Investors and election bettors were caught on the wrong side of Brazil’s presidential election, when leftist president Luiz Inácio “Lula” da Silva defied expectations to secure victory. Brazilian stocks crashed, with the iShares MSCI Brazil ETF briefly falling 25 per cent, a loss amplified by the real’s sharp decline against the dollar.
Investors were not the only ones wrong-footed by Lula’s surprise victory over Flávio Bolsonaro. Betting on prediction markets such as Polymarket and Kalshi gave Bolsonaro roughly an 80 per cent chance of winning, prompting Bloomberg’s Joe Wiesenthal to ask if this was “one of the biggest prediction market whiffs of all time”.
Statistician Prof Harry Crane responded with a pithy question of his own: was this the first time in history that a four-to-one shot had won?
Crane has a point. Commentators often bemoan that betting markets get it wrong when an unexpected result occurs. However, unexpected things happen all the time.
READ MORE
Everyone knows the best football team will occasionally lose to an unfancied opponent. A bookmaker who gives them an 80 per cent chance of winning isn’t proved wrong by the 20 per cent of occasions when they don’t.
What about the possibility of political bias distorting market prices? Crane’s research shows that even when many bettors are politically biased, their influence on prices is usually outweighed by profit-motivated traders.
In 2024, prediction markets comfortably outperformed leading polling-based forecasts across the US presidential race.
That does not mean they don’t have flaws, or that Lula’s odds were justified. However, one upset doesn’t condemn a forecasting method. Investors should be wary of treating unexpected market moves as proof that the consensus was wrong.
