Hypermind

Lend your brain to Ukraine! Glimt is a joint project of the Swedish and Ukrainian governments, powered by Hypermind, open to everyone. Perhaps the first use case of crowd forecasting as a weapon of war against evil. Join us here: glimt.nu/glimt/en/welco… @kikollan @NunoSempere @jorgebmontanes @PTetlock @robinhanson @JamesSurowiecki @JustinWolfers @IgnatiusPost @pennockd @DavMicRot @Scott_E_Page @page_eco @GlimtCommunity @FOIresearch

Tuesday morning Election 2024 forecast landscape: Skilled forecasters panels favor Harris, betting platforms favor Trump, and statistical models have no idea.

@ochlosophe @Onefora29958575 Agreed.

@Hypermind_com @Onefora29958575 I can’t help but think that Trump being over 65% was seriously biased, even if it’s still basically a coin flip overall.

Kamala #Harris is back to a 60% victory chance on @Hypermind_com. She hasn’t dropped below 50% since Biden stepped down. #USElections2024 🇺🇲🗳️

@ochlosophe @Onefora29958575 Whatever happens, nobody will be totally right or totally wrong. It's just probabilities on a single event.

@Onefora29958575 @Hypermind_com Look, betting on #Trump with the cryptobros? Huge mistake. Total scam! Way overpriced, believe me! #Polymarket #Kalshi

Final pre-election day forecast landscape. Skilled forecasters give an edge to Harris. Betting markets (except PredictIt) give an edge to Trump. Statistical models occupy the toss-up middle-ground.

Harris' chances continued to improve overnight. Interestingly, the US-based mrkts give her better odds (51% avrg) than the foreign-based mrkts (46% avrg). She's the favorite of the skilled forecasters panels (Hypermind, Good Judgment, Metaculus), with 55% probability avrg.

Accelerating Kamalamentum in the last few days before the vote.

@JamesSurowiecki Kamalamentum, forecasting platforms edition

The race had already tightened on other prediction-markets platforms, like Kalshi and PredictIt. Now noticeably tightening on Polymarket, with Harris' chance of winning above 40% for the first time in weeks.

@peterwildeford It's just a correction back toward the true odds.

Prediction markets have moved 9 percentage points towards Harris over the past three days with basically no reason (especially given the magnitude of the shift)

With under a week to go until the election, Polymarket has become a mainstream source of data — but the suspicious activity raises questions about the accuracy of the site. You can read my full article here: fortune.com/crypto/2024/10…

Scoop: Blockchain researchers have found evidence of rampant wash trading on the leading electoral betting site Polymarket, with Chaos Labs concluding that one-third of volume on its presidential market is likely artificial

@ochlosophe Garbage in, garbage out.

2 salles, 2 ambiances #polymarket 🆚 #Hypermind

Comparative analysis of Harris vs Trump forecasts over the last month. @GJ_Open @superforecaster @metaculus @FiveThirtyEight @TheEconomist @ManifoldMarkets @PredictIt @Betfair @Polymarket @Kalshi

@ochlosophe « Prediction polls » est le terme technique. Voir researchgate.net/publication/28…

@Hypermind_com Comment les qualifier ?

La Russie 🇷🇺 va-t-elle étendre son territoire en Ukraine 🇺🇦 en 2026 ? (+ 2 localités parmi une liste) 🔴si Trump est élu, oui à 80% 😵 🔵si Harris est élue, oui à 45% 😑 source : marché prédictif de Metaculus metaculus.com/questions/2850…

60/40 is not even, but its hardly settled either. Race could still go either way. electionbettingodds.com/President2024_…

It is remarkable that our prediction market is now (Oct 26) the only crowd-forecasting or statistical model that still gives Kamala Harris slightly more chances of winning (54%) than losing. All others are now favoring Trump more or less boldly.

Join the FIGHT to make the most accurate forecasts about swing states and congressional control! New contest launched today with > $1,000 reward. predict.hypermind.com

@nikosbosse you might also be interested in this research on infectious disease crowd forecasting that we conducted with Johns Hopkins : drive.google.com/file/d/1T_MBIn…

@nikosbosse Thanks @nikosbosse. We have not drawn any conclusions yet. Just getting started. Need to gather more data, including 2024.

I don't quite agree with the conclusions they draw. In particular, I think it's ~impossible to draw conclusions about relative performance based on forecasts for a single event. But I appreciate the effort of providing data and engaging with the forecasting community

Shoutout to @Hypermind for sharing some of their data

@NunoSempere @nikosbosse @TLiptay And this is Hypermind's calibration on all questions over the last 10 years.

This is n=1 election. The 538 track record in contrast has many electoins, and so one can meaningfully speak of "calibration". projects.fivethirtyeight.com/checking-our-w… What you are saying amounts to "Hypermind was more bullish on Trump in 2016, or just more uncertain". This is somewhat informative. But it's not really a track record, in that one election alone is a poor discriminator

Strong claim! @Hypermind_com do you have evaluations you could show to support that claim?