Polls, Sentiment, Prediction Markets

It’s of some interest to see how polls and prediction markets are viewing the presidential race, and how this links with economic sentiment.

First polls and sentiment.

Source: TheHill, accessed 4/17/2024.

Source: U.Michigan via FRED, TradingEconomics, accessed 4/17/2024.

Now, a prediction market. Below, I show PredictIt odds for Biden vs. Trump (omitting RFK Jr.), for the last 90 days (corresponding to the period after the green lines in the top two figures).

Source: PredictIt, accessed 4/17/2024.


4 thoughts on “Polls, Sentiment, Prediction Markets

  1. Macroduck

    Y’all may have heard about Stephanopoulos interviewing Sununu about his flip-flopping on support for Trump. One bit of mendacity Stephanopoulos neglected to point out was Sununu’s repeated claim tath Trump has the support of 51% of the electorate. Trump is polling at well below 50% now. He lost the popular vote in both 2016 and 2020, and his approval rating never cracked 50%:


    The Trump crowd, including it’s sometime members, really don’t want the ignorant middle to understand how unpopular Trump is. The ignorant middle, in a few states, have a very large say in the choice of president.

  2. Macroduck

    G. Elliot Morris from 538/ABC has reviewed presidential preference polling accuracy as of mid-March in election years. Here are a couple of tidbits:

    “At this point (237 days until the election) in the 1980 election cycle, for instance, incumbent Democratic President Jimmy Carter was ahead of Republican Ronald Reagan by 14 percentage points. But on Election Day, Carter actually lost by 10 points. That’s a 24-point error.”

    Such a miss is not uncommon in the history of pre-election public opinion polls. In 1992, the early polls were off by 19 points; in 1972, by 12; and in 2000, by 8. Over our whole dataset, polls from mid-March have missed the final outcome of the presidential national popular vote by an average of about 8 points.”


    Morris provides a graph of mid-March polling vs actual popular vote results over twelve presidential elections from 1972 to 2020. Democrats outperformed their mid-March polling numbers 7 times, under-performed 3 times and matched 3 times. Under-performances were all 1984 and earlier. Kinda looks like a modest systematic bias toward underestimation of Democratic presidential candidates. Could be a small-numbers problem.

    1. Moses Herzog

      538 has always been one of my favorite sites to gauge elections ahead of time. That being said, they are crappy at gauging the rural vote, and seemingly have no interest in fixing this error that plagues them in the last 3 presidential elections. I suspect Nate Silver is just too cheap to do something other than live off years past windfalls.

      1. Macroduck

        Nate Silver has cashed out. NYT couldn’t handle him, so he moved to the broadcast world. 538 is an ABC product now and ABC cut staff at 538. ‘Cause news is important?

        I never much cared for 538’s transcipts of long conversations, nor its sports coverage, nor its scant efforts at economics. “Just the facts, Ma’am” political coverage is (was?) good, though as you note, it has flaws.

Comments are closed.