Polymarket and Kalshi are about to be marked in public. The United States midterm elections will show whether two betting sites that studies say can predict races correctly can still do so when a national ballot is counted in full view.

The claim behind both platforms is stark. Put money on a political outcome and the price becomes a forecast. Researchers have found that this method can call races correctly. A midterm night is a harsher setting than a paper. Many contests resolve together. Late swings land in public. There is nowhere to park a bad call.

“A national election will not grade these markets on style. It will grade them on whether they called the races.”

— Kings Journal

From academic papers to election night

Betting sites of this kind run on a blunt incentive. Traders lose money when they are wrong, so prices should move towards the likeliest result. That logic has been enough to impress studies that checked market calls against later outcomes. What those studies have not had to survive is a single nationwide round of voting in the United States, watched in real time by people who will remember every miss.

This is why the midterms matter beyond the circle of people placing bets. If Polymarket and Kalshi track the results with the reliability that research has described, it becomes harder to treat the odds as a parlour amusement. If they stumble, those studies will look less like a method and more like a streak that ran out.

“The midterms are less a novelty for betting sites than a public audit of their forecasting claim.”

— Kings Journal

What the test will measure

  • Whether race calls that looked sound in studies still hold across a national ballot
  • Whether the platforms stay accurate when many contests resolve on the same night
  • Whether a correct forecast in research settings survives the noise of a live campaign

None of this needs a grand theory of politics. It needs a count. When the ballots are tallied, the prices Polymarket and Kalshi showed before the close will either have pointed at the winners or they will not. Studies already say these sites can predict races correctly. The midterm elections in the United States are where that statement stops being a finding in a paper and becomes a result anyone can check.