🏅 Top traders

From the Manifold team’s meeting notes on Notion:

Seems like they’re aggressively pursuing him.

@oh LMFAO!!! I guess if it’s on notion maybe he will see it next time he googles himself or whatever. But the only thing they do aggressively at Manifold is defend SBF.

So Nate Silver came to the Manifest but still hasn’t used the app, which means they paid him to come and he has probably already forgotten about it so long as his check cleared. Same with Robin Hansen. The reality is that these dudes are not believers in or promoters of prediction markets, they are self promoters interested in forecasting who occasionally blog about their base rates and Bayesian statistics (more virtue/value signaling for nerds).

@BTE its also possible that betting fake money for effectively nerd credit doesn't feel like an actual prediction market to him

@BTE Did you seriously just call Robin Hanson "not a believer in or promoter of prediction markets"?

@IsaacKing I know he is an expert on prediction markets but he is obviously scared to put his fake money where his mouth is because he is probably scared he will lose credibility if he cant hold a Brier Score under 0.05 or shoot to the top of the leaderboard.

@BTE this is a great marketing campaign idea. Daring high profile users to join by insinuating that they'd be bad at it.

@BTE Believing that prediction markets are useful is not the same as believing that you'd be good at turning a profit in them. Those seem like largely unrelated propositions to me.

@Quinn Worked for Eliezer

@IsaacKing You can be a very good forecaster and have hugely negative profits. Check out my profile because that is basically the one sentence description of it.

@IsaacKing Yes it is working for the AI crowd generally at the moment. I think they (AI stakeholders) should be exploring this much more aggressively. As of now it accounts for a fraction of a fraction of a fraction of one percent of the capital spend by the 500 largest AI stakeholders. That segment alone could be a bigger business than Snap if it gets deep penetration into the core AI labs. I am almost embarrassed to admit how much time I have spent thinking about this honestly. I would go work for Manifold to implement my vision tomorrow if they asked me. Between you and me, our markets, we account for like half of all the bets on all the site I bet. The fact that they essentially forgave all your self inflicted losses and returned your fiat which accounted for like ALL OF THEIR REVENUE in 2023 and put you into the black just to get you making markets again the most egregious moral hazard in the history of fake money. LMAO. Maybe we should take the party to Canada and use regulatory arbitrage to explore some new approaches that raise the stakes before the music stops??

Between you and me, our markets, we account for like half of all the bets on all the site I bet.

I'll accept that bet. How much? :)

@BTE FWIW Robin Hanson has an account now https://manifold.markets/RobinHanson

@BTE I very much doubt you can be a good forecaster if your profits continue to be negative and you have bet according to your forecasts.

Also what about ai stakeholders are you saying?

@AdriaGarrigaAlonso You didn’t look at my profile. I have better calibration score than the top trader on the leaderboard. My Brier Score on Metaculus for 400 resolved questions is 0.049 and I am ranked 111th. Profit here is a measure of some skill, but it’s not forecasting accuracy. More likely trading discipline, both timing and sizing.

[Deleted]

@BTE while Metaculus is arguably more serious, it does have its own issues as well
(I have a brier Score of 0 after 40 questions)
https://www.metaculus.com/accounts/profile/133153/

And you can get a good calibration in the long run by using "defensive forecasting" -- basically correcting for past deviations with future forecasts. I'm pretty sure showing a calibration curve to users induces this behavior. See the article of V Vovk, A Takemura, G Shafer.

@IsaacKing That's not true, it actually is a measure of forecasting accuracy. It's the mean-squared-error of the probabilities (https://en.wikipedia.org/wiki/Brier_score).

Calibration is the thing that you can fulfill using a random predictor.

@BTE okay, I think if you used your probabilities to determine the sizing of your trades (and nevermind the timing) you'd do better.

Though TBH a good strategy here is to get in on a market before it updates due to recent news that should put it at 0% or 100%, which doesn't need forecasting.

As a user right? Thought you meant as an employee for a second

He'll be at Manifest. I'd be surprised if he didn't sign up.

@NoaNabeshima Are you surprised yet?