FTX recently launched 2020 presidential election markets; TRUMP, for instance, is a future that expires to 1 if Donald Trump wins the 2020 presidential general election, and 0 otherwise.

How should you think about these markets, and what makes them different from others?

Comparison of markets

Well, to first order, the expected value of TRUMP is equal to the odds he wins the election. So if you think there’s an 80% chance he’ll win reelection that would imply that it should trade at $0.80. It’s currently trading at around $0.62, implying a 62% chance of winning; so you might want to get long if you think the odds are 80%, and short if you think the odds are 20%.

This thinking of course ignores interest rates, inefficiencies, fees, spread, etc. but one of the advantages of the FTX markets is that it’s pretty easy to trade based on the odds you think someone will be elected.

How does that compare to other markets? High level it’s similar to the consensus, though slightly more bullish on Trump; other markets have him trading around 58% as of February 17th. But as you drill in you can see a lot of design differences between the venues.

First of all, FTX presents election markets similarly to general financial markets: an orderbook with a price, sizes, and trade histories. You can quickly see that you can buy Trump at 62.5% or sell him at 62.1%; similarly you can see that Bloomberg has traded 160,000 contracts today, with the last trade happening at 11% implied odds. If you look at betfair, one of the highest profile betting venues in the world, you’ll see quite a different layout:


While this might be intuitive to people used to betting lines, for those coming from other areas it’s hard to understand. If you want to bet on Trump, you’d have to think his odds of winning are…. 1.69? As it turns out the right way to interpret these is 1/X–so 59.1%. Which is fine if that’s what you’re used to! But if you’re used to trading basically anything else, it’s way easier to interpret markets phrased in terms of the probability that something happens.

Another issue is fees. Predictit, the most popular US prediction market, charges 10% of all winnings, along with 5% of all withdrawals. That means that they’re charging $0.05 - $0.1 on every contract, along with 5% when you withdraw; that’s over 10 times as high as FTX’s fees! Betfair charges 5-7% on all winnings, once again way higher than FTX.

Then there’s liquidity. FTX is generally known for its liquidity and the election markets are no different. You can buy or sell roughly 80,000 TRUMP contracts within $0.02 of the market right now, that’s about five times what you see on Predictit and Betfair:


But it gets worse; Predictit, in fact, limits you to just 800 contracts per market! FTX has no limit.

Finally, there’s volume. Betfair has traded roughly 16m since inception; FTX has traded roughly 4m in the last week!


So what trades are there to do?

The most obvious trade is to buy underpriced candidates and sell overpriced ones. Right now Biden is trading around 4% to win the election. If you think he’s totally done, maybe sell; if you think his drop from over twice that a few weeks ago makes no sense, maybe it’s time to scoop some up!

There are also arbitrages. Trump is trading at $0.62 on FTX, but around $0.05 lower elsewhere; you could sell Trump on FTX and buy him on Betfair or Predictit (though watch out for their fees!). Or you could collect some $0.04 Biden on FTX and sell him higher on Predictit or Betfair.

Finally, you could try to play the news. Think Bloomberg is going to rally on Super Tuesday? Buy before!

Put your money where your mouth is

Above all, prediction markets give you the opportunity to profit if you’re correct–and this in turn lets others turn to them as an indication of what might happen.

Sam Bankman-Fried