Severe dislocation between Prediction Model Odds and Betting Market Odds in the 2020 US Presidential Election
Is there a Positive Expected Value bet in the market right now?
It’s been a while since I engaged in any type of betting, but something caught my eye over the weekend while doing my usual reading and perusing PredictIt.org, a website that allows American citizens to wager small amounts on US Politics outcomes.
In PredictIt, unlike traditional betting websites, the odds of betting are shown in cents. That makes it easy to calculate percentages: For example Biden to win the election at 65 cents means PredictIt market is pricing a 65% chance of Biden winning.
I decided to look at what election models were predicting for the 2020 election.
Here’s what I found:
As of October 20, FiveThirtyEight gives Biden an 87% chance of winning.
As of October 20, The Economist gives Biden a 92% chance of winning.
I couldn’t believe it. I could bet something at an implied probability of 65% when in reality the actual probability is closer to 87%. That kind of dislocation has to be rare!
I then decided to compare the Betting Odds of Hillary Clinton winning the 2016 election vs FiveThirtyEight 2016 election forecast. After converting the betting odds of Hillary Clinton winning to their implied probabilities, here’s what I found:
August 18, 2016
Betting Market probability of winning: 81.82%
FiveThirtyEight prediction model probability of winning: 86.40%
September 26, 2016*
Betting Market probability of winning: 68.25%
FiveThirtyEight prediction model probability of winning: 54.80%
*This was debate night that year
October 25, 2016
Betting Market probability of winning: 84.62%
FiveThirtyEight prediction model probability of winning: 85.00%
November 1, 2016
Betting Market odds of winning: 73.33%
FiveThirtyEight prediction model odds of winning: 71.20%
I can conclude that in 2016 betting markets and prediction markets were largely inline.
What are some potential causes of the dislocation 14 days before the 2020 election?
Someone has some insider information about the election and is betting on Trump, keeping Trump’s price high.
Trump supporters are overconfident and/or believe polls are ‘fake’, thus flooding the betting market on Trump’s side.
Biden supporters are concerned about the polls/prediction models being wrong again, as they were in 2016, thus they are staying in the sidelines out of fear of losing again.
I find cause 1 to be highly improbable. Someone with insider information would be betting large amounts, which would be hard to hide. Furthermore, why wouldn’t that person allow the betting market to converge with the prediction market, reducing the cost of betting on Trump, thus increasing the return of investment of that bet?
I am leaning towards the probability that it is combination of cause 2 and 3. It is no secret that there is fanaticism in Trump’s base, and that could be affecting the market. The Biden/Democratic base is still reeling from the 2016 election results and they likely don’t trust prediction models as much, hence why they might be sitting on the sidelines.
I have no reason to believe prediction models are wrong this time. I trust that statisticians like Nate Silver learned from their mistakes in the 2016 election and adjusted their models accordingly. This was reflected in his 2018 Midterm Elections Forecast Results.
I have concluded with a high degree of confidence that at current odds there is positive expected value in betting that Joe Biden will win the 2020 Presidential Election.
Is this free money? Absolutely not. Even if you believe the most favorable prediction models, there is still a 8% probability, or roughly 1 in 10 chance of Donald Trump winning the election. There is a possibility that I lose every cent I wager.
So what is the point of betting? I will use an analogy to explain:
A fair coin has a 50% chance of landing heads. A well priced betting market should price the coin flip bet at even odds (50% implied probability), meaning that for every $100 you bet on heads, when it lands on heads you win $100.
What if I proposed to you the following: If you bet heads and it lands heads, I will pay you $100, but if you bet tails and it lands tails, I will pay you $150.
What should you bet?
Tails. Every single time. This does not mean that you will win every time. The coin will land heads 50% of the time and you will lose. But if you keep betting tails and I keep flipping, you will get ahead. Betting tails is a positive expected value bet. Same with betting Joe Biden at current odds.
The amount I am wagering is an amount I am comfortable to lose. I also want to point out that there is no partisanship in this bet: I am just taking advantage of a Positive Expected Value situation. If the names were switched, I will still take advantage of this opportunity. In fact, if we assume prediction markets stay the same and betting markets move to converge with prediction markets, I will start betting on Donald Trump to win the 2020 US Presidential Election.
Why would I do that? I would be hedging and locking in gains.
Let me explain using real numbers.
Because I am Canadian, I am not allowed to wager on PredictIt, so I am using Bodog to place my bets.
On Bodog, the odds of Biden winning the election are -170. The implied probability at those odds is 62.96%, even better than PredictIt. Sweet.
Odds at -170 means that for every $100 I wager, I can win $58.82.
Lets say that tomorrow the odds at Bodog move to reflect the prediction model forecast of FiveThirtyEight. With Biden at 87% probability of winning and Trump at 12%, odds on Donald Trump winning would be +730.
Odds at +730 means that for every $100 I wager, I can win $730.
For every $100 I wagered today on Biden winning at -170 odds, I would wager $19 on Trump winning at +730 odds tomorrow. This guarantees me a profit of ~$40 for every $119 I wager, as long as either Biden or Trump wins. That’s a 33.61% return in 14 days.
In conclusion, there seems to be a rare opportunity to place a wager on a Positive Expected Value outcome. It makes sense to take advantage of this opportunity. Regardless of the outcome, I feel confident that my decision to bet on Joe Biden to win the 2020 US Presidential Election at current odds is rational.