The arena of corporate earnings forecasting is undergoing a quiet revolution. Each quarter, numerous Wall Street analysts build financial models, mine alternative data, and compete for communication with company executives, all to accurately predict corporate earnings. However, a recent study suggests that anonymous bettors on Polymarket may be more accurate at predicting company profits than Wall Street analysts. Why are prediction platforms more accurate? A report from brokerage firm Wolfe Research indicates that when Polymarket users bet on a company's earnings falling short of expectations, their accuracy rate reaches 44%, more than double the historical baseline of 18%. Conversely, when traders are highly confident that a company's results will exceed expectations, the accuracy of this judgment can reach 90%, surpassing the industry average of 81%.

Yin Luo, Head of Quantitative Research at Wolfe, stated, "This high accuracy may stem from the crowdsourcing model. When it comes to earnings predictions, the investors placing bets on the Polymarket platform are likely more diverse than the market consensus expectations based on sell-side analyst forecasts." This result further indicates that prediction platforms have the potential to become an important information source for investors and could even become competitors to sell-side analysts, whose core function is earnings prediction.

A working paper updated in early April by researchers from London Business School and Yale University pointed out that these emerging platforms boast high prediction accuracy, can integrate new information faster than analysts, and can simultaneously avoid some biases inherent in Wall Street earnings forecasts. The researchers believe the exceptional accuracy of prediction platforms is due to participants betting with real money; they also found that users participating in earnings predictions are unusually professional. The researchers noted that insider trading could also be a factor influencing prediction accuracy.

Despite being small in scale, the potential is significant, and Wall Street capital is already taking notice. Regardless of the underlying drivers, these two studies highlight the substantial potential of event contracts tied to earnings. Although such contracts currently represent a very small portion of trading volume on platforms like Polymarket and its main competitor, Kalshi, the research confirms their value. This further supports the platforms' view that while sports betting currently constitutes the bulk of their trading volume, these new financial derivatives will ultimately play a significant role on Wall Street.

Predicting whether a company will beat earnings expectations can be complex. Primarily, the vast majority of companies do exceed earnings expectations, partly because company executives often deliberately lower expectations to create positive earnings surprises. Since last September, Polymarket has allowed users to place "Yes/No" contract bets on whether certain large-cap stocks will beat earnings expectations. To compare the accuracy of these bets against Wall Street expectations, Wolfe studied approximately 430 earnings reports covered by Polymarket, representing about a quarter of the Russell 1000 index constituent reports during the same period.

The Wolfe research team wrote regarding prediction platforms: "The signals they generate will provide an increasingly rich and high-frequency perspective. Using this perspective, we can delve into information aggregation, belief formation, and the pricing of uncertainty across almost all event areas that drive market volatility." Although exchange operators and other financial institutions are investing heavily in prediction platforms, the field is still in its early stages. According to user-compiled data on Dune Analytics, the trading volume for earnings predictions on Polymarket was only $795,315 in a recent week, accounting for 0.03% of the platform's total trading volume.

Vinesh Jha, founder of alternative data platform ExtractAlpha, which also offers earnings prediction betting, stated that he sees Polymarket as a complementary source of information. However, it might be too early for widespread integration into quantitative fund managers' trading systems. Jha said, "The market is more concerned with the specific earnings figure and how the market will react, rather than just whether results beat expectations. It's premature to draw conclusions, and the relevant data is still too thin."