works
Scott Alexander Prediction Market FAQ online Prediction markets, similar to stock markets but for beliefs about future events, possess accuracy and canonicity. Their prices reflect the aggregate belief in the likelihood of an event. This mechanism efficiently incorporates information, as mispricings create arbitrage opportunities, incentivizing correction. Ideally, prediction markets provide a unified, unbiased truth source, resistant to manipulation due to financial incentives for accuracy. However, real-world limitations include transaction costs, limited liquidity, regulatory hurdles, and distrust. These factors can lead to persistent mispricings, especially for long-term or low-value events. Despite these limitations, empirical evidence demonstrates prediction markets often outperform average individuals, experts, and polls, offering a potentially valuable tool for decision-making and social consensus on various factual questions, from vaccine efficacy to climate change. – AI-generated abstract.

Prediction Market FAQ

Scott Alexander

Astral Codex Ten, January 10, 2022

Abstract

Prediction markets, similar to stock markets but for beliefs about future events, possess accuracy and canonicity. Their prices reflect the aggregate belief in the likelihood of an event. This mechanism efficiently incorporates information, as mispricings create arbitrage opportunities, incentivizing correction. Ideally, prediction markets provide a unified, unbiased truth source, resistant to manipulation due to financial incentives for accuracy. However, real-world limitations include transaction costs, limited liquidity, regulatory hurdles, and distrust. These factors can lead to persistent mispricings, especially for long-term or low-value events. Despite these limitations, empirical evidence demonstrates prediction markets often outperform average individuals, experts, and polls, offering a potentially valuable tool for decision-making and social consensus on various factual questions, from vaccine efficacy to climate change. – AI-generated abstract.

PDF

First page of PDF