It’s December 2020, a month after the election. Not only has Joe Biden cinched the popular vote, but also the electoral college. Why are prediction markets still giving President Donald Trump a 10% chance of being inaugurated for a second term in 2021?
Not Just a Wording Artifact
One theory to first debunk about prediction markets is that it’s a wording problem. For example, the PredictIt market doesn’t define in clear terms what a winner is. It gives gives Trump about a 10% chances. Yet other markets like FTX and PolyMarket have a clear definition. The 2020 winner is the person who appears on the whitehouse.gov website as president on February 1st, 2020. Those markets still trade around 10%.
Prediction Markets are Surprisingly Consistent with Each Other
Another possibility is that it’s due to wild market inefficiencies, or limitations of a market. For example, FTX markets towards Europeans and so might miss American opinions. But PredictIt allows US opinions in a relatively accessible way. Yet some might complain PredictIt has both user limits and dollar limits (just a measly $1,000). PolyMarket, on the Ethereum blockchain has no user count or dollar limit, and is anonymous to boot. All three markets agree!
So why are prediction markets still giving Trump a 10% chance today? The answer has lessons for prediction markets across the world.
Trump Buyers are Bidding Up Prediction Markets for Narrative Utility
Ther are two effects. First, a number of Trump buyers receive utility beyond the objective chances of Trump winning. This “cheerleader” reason why people buy alumni sweatshirt at prices way higher than generic. They’re not making an investment, as much as showing support.
After all, Trump supports have donated hundreds of millions of dollars to Trump after the campaign. No one criticizes political donors for making poor investments: because investments is the wrong lens to view things through.
It’s not a surprise then that there are at least a few million dollars of Trump cheerleaders in the market. And that’s all you need — just a few million dollars. Because each $1 of Trump buyer there is, you need $9 of Trump sellers to push back against them.
Limits to Arbitrage
The second effect is that there isn’t a ton of money selling Trump to neutralize the cheerleaders. This is known as limits to arbitrage.
With the above few million dollars of Trump buyers, you need a few dozen million dollars of Trump sellers. Reasons for this limitation include:
- Regulatory constraints. Polymarket allows betting anyone to bet. This is not sufficient to attract large financial bettors in jurisdictions where betting is a grey space.
- Fees. PredictIt takes a 5% fee on withdrawal. This means if the market is trading at 90%, even if you are sure of a 100% resolution, you’ll make only 5% net.
- Taxes. This can be a big one depending on your tax laws. If you put in money at 70% and the market resolves at 100%, you get taxed on the difference, grossing you only 90%. But if you lose, you may not be able to deduct your losses, meaning you get 0%. Thus your threshold for betting is 1/3 higher!
- System risk. What if PredictIt goes bankrupt? What if your government bans Polymarket?
- Time needed to make the bet. This is pure overhead.
Ay one of the above may not be enough to make a difference. The totality of the above stops most dollar volume from entering the market.