Betfair and the 2016 Presidential Election
Betfair is the largest (by far) betting exchange in the world. It operates most of its business outside of the USA due to legal restrictions on betting, but the US Presidential election is a big enough worldwide issue that nearly $130M USD has been wagered on the question of who will win it.
[While we are at this, a request: please vote, and vote against the Trump and all he stands for. Ideally vote for Hillary, especially if you are in a swing state.]
Many people looks at the odds in these markets to make a prediction about what will occur on Tuesday. Polls are based on calling people and asking what they would do. Predictors such as 538 try to translate this and other data into a prediction of what will occur. People who like to heavily weight a betting exchange like to do so because the participants have real skin in the game: they believe Nate Silver likes to call a “close election” because a landslide yields no page views or relevance for him. But if you have to put your money where your mouth is, it really truly is what you believe will occur.
I was a founder of Betfair and on the board of the company for 14 years, so I am intimately familiar with the marketplace, though not directly involved today. Because Betfair is transparent about odds and the terms of every bet wagered in every market, it is free from manipulation and great for understanding the market.
Much like the pollsters, the bettors who drive the real money volume on Betfair do so based on elaborate models and the best statistical thinking they can bring, and are able to win consistently in markets which they are knowledgeable. Consequently, I am willing to place a lot of faith in what the markets say.
So, what do the markets say?
Here is the status of the Betfair market for the 2016 US presidential election today, three days before election day:
What does this market tell you? Look at the blue: It says if you bet on Hillary Clinton and she wins, you will receive 1.29 times your money (0.29 of profit plus your 1.0 stake.) The number underneath tells you here is committed capital to support a bet of up to $11,235 at this instant. If you bet on Donald Trump to win you will receive 4.4x your money if he wins, and there is $20,895 available. Also note nearly $130M has now been matched on this event. The pink section are the odds you get to take the other side — betting Hillary or Donald to lose, not very interesting when down to two but a uniquely useful feature of Betfair six months ago when we had a number of other candidates.
So we know the market thinks Hillary is more likely to win (because the market must pay you more to get you to bet on Trump), but without doing a bunch of math, it’s hard to estimate how much more likely.
You can get some more useful information by clicking on the little graph icon next to the candidate’s name. This opens up the history of money wagered in the market on each participant over time. This graph shows you how the price has moved for Hillary Clinton over the entire history of the market. If you click on Hillary Clinton, you can pull down and see the same info for Trump, Sanders or who ever has been in the market.
The key button for the casual viewer is to check the Inverse Axis button. Now the Y axis on the left has Betfair doing the math for you. The graph tells you “Implied % chance of winning” — just what you wanted to know! We can see the implied chance of winning for both Clinton and Trump.
Currently, the market is saying Clinton has a 78% chance of winning on Tuesday, and Trump has a 22% chance of winning on Tuesday. As I look at 538, it presently says:
So the betting markets are good news if you (like me) want to see The Donald lose. Smart money thinks Hillary is strongly positioned for Tuesday.
The Mistake EVERYONE Makes
People think a market like the current Betfair one is equivalent to predicting that Hillary will win. If you do this, don’t worry, nearly every single person I discuss prediction markets with makes this mistake, but let me try to clarify this for you.
As an example, I had a conversation with my Dad this morning, when I brought up the odds in the current Betfair market. His reply was: “Betfair was wrong about Brexit, the odds the day before were 68% chance of Brexit not passing but it did. Do you think it’s really more right about Clinton/Trump?”
This is the mistake I am discussing: to think the market was “wrong” if it says something is >50% likely, but doesn’t happen. Betfair said the day before Brexit there was a 32% chance of it passing. If the market was correct, there was a very good probability of that happening (and it did). Unless the market says Clinton has a 99%+ chance of Trump winning, the market is NOT saying “Clinton will win” and the market is not “wrong” if Trump wins. You would have to look across a wide number of events and how often favorites came in vs. longshots to form a conclusion on whether the market tends to be right or not. And trust me, in large liquid markets like this, a majority of the money is smart money that has done its homework, and has more riding on finding the “truth” than just about anyone else commenting on the state of the election.
Ideally, we would see markets converge to the truth (and thus converge towards 100% to the winning side as the event approaches.) This does happen not linearly with time, but with the rate of new information being released. Imagine a market on a coin flip: the odds would be 50/50 and stay there, until the result was revealed — then it would jump to 100%.
In this election, the bettors cannot possibly know the “truth” — no one does, yet. As more information is revealed, the market will converge to the answer. In some events, such as a live market on a soccer game, the market will converge to certainty as time ticks away and goals are scored — when it’s 2–0 and 5 minutes left, the winner will be >90% likely. Elections are tougher — we won’t get much concrete new information until early poll information comes out on election day, and the results in swing states will move the market a lot. But that data historically has been super volatile, and the market often jumps around a lot. Watching it during Bush / Gore, early polling went for Gore and the markets got up to 80% certainty he would win, but ultimately we know he did not.
Watching the market live on election day can be a great way to see the information being released translated into market pricing — keep it open in a browser window as you follow the news!
How to View the Market
Here is a link to the US Presidential market. Note it may not work as well on a mobile browser.
