Now that the midterms are over, major election races have been settled. Many of the Senate races have been settled with the winner priced at 99 cents on PredictIt and the loser at one cent. Analyzing what bettors missed in these races can inform betting strategy for election markets in the 2024 presidential race.
However, some markets had prices that add up to 99 or 101 cents. On November 10, the Senate control market priced Democrats at 91 cents and Republicans at 11 cents. On November 28, Democrats were priced at 98 cents and Republicans at 1 cent. Even as results were announced and results were finalized, PredictIt probabilities didn’t always add up to 100%.
Why Pricing Sometimes Doesn’t Add Up to 100
The PredictIt pricing implies uncertainty, but these small price differences really come from noise. Current prices in any market may be 98 cents, but the best price may be 99 cents. That means some traders will buy a Democratic contract at 98 cents and some will buy at 99 cents. Buying at different prices causes price movements. Those small price movements are more visible as a market comes closer to settling and can result in 99-cent and 101-cent totals.
Once markets settle, the correct outcome is priced at 99 cents and the incorrect outcomes are almost always priced at one cent. (Utah’s Senate race settled with a one-cent overround, with the Republican winner at 99 cents and the Democratic and Independent candidates at one cent each.)
Prediction market prices are not evidence of voter fraud or election meddling. But now that the midterms are almost over, high-profile election deniers are trying to overturn the fair elections they lost in.
Arizona’s Election-Denier Conflicts
One of the dominant media narratives over the past few weeks has been that democracy triumphed over election denial. Election-denying governors, senators, and representatives lost in swing states. These stories make it easy to forget about the brewing conflicts from the election deniers who made it to office.
One conflict has already begun in Cochise County, Arizona.
On November 28, Reuters reported that two of the county’s three board of supervisors voted to “postpone certifying the county’s election results.” The two Republican board members out-voted the one Democrat.
Election deniers pushed claims that Arizona’s voting machines lacked proper certification. Not only has no evidence of improper certification been presented. Arizona’s Secretary of State plans to file a lawsuit on Dec. 5 to force the board to certify the results. The office of the Secretary of State has already sent the board proof of the voting machines’ documentation. The only thing holding the vote up is irrational election denial.
Republican gubernatorial candidate, Kari Lake, has also refused to concede her loss. Donald Trump has also called for Lake to be “installed” as governor. The election denial movement hasn’t gone away because of one election cycle. It’s one of several forces that election bettors will have to weigh in the next election.
Betting Strategy for Election Markets
Relying solely on media reports about what issues are important to voters is a poor way to make accurate political predictions. It’s not that all media sources are untrustworthy. It’s that media reports are an incomplete picture of the entire political landscape.
Leading up to the midterms, many news outlets predicted inflation and other economic issues would decide the midterms. However, the polls underlying these conclusions underrepresented young people who were motivated by Roe v. Wade’s overturning and by opposition to election denial.
The challenge for political forecasting is not only catching these leftover pieces of information. Predictors also have to figure out whether missing bits of information will determine close races.
Always ask what’s missing from credible political analysis. Credible coverage can help viewers understand new issues and add important context. But no single story or station can fit every relevant piece of information into one story or separate noise from useful predictive signals. There’s always something else to learn.