Prediction market giants Kalshi and Polymarket are showing a recent surge in bets on right-wing populist Abelardo de la Espriella to be the eventual winner of Colombia’s presidential election at the end of May.
The markets are out of line with conventional polls in Colombia, which have leftist candidate Ivan Cepeda comfortably leading the race with de la Espriella in second, followed by center-right candidate Paloma Valencia.
In the last week, however, bets on de la Espriella to win the election have increased relative to bets on his rivals.
On May 1, de la Espriella was given a 28.8% chance of victory whilst Cepeda was given 38% on betting market Kalshi. As of May 8, de la Espriella has overtaken Cepeda by 1 percentage point, reaching 42%.
Polymarket shows a similar trend. De la Espriella’s odds have risen from 28% to 39% over the last seven days, though he still trails Cepeda, who remains on 41%.
Over time, the race has been more measured. Since September of last year, Kalshi bet data show de la Espriella and Cepeda wrangling for first and second places until March when Valencia exploded onto the scene following her win in the right wing political party primary in which de la Espriella did not participate. She has since dropped – both on prediction markets, and in the polls.
Sergio Guzmán, director of Colombia Risk Analysis, isn’t convinced of the predictive power of betting markets in the case of Colombia’s elections.
He observes that, before the Colombian primaries in March, Valencia was unfancied on betting markets. She stormed to victory with 3,236,286 votes in the primaries while her party obtained 16% of the vote share in legislative elections. Meanwhile, de la Espriella declined to take part in a primary, but the party he leads, Movimiento de Salvación Nacional, won only 4% of the vote share. Cepeda’s party, Pacto Historico, held a primary in October 2025 which he won with 1,186,095 votes.
“The best information we have suggests that Cepeda and Paloma have the highest proven ability to get votes,” Guzman told Latin America Reports. “To me, that suggests something is off in the prediction markets.”
The Poll Weighting Index from news website La Silla Vacia combines polling data from the five authorized polling companies and finds that de la Espriella has 24% of voter intention, while Cepeda has 38%.
This would be sufficient to earn the right-wing populist a place in the second round, but the same poll index suggests de la Espriella would lose the second round by three points to Cepeda.
The trouble with prediction markets for polling
Whilst prediction markets are not new, the recent rise in the popularity of sites such as Kalshi and Polymarket, presents challenges to democratic processes that require greater public understanding.
Unlike polls, prediction markets do not measure voting intentions but the hunches of investors. They allow users to trade cryptocurrency-based shares on future events. Correct predictions pay $1 per share; incorrect ones lose their value.
Thus, rather than being an average of users’ probability beliefs, prediction markets are skewed towards the predictions of those with more capital to commit and are vulnerable to manipulation by those with the means to influence the perception of the likelihood of an outcome.
There is some evidence that prediction markets have outperformed polls on electoral margins forecasting, but both get significantly more accurate closer to the predicted event and should not be treated as reflecting the probability of an outcome.
Kalshi’s market for bets on the Colombian presidential election is relatively shallow, with only USD $385,390 bet at the time of writing, making the odds particularly volatile, and potentially vulnerable to manipulation.
Coljuegos, the state-owned body responsible for regulating online gambling in Colombia, does not recognize either Kalshi or Polymarket as an authorized operator.
Changes to Colombia’s poll process
The discrepancy between poll data and prediction market data may partly reflect the age of the available polling data. In Colombia, the most recent authorized surveys were conducted no later than April 30, meaning all publicly available polling is now more than a week old.
Prediction markets such as Kalshi and Polymarket, by contrast, update continuously, with odds shifting hour by hour as traders react to new information.
While most polling data isn’t instantaneous, the increased delays stem from electoral polling regulations introduced in Colombia last year. The rules effectively impose a gagging order on polls until the National Electoral Council (CNE) has reviewed their methodology and respondent metadata.
As a result, the publication of voter intention polling has slowed considerably, often lagging by more than a week.
Spanish polling and consulting firm GAD3 has suspended polling operations on Colombia’s presidential elections, arguing that the new requirements make reliable research impractical.
In a statement, the company said it was “unfeasible to carry out social research with the minimum guarantees of rigor and operational viability.”
The prediction market phenomenon entering Latin American politics – as it has around the world – will be an interesting trend to watch.
For his part, Guzmán suspects that the prediction markets are more reactive than predictive, but understands that people need “a place where they can go to soothe their anxiety,” in the absence of more regular polling updates.
Featured image: Screenshot of Kalshi market for Colombia’s upcoming election taken on May 9.
Image credit: Kalshi
The post The prediction markets betting on Colombia’s upcoming election appeared first on Latin America Reports.