warming, climate science, global temperatures, carbon emissions, renewable energy',
Biotech:
'biotech, biotechnology, CRISPR, drug trials, pharmaceuticals, vaccines, gene editing',
Math: '
mathematics, Millenium Prize, math olympiad, proof, puzzle, paradox, numbers, equation',
Nuclear: '
nuclear, nuclear power, fission, fusion',
Space:
'space, astronomy, astrophysics, space exploration, NASA, SpaceX, Mars, Moon, solar system, planets, stars, galaxies',
Racing:
'racing, Formula 1, NASCAR, IndyCar, MotoGP, World Rally Championship, WRC, Le Mans, Dakar Rally, Formula E',
Soccer:
'soccer, football, Premier League, La Liga, Serie A, Bundesliga, Ligue 1, MLS, Champions League, World Cup, Euro Cup',
Finance:
'finance, investing, Wall Street, S&P, hedge funds, private equity, venture capital, IPOs, SPACs, mergers and acquisitions, M&A, bonds, options, futures, commodities, derivatives',
Economics:
'economics, macroeconomics, game theory, GDP growth, inflation',
Startups:
'startups, entrepreneurship, founders, angel investors, VCs',
Gaming:
'gaming, video games, esports, Twitch, Steam, Xbox, PlayStation, Nintendo',
'Sex and love': '
sex, relationships, dating, love, marriage, divorce',
'Personal development': '
personal development, self-improvement, self-help',
Programming:
'programming, software development, coding, computer science',
Baseball: '
baseball, MLB, World Series, home runs, strikeouts',
NFL: '
NFL, football, Super Bowl, touchdowns, interceptions',
Chess:
'chess, chess openings, World Chess Championship, Magnus Carlsen',
@ShitakiIntaki That checks out for sure. I know that while my time here is a bit shorter than most everyone else in this league, I’ve created or bet on a market for just about every topic listed there.
The way leagues are structured, you don't want to promote into a league where you are the middle of the pack, if you are motivated by league rewards that is. There are bragging rights to being able to say you promoted to Masters, but the return on league rewards are better if you are just on the bubble of promoting and do not promote. Middle of the pack is actually worse than demoting and promoting every other month.
This was a chart I constructed based off of last seasons promotion structure and this season's reward schedule. I modeled players naively as if they had a constant percentile ranking which would always determine their season rank within their cohort. Average tier plot shows some players alternating between different tiers, some stuck in the same tier every month. The red line is average income plotted with a log10 scale on the right hand side.
You can see being ranked 10th in Masters is just enough to keep you from demoting back to Diamond and you make less than if you were demoting and promoting, but better still is not promoting at all so that you don't run the risk of a dry month waiting to demote out of Masters.
So these doldrums might be pretty small depending on how much neighborhood mixing there is. If the population interest neighborhoods are laminar then then this pattern might show up in each interest locus, but I suspect there might be quite a bit of cross interest mixing as the seasons progress so long term performance may be much more volatile.