The election of Donald Trump as the 47th President of the United States has dealt yet another blow to the reputation of the American polling industry. The polling average had Harris beating Trump by a point nationally, and winning or tied in virtually every swing state. Until the eleventh hour, the polling consensus maintained that Harris was winning in each of the key ‘Blue Wall’ states of Michigan, Wisconsin and Pennsylvania. Within minutes of the first election day polls closing, it became clear they had got it badly wrong.

President Trump won all seven swing states – a bigger electoral college victory than he secured in 2016. As of the time of writing, he commands a decisive lead in the popular vote, with more than four million votes than his Democrat rival. And with votes still coming in, his party has secured a majority in The Senate and looks bolted on to win in the House of Representatives.

But even that doesn’t tell the full story. Trump came within just 200,000 votes of taking the Democrat stronghold of New Jersey and within 150,000 votes of winning deep blue Minnesota, the home state of Vice President Harris’ running mate, Tim Walz. With results coming thick and fast on election night and Democrat hopes of securing a path to victory fading, CNN analysis revealed the Vice President had at that time failed to outperform her predecessor by three points or more in a single county anywhere in the country.

The famous American pollster Frank Luntz, who had called the election for Harris on Piers Morgan’s Uncensored, returned to the programme on November 6 to admit he had got it wrong, saying ‘half the country are mad as hell and they aren’t going to take it anymore.’ He also pointed the finger at colleagues, labelling Ann Selzer’s Iowa poll, which ignited enthusiasm among Democrat ranks when it predicted Harris would win the red state, as ‘the worst survey of all.’

Those who were looking at the polls for an insight into the mood of America would not have come away with much in the way of useful information. But were the markets any better at predicting the outcome?

How did the polls get it so wrong?

Polling is a complex science. Although prima facie it simply involves contacting potential voters and asking them who they support, there are a lot of complications. Traditionally pollsters have contacted voters on their landline telephones. But in 2024 only a quarter of Americans own a landline telephone, and those who do will skew toward an older demographic. So, any poll carried out with this method will fail to register the support of younger cohorts.

In September, pollster Matt Taglia told The Hill that Emerson polls use a combination of telephone calls, text messages and online panels to try and capture a broad range of views. But this still poses problems for pollsters because some groups are more likely to respond to surveys than others – and even when they do, they might not be telling the truth.

The so-called ‘shy Trump voter’ is thought to account for as much as a 4 per cent polling error. To remedy this, some pollsters use automated messages to contact voters, believing they will be more likely to share their authentic views with a dispassionate machine than a potentially judgmental human. Pollsters must also grapple with the challenge of reaching different demographics at different times of the day. People in work, who are more likely to vote Republican, are less likely to respond to polls, let alone daytime landline polls. Conversely, young college educated women who are enthusiastic about voting Democrat may be keen to complete an online poll expressing their support.

So polling is a challenge – but it is a challenge pollsters are paid not insubstantial sums to overcome. And for the third general election running, they appear to be no closer to squaring the circle in the post-landline age.

Did the stock market predict the US election?

The stock market did a better job of predicting the US election than the opinion polls, but it didn’t predict the scale and surety of the Trump win. In early October deVere reported that the markets believed the race was far closer than the polls suggested, with investors betting on assets and stocks likely to benefit from a Trump Presidency.

While the so-called Trump Trade was riding high over the summer, it experienced a drop-off after Kamala Harris replaced Joe Biden as the Democrat nominee. She was said by pundits to be enjoying a honeymoon, and pollsters delivered one survey after another appearing to show she was building a commanding lead over her Republican rival.

It is not clear whether those honeymoon polls were accurate snapshots of a moment in time which has since faded, or were just as wrong as October and November’s polls. But it does appear that right or wrong they impacted market sentiment, causing a drop off in Trump Trade – but not a total halt. As deVere said at the time:

“Investors are betting on an ‘extremely close’ election – defying the polling consensus which has Harris up three points on Trump nationally.”

There was also some confusion among observers of what they should read into stock market movements. For example, Sam Stovall at CFRA Research told CBS that incumbents tend to win when the stock market advances in the run-up to the election. He told the outlet:

“I believe we will see a Harris victory ultimately, because I’m a very big believer in history and rules-based investing.”

However, it now appears more likely the stock market gains were made not despite Trump – but because of him – with investors continuing to believe his presidency will fuel economic growth and prosperity – and that he was likely to win and deliver it.

In short, the runes were there to be read, but not everybody drew the right lessons from the movement of the markets.

How the betting markets predicted a Trump victory

The real Nostradamus in this story isn’t the stock market – but the betting markets, which had Trump as a strong favourite and hardly wavered. The betting platform Polymarket, where millions of dollars were being wagered on the outcome of the election, had Trump at an almost 60 per cent chance to win the election on the eve of poll. Speaking to CNN, professor Eric Zitewitx explained why betters are often a useful weather vane for public sentiment:

“Financial markets are generally pretty efficient, and the evidence suggests that the same is true of prediction markets.

“There’s no virtue-signalling in an anonymous market when you’re betting.”

One French trader is reportedly set to win $50 million from an enormous bet on Polymarket, which he placed after conducting his own polls – with a canny twist. The trader, known as Theo bet that Trump would win the blue wall and the popular vote – a bold move given that no Republican has managed the latter feat since George Bush in 2004.

However, his confidence was boosted by a series of polls he commissioned himself. Unlike many other polls, the survey did not ask respondents who they were going to vote for and instead asked who they thought their neighbour was going to vote for. The results pointed toward an overwhelming Trump victory.

The former banker, who remains anonymous, spoke to the Wall Street Journal four days before the election and said “I know a lot of Americans who would vote for Trump without telling you that”, adding that he was not satisfied polling companies had done enough to correct their previous mistakes in 2016 and 2020.

He is now recommending that US pollsters should rely on the neighbour question for future surveys to secure more accurate responses.

Money talks

American pollsters continue to get it wrong. While the flawed nature of polling and the bias of pundits suggested a knife-edge race or a Harris win, financial and betting markets were closer to the mark.

Since 1928, the stock market has correctly anticipated the election winner approximately 87 per cent of the time. This is based on the trend that if the S&P 500 performs positively in the months leading up to Election Day, the incumbent party tends to retain the White House; a decline, conversely, has often signalled a party switch. In 2016, for example, the S&P 500’s fall predicted the political shift as Donald Trump took office, despite forecasts pointing to a Democratic win.

However, this metric became slightly confused in 2024, when the chance of a Trump victory was actually driving the stock market higher. This might be because, as the first President to serve two non-consecutive terms in more than a century, investors were clear about what Trump would do once in office, him already having served a term. Typically, oppositional figures can bring volatility and worries for investors, whereas the increasing likelihood of a Trump win was boosting investor confidence, and turning the conventional wisdom about a rising S&P 500 being good for the incumbent on its head.

This year, while many betting markets pointed to Trump, the stock market’s recent trends suggested a closer race – and the polls got it completely wrong. The stock market can be a valuable tool for gauging public sentiment and economic optimism, especially when traditional polling methods struggle with biases or structural challenges, such as the “shy voter” effect. Ultimately when people are wagering their own money, they will do so based on cold hard calculation, rather than with the ideological skew of a pundit.