US election day is tomorrow and if I Google search for “Trump vs Kamala polls”, I find the New York Times poll results:
Source: nytimes.com
It seems that Kamala Harris is currently the more likely candidate to win over Donald Trump (by a mere 1%). This seems odd to me, not because I am trying to inflict any bias onto you as my reader (because I am not) but because I recently checked the betting odds and the markets have a completely different idea.
What are betting odds and how are they different from the polls?
Regular polls rely on surveys given to the population to determine their preferences on candidates and positions on certain issues. They are then used to build a model to predict which candidate might win the election.
Betting odds on the other hand, are determined initially by bookmakers who may include polling data as well as a whole range of factors such as analysis and historical data. The odds are then influenced by the volume of people betting on a particular candidate.
Betting odds have predicted the next president with 74% accuracy since 1872 and 91% since 1980
I decided to dig into historical US election odds to see just how accurate they are. I found data from Sports Odds History (source) which provides US election odds from 1872 to 2020.
Here is a sample of the data after I load it into Excel:
I created 3 columns to count the number of times the betting favorite actually won. Here are the results:
| # times winner was favorite | # times winner was not favorite | Accuracy of Odds |
| 28 | 10 | 74% |
This means that since 1872, the betting market has predicted the next president with 74% accuracy.
But what if we look more recently? The rise of the internet has provided people with an extreme level of information that voters can use to make better decisions about who to not only vote for but put their money on. I am going to look exclusively at the last 40 years. The results are staggering:
| # times winner was favorite | # times winner was not favorite | Accuracy of Odds |
| 10 | 1 | 91% |
Woah. Crazy right? 91% accuracy at predicting the next president. You may be wondering who that 1 winner was who was not the betting favorite… Well, it was Donald Trump in 2016. Interestingly, the betting market predicted his 2020 loss against Joe Biden.
If we exclude Trump winning in 2016 (which many can argue was an anomaly), the betting accuracy of predicting the next president over the last 40 years would be 100%.
So who does the betting market think is going to win in 2024?
The answer: Trump.
If I had to bet on either Trump or Harris, regardless of any political views I may have, it would have to be Trump (but please, do not make any bets based on this information).
What percent chance does the market think Trump has at winning?
Well this is perhaps where it gets even more interesting (or strange, however you want to look at it).
If we look on RealClear Polling we can see that as of Nov 4th 2024, Trump is in the lead with a staggering 57.9% chance of winning.
What is even more staggering is that less than 1 week ago, Trump had a 63.9% chance of winning.
Some could argue that Kamala is closing in on him, but with less than a day until the election, I think Trump is going to be the winner. Let’s hope this post doesn’t come back to bite me and I hope you enjoyed reading.
This isn’t my usual sort of post as I am focused primarily on the AI niche, however as an AI & data professional, I love digging into the numbers – especially when it comes to an event that we won’t see for another 4 years. Have a nice day!
