Prediction site is accurate in 32 out of 35 Senate seats, maintains a certain level of accuracy - US midterm elections

[New York Current Affairs] It was revealed on the 4th that predictions by the leading prediction market platform ``Kalshi'' were accurate in at least 32 of the 35 seats up for grabs in the U.S. Senate election in the midterm elections in which the opposition Democratic Party won a landslide victory. This election has been described as the "biggest test in history" (Reuters) for the accuracy of prediction markets, but a certain level of accuracy has been maintained.

Candidates who invested huge amounts of money lost the election one after another, and the Republican Party's crushing defeat was the defeat of billionaires such as Musk.

In prediction markets, you can bet money by predicting the outcome of various events such as politics and sports. A comparison of Calci's predictions as of midnight on the 2nd, just before the midterm elections, with the actual wins and losses showed that as of 10pm on the 4th, he was correct in 32 of the 35 Senate seats. That includes the close state of Iowa, where Democrats had a 53% chance of winning.

However, in the southern state of South Carolina, which had been expected to have a 67% chance of being won by the Republicans, the Democratic Party won the seat and the prediction was wrong. Results are not yet known in Alaska and Mississippi.

In the gubernatorial election held at the same time, he lost in Texas, where he had predicted a Republican victory with a 60% chance. Results in other states were largely in line with expectations.

We predict a landslide victory for the Democratic Party in the House of Representatives election. Although the results are close, there are some cases in which candidates who expected a high chance of winning lost.

Prediction markets attracted attention for predicting that Trump would have a high chance of winning the 2024 US presidential election, despite major media reporting that it would be a close race. However, problems have been pointed out, such as when some users use the system for gambling purposes, or when the overall amount of bets is small, the predictions may be ``manipulated'' to make large bets.

According to Reuters, more than $1.48 billion (approximately 230 billion yen) was bet on the midterm elections across prediction markets.