@Ariston_Macro: The ceiling of analysts, Michael Mauboussin's new book *The Wisdom of Crowds in Markets* systematically analyzes the information aggregation mechanisms of prediction markets, sports betting, horse racing betting, and stock markets, answering a core question: when do markets exhibit "wisdom of crowds," and when...

Analyst’s Ceiling: Michael Mauboussin’s latest book, The Wisdom of Crowds in Markets, systematically analyzes the information aggregation mechanisms of prediction markets, sports betting, horse racing wagering, and stock markets, addressing a core question: when do markets exhibit the “wisdom of crowds,” and when do they descend into “crowd madness”? The book explains bubbles, market efficiency, and the trading value of prediction markets, offering significant reference value for understanding price discovery and market behavior….

1. The “wisdom of crowds” and “crowd madness” in markets are not mutually contradictory, but rather two states within the same framework. Most of the time, markets can effectively aggregate dispersed information, so prices are generally efficient; however, when the key conditions for the wisdom of crowds break down, the market enters a phase of crowd madness, and prices begin to deviate significantly from value. Overall market behavior can be uniformly explained by whether the conditions for the wisdom of crowds hold, rather than as a dichotomy between efficient markets and behavioral finance.

2. The wisdom of crowds relies on three core conditions: Diversity, Aggregation, and Incentives. Here, cognitive diversity does not refer to social attributes such as age or gender, but to differences in information, knowledge, heuristics, cognitive frameworks, and mental models. Scott Page’s Diversity Prediction Theorem provides a mathematical expression: Collective Error = Average Individual Error − Prediction Diversity. This means that overall predictive power derives not only from individual ability but also from the diversity among predictions—the higher the diversity, the smaller the collective error.

3. A market is fundamentally an information aggregation machine. Although different markets employ different mechanisms—such as Continuous Double Auction, bookmaker pricing, and Parimutuel Pools—they all share the same objective: to continuously aggregate dispersed information into prices. Research shows that simple averaging can improve forecast accuracy, while aggregation methods such as weighted averaging and extremizing can further enhance predictive accuracy. Thus, the price formation mechanism itself is a critical source of market efficiency.

4. Incentives determine information quality. Real markets incentivize participants to seek information advantages (Edge) through profits and losses, and the size of a wager should reflect one’s confidence in that edge. At the same time, the Grossman-Stiglitz paradox demonstrates that a fully efficient market cannot exist—if prices already contained all information, no one would bear the cost of acquiring it; and without anyone acquiring information, the market cannot be efficient. Therefore, markets can only remain “Efficiently Inefficient”: edges exist, but they must be sufficient to cover information costs, transaction costs, and market impact costs.

5. The author classifies market problems into four categories: Needle-in-the-Haystack (finding a known answer), State Estimation (estimating the true state of the future), Prediction with Resolution (forecasts with a definitive final outcome), and Prediction without Resolution (forecasts with no final resolution). Stock markets fall into the last category, making them the most complex and the most prone to sustained deviations, whereas prediction markets and sports betting are markets where outcomes can ultimately be verified.

6. Galton’s ox-weight guessing experiment validated the most classic phenomenon of the wisdom of crowds: among 787 participants, the vast majority of individual predictions were inaccurate, yet the average of all predictions was almost perfectly aligned with the true weight. What truly drives this effect is not that every individual is skilled, but that independent errors continuously cancel each other out. As a result, the collective prediction outperforms the vast majority of individuals—this is the most important statistical foundation of the wisdom of crowds.

7. Prediction markets are, in the author’s view, among the closest approximations to an ideal information aggregation mechanism. Tens of millions of trades on Kalshi show that market prices closely track final true probabilities, and long-term accuracy significantly outperforms most polls and individual expert opinions. Although statistically aggregated Polls and Superforecaster teams can match or even exceed prediction markets, prediction markets as a whole remain one of the most effective information aggregation tools currently available.

8. The greatest trading value of prediction markets comes from three findings: First, market prices can almost be directly treated as probabilities; second, the Favorite-Longshot Bias persists over time—markets undervalue favorites while systematically overvaluing longshots; third, although market manipulation can temporarily move prices, genuine information ultimately reprices the market, and manipulators typically exit with losses, making sustained manipulation difficult. Meanwhile, profits are highly concentrated—the top 1% captures 77% of profits, Makers are profitable over the long term, while Takers lose money over the long term.

9. Sports betting markets are similarly highly efficient. Bookmakers not only balance both sides of their positions but actively exploit public biases when setting odds. Although the Favorite-Longshot Bias is widespread, the vast majority of deviations are too small to overcome the Vigorish, and thus cannot form stable arbitrage opportunities. Only about 3%–5% of sports bettors are genuinely profitable over the long term, and Sharps often need to resort to Runners, Beards, and other methods to circumvent betting limits.

10. The parimutuel horse racing market offers another form of information aggregation. Odds are entirely determined by the distribution of wagered money, and final Closing Odds are typically highly efficient, yet the 15%–24% Takeout consumes nearly all arbitrage opportunities. Smart Money typically enters in the final moments before a race concludes, so Closing Odds significantly outperform early odds, while major public events (such as the Triple Crown) tend to produce tradable deviations due to a large influx of non-professional capital.

11. The biggest difference between stock markets and betting markets lies in the positive long-term expected value. Betting is a zero-sum game minus costs, whereas stocks represent ownership in companies that generate cash flows over the long term, allowing aggregate wealth to grow persistently. Over the past century, the U.S. stock market has created approximately $91 trillion in wealth, while betting has accumulated long-term losses of roughly $5.8 trillion. The vast majority of gamblers lose money over time, while the vast majority of stock investors make money over time.