'Jury is out' on whether betting markets are a better guide to election outcomes than poll-based models: researcher.
Political betting markets aren't new, but a too-close-to-call, high-stakes presidential election and wariness over traditional polling has led Wall Street investors to look to the shifting probabilities offered by wagering platforms for guidance.
Those platforms show an increasing probability of a victory by Donald Trump, the Republican nominee and former president, over Vice President Kamala Harris, the Democratic nominee. Investors have reacted accordingly, analysts said, with the data helping to drive bets on the stock market and sectors, while adding fuel to a bond-market selloff that's seen Treasury yields jump in October. Shifting probabilities have also been cited as a factor in the U.S. dollar's October rally.
Is that the smart thing for investors to do?
It's too soon to tell, though the 2024 election will offer an important test not only of how well betting markets perform relative to polls but also against each other, said Rajiv Sethi, an economics professor at Barnard College who has closely tracked the markets.
Sethi and fellow researchers used a "profitability test" to see how a hypothetical traders would have fared around the 2020 presidential election and the 2022 midterm congressional elections if they made trading decisions based on polling-based election models or a prediction market.
The results showed the virtual trader came out slightly ahead by trading according to the betting markets, but it was close, coming down to the outcomes in three tightly contested states, he said.
"I would say the jury is still out" when it comes to whether betting markets are a better guide than polling-based models, Sethi told MarketWatch, in a phone interview.
While polls have shown a recent increase in support for Donald Trump over Vice President Kamala Harris, national and crucial battleground-state surveys remain largely on a knife-edge. Betting markets, in contrast, have shown participants growing increasingly confident of a Trump victory alongside a rising probability of a Republican sweep of both the House and Senate.
On the eight platforms tracked by RealClearPolitics Trump's probability of victory ranged from 57% on Smarket to 62% on Polymarket. In late September, Harris had a solid lead on most markets.
Contrast that with the site's average of national polls, which showed a tie with Trump and Harris at 48.4% each. A polling-based election model produced by website 538 also showed a close race; run the model 100 times and Trump wins 53 times, while Harris wins 47 times.
Meanwhile, controversy has surrounded Polymarket, a crypto-based betting market closed to U.S.-based investors, which shows Trump with a 62% probability of victory. On Thursday, Polymarket acknowledged what many had suspected, that a single bettor was responsible for a number of very large wagers on a Trump victory. Polymarket said the trader was a French national making bets based on personal belief and that the platform had found no signs of manipulation.
Prediction-market startup Kalshi and popular investing platform Interactive Brokers $(IBKR)$ allow U.S. investors to place bets in prediction markets. A court ruling in September cleared the way.
Skeptics have argued that betting-market participants are often overwhelmingly male, Republican-leaning or, in the case of Polymarket and some other platforms, not even U.S. residents, and so don't reflect the broader electorate, meaning the signal should be viewed with suspicion. On the opposite side, some proponents argue that markets are clearly superior, reacting more quickly to developments than polls while providing a clear-eyed, unemotional assessment of the state of the race in the pure pursuit of profit.
Neither argument holds up well, Sethi said. Meanwhile, the 2024 election will provide some insights into what sort of market structure works best. Individual markets operate by different rules. Predictit, for example, limits individual traders to $850 wagers on any question, perhaps making it less prone to accusations of individual traders attempting to manipulate the market.
Frustrated poll watchers have learned to rely more on averages of surveys, such as those compiled by RealClearPolitics, the Economist and 538, in hopes that the collective weight will help cancel out biases and other distortions in individual polls.
Sethi cautions against using the same approach when it comes to betting markets.
While results from one poll won't influence another survey, betting markets offer traders the opportunity to arbitrage anomalies away. For example, a trader could buy a cheaper Harris contract on one exchange and sell a more expensive Harris contract on another. As a result, when one market moves sharply, it tends to pull the other markets along with it. As a result, markets "are somewhat aggregating each other," he said.
Betting on elections and other events is an age-old past time. U.S. elections have long been a popular wagering topic overseas. The Iowa Electronic Markets, an academic platform, opened in 1988.
In a 2008 paper, economists Paul W. Rhode of the University of Arizona and Koleman Strumpf of the University of Kansas noted that political betting markets were a feature of U.S. presidential elections from just after the Civil War until the 1940s.
Meanwhile, some analysts caution that the influence of the betting markets on asset prices may be overstated.
Tom Essaye, founder of Sevens Report Research, argued in a note earlier this week that it made little sense to tie the bulk of an October run-up in Treasury yields to concerns a Trump presidency would lead to a relatively larger fiscal deficit than a Harris presidency.
While Trump's plans are seen increasing the government's deficit, both candidates are seen significantly adding to the nation's debt load if their policy proposals are implemented.
"My point here is that neither candidate has a plan to address the deficit so while Trump could be worse for the deficit, it's unclear why the bond market would suddenly care so much about his deficit but not Harris's," he wrote.
Instead, it's much more likely the 10-year yield BX:TMUBMUSD10Y has risen from around 3.6% in mid-September to over 4.2% because economic growth and inflation have been stronger than expected, Essaye argued.
The S&P 500 SPX and Dow Jones Industrial Average DJIA snapped a streak of six straight weekly gains on Friday, though the Nasdaq Composite COMP managed a seventh straight weekly rise. Stocks have largely defied pre-election seasonality.
Long-term investors will and probably should look past near-term election noise, analysts say. As far as the signal from the betting markets, investors can take their choice.
As MarketWatch reported earlier this week, Fundstrat's Tom Lee argued in a note that the important thing is that "it seems like the betting market movements are impacting equity markets. And in this sense, it probably makes sense for investors to pay heed on this."
It's possible that betting markets are biased, he acknowledged. "This could be true, and as such, if one believes this is the case, then one can simply bet the opposite whether in equity markets, or on the betting markets themselves," he said.
Comments