Want Timely, Accurate Election Predictions? Look To The Betting Markets, Not Polls Or Pundits

Wednesday, October 24, 2012

Are you putting election results into your investment thinking? Betting markets provide excellent insights that are better than polls and pundits. Here’s my experience‚Ķ

Many years ago, I read (I believe in The Wall Street Journal) about betting’s accuracy in making election predictions. It turns out that, once again, watching what people are doing with their money is better than listening to what they are saying.

So, I began watching online betting to evaluate elections beginning in the early 2000s. The results were uncanny. Winners were foretold and margins of victory were anticipated. Additionally, the betting revealed trends and forecasts quicker and better than polls and pundits.

Where’s the proof?

A good academic piece is “How to Forecast an Election (And How To Win One!)” by a unique expert:

Leighton Vaughan Williams is director of the Betting Research Unit and director of the Political Forecasting Unit at Nottingham Business School, Nottingham Trent University.

In his write-up, he explains the theory behind betting markets and provides relevant examples of their usefulness.

The accuracy he describes matches my experience. In the 2008 election, the betting market not only had Obama winning, but also correctly forecast the results of every state. In fact, the betting results foretold not only each swing state’s final choice (Obama or McCain), but also how close the voting would be.

Betting market’s process provides the value

Before looking at the results, let’s discuss how the betting gets converted into usable information. Intrade.com, a major provider of betting markets, has a fulsome explanation in “How it works.” Here is a summary description, using the Obama/Romney upcoming election as an example:

Three yes/no contracts exist: (1) Obama will win, (2) Romney will win and (3) Someone else will win. Note that the contracts are all-inclusive, meaning 100% of the possibilities are covered, and that only one contract will be true – i.e., “win.” At expiration, the correct contract will be worth $10.00 and the others $0.00. Thus, through arbitrage, the sum of the three contracts is always close to $10.00 Importantly, these amounts can be converted easily to odds (or, more familiar to us, probabilities), ranging from $10.00 = 100% to $5.00 = 50% to $0.00 = 0% (and all points in between).

Real-time data quickly measures effects, like Tuesday’s debate

Instead of polling’s delay from question to compilation to publication, betting’s 24-7 markets ensure that the latest results are relevant. So, want to know who “won” last night’s debate? Ignore the talking heads, and focus on the betting market’s latest changes. Those results not only indicate the winner, but also the extent of the win.

Below are Obama’s and Romney’s election probabilities since the beginning of August, showing the effects of the conventions, the debates and the ongoing campaigning. (Note that they are mirror images, always adding up to about 100%, because the third contract for someone else being elected is virtually 0% probability.)

Example of betting market’s superiority

The Wall Street Journal’s front-page article Monday was “Dead Heat for Romney, Obama,” explaining “WSJ/NBC Poll Shows GOP Contender Gains.” The poll results: Obama 47%, Romney 47%.However, today’s betting has Obama at 57% and Romney at 43%.

Why the discrepancy? On the surface, there are the usual polling problems of small sample (816 likely voters), “old” data (Oct. 17-20 polling) and NAs (Obama’s 47% + Romney’s 47% = 94% total). Also, polls lack state-by-state (electoral college) vote distribution analysis. Most importantly, polls are number compilations only and lack analytical insights.

Betting markets’ bonus: More information available

Besides addressing the big question of who will be president, the betting markets have sliced and diced the election into meaningful components. For example, there is a betting market for each state’s outcome. While many fall into the near-100% camp (i.e., strongly Republican or Democratic), the important swing state results provide the foundation for the overall election. Here are the swing states currently, showing both polling and betting market information. The results are sorted by the betting market’s Obama/Romney “spread”, highlighting the truly close races.

Additionally, there are markets for senator and governor elections as well as which party will control the Senate (current probabilities are 64% Democratic, 25% Republican, 11% neither) and the House (96% Republican).

The bottom line

The best way to factor elections into your investment thinking is to watch the betting markets, not the polls or pundits. Through a trading market, we get a sound measure of the expected outcome for a specific event. History has shown that this process gives a better measure of probabilities than other approaches. Moreover, as time passes, probability changes give us data relating to relevant events and trends.

With only two weeks to go, the betting markets are telling us that we should expect a post-election U.S. government composed of President Obama, a Democratic-majority Senate and a Republican-majority House. Certainly these outlooks can change, and the best way to stay on top of developments is to watch the betting markets.

Leave a Reply

You must be logged in to post a comment.


John Tobey on Seeking Alpha

Seeking Alpha Certified

October 2012