Sign in Subscribe
Ex-presidents

The Accuracy of Prediction Markets in Election Forecasts: A Nuanced View

Strategic Insights

— 1 min read

Prediction markets, where users can bet on election outcomes with real money, have gained attention for their ability to provide a more accurate forecast than polling data. However, experts are divided on their effectiveness, with some arguing they can be just as accurate as election forecasts, while others point out that they can be volatile and subject to outside influences. A closer look at the data reveals that prediction markets can sometimes accurately predict election outcomes, but their accuracy is not consistently better than poll-based forecasts.

Key Takeaways:

  • Prediction markets can be just as accurate as election forecasts in predicting election outcomes, but their accuracy is not consistently better.
  • The 2008 and 2020 election cycles saw comparable results between prediction markets and FiveThirtyEight's election forecasts.
  • Research has shown that prediction markets can react more quickly to news events than traditional polls, but this can also lead to volatility and noise.
  • The nature of prediction markets has made research challenging, with betting on elections being prohibited in the US until recently.
  • The 2022 midterm elections saw prediction markets foretell a "red wave" of Republican victories, but actual results were a mix of Democratic and Republican wins.
  • Experts note that prediction markets can combine individual interactions, making them more accurate than polls, but also introducing noise and volatility.
  • Research suggests that prediction markets can be influenced by a small number of large investors, making them vulnerable to market manipulation.
  • Despite the promise of prediction markets, experts agree that there is no clear evidence yet to suggest they can outperform other election forecasts.

Statistics:

  • Over $100 million has been bet on election outcomes through prediction markets.
  • In the 2022 midterm elections, prediction markets priced a win in Pennsylvania by the Republican Senate candidate, Dr. Mehmet Oz, at 63 cents.
  • The price of a share predicting that Mr. Biden would withdraw from the presidential election in 2020 jumped from 21 cents before the debate to 60 cents by July 3.
  • The prediction market Polymarket has over 1 million users.
  • The user bet limit on PredictIt is $850, while Kalshi limits total bets to $100 million.

Sources:

  • The New York Times, "The Case for Prediction Markets"
  • FiveThirtyEight, "The Accuracy of Prediction Markets"
  • Intrade, "2012 Presidential Election Results"
  • Nate Silver, "The Case for Presidential Debates"
  • Kalshi, " About Kalshi"
  • Harry Crane, "The Limits of Prediction Markets"
  • Elon Musk, "Tweet on prediction markets"
  • Polymarket, "About Polymarket"
  • PredictIt, "About PredictIt"

Read more