Kalshi banks on its predictions’ accuracy
The company studied its bets and found they came closer to reality than most AI models and track human “superforecasters.”
Wall Street’s preferred betting arena, the Fed funds futures market, recently predicted a 2/3 chance of the central bank keeping interest rates unchanged during its last meeting. Kalshi, a prediction market, had odds up to 10 percentage points higher. Both predictions proved correct when the Fed decided to maintain steady rates.
However, a discrepancy exists, as a 10-point advantage could significantly benefit those trusting Kalshi’s data over CME’s calculations. Kalshi aims to evolve beyond its casino image, combat political opposition, and establish itself as a credible participant in financial markets. To achieve this, the company has recruited a former Standard Chartered executive to expand its institutional business, encouraged market-makers such as Susquehanna to offer liquidity, and is constructing a terminal to compete with Bloomberg for traders’ desktop space.
In addition to maintaining a secure environment free from insider trading, Kalshi must also deliver accurate predictions. To prove its claim, Kalshi released data from 2.3 million bets made on its platform from its inception in 2021 up to the middle of 2023. The research, conducted by Kalshi’s internal team but using publicly accessible data, reveals a high level of accuracy, particularly for economic and financial markets like inflation, unemployment, and commodity prices.
The Brier score, a widely accepted metric for evaluating forecasts, measures the precision of predictions. A score of zero represents perfect accuracy, 0.25 indicates a random guess, and 1.0 signifies consistently incorrect forecasts. Kalshi markets scored 0.02 shortly before a market closes, similar to the precision of human "superforecasters" and even outperforming leading AI models.
According to the study, Kalshi’s accuracy is highest in markets that are difficult to manipulate, such as elections and economic indicators, and lower in easier-to-manipulate markets like sports and "mentions" markets, which involve predictions about what a public figure, such as Donald Trump, will say next. While Kalshi strives to be a respected member of the financial industry, its success will depend on providing highly accurate predictions.
Written by urgent.news from Semafor's reporting — not their text. Machine-written; read the original for the full account.
