Thursday, December 31, 2020

Betting on the 2020 US presidential election *after* the election

A couple weeks after the 2020 US presidential election was held, Nate Silver tweeted several times about the irrationality of betting markets (for example PredictIt), since they still had a 10% chance of Trump wining various states.

Screenshot of tweet - text of tweet:  Political betting markets still give Trump a >10% chance, both nationally and in several states where *results have been certified*. As I've said before, it's a bit alarming there are so many delusional people out there that the market equilibrium is this detached from reality.  Tweet contains screenshot of betting market odds / map of US colored by odds of who wins each state.

Other versions of this tweet talked about "free money" and "money left on the table".  I decided to try it out for myself discovering along the way that Nate failed to take into account transaction costs and market limits/barriers that contribute substantially to maker distortion.  In other words, it's not solely as Nate says irrational people, but also other well know sources of market inefficiencies..

TL;DR I made a net profit of 6.8% in 1.3 months (before taxes, ~4.5% after taxes), from a gross profit of 13.8%.  The 13.8% is close to (slightly higher than) the headline-grabbing number about irrational people in the markets, but from this example you can see that 67% of that "irrationality" is wiped out by simple transaction fees. 

Here's a brief timeline of what I did, and how I discovered on the fly the transaction costs.  Yes, I know:  if I had RTFM beforehand... bold italics indicate transaction costs/barriers encountered.
  1. transferred in X dollars as investment to PredictIt (11/11/2020)
  2. Attempted to bet on most profitable markets - Georgia, Pennsylvania, Michigan - blocked because market had reached maximum number of traders
  3. Invested in next most profitable market (Arizona) - hit maximum allowed investment (Y)
  4. Repeated by investing in next most profitable markets (Wisconsin, Nevada)
  5. Later, Georgia market opened up, invested smaller remaining amount (~Y/2)
  6. Most of these markets closed around December 15, 2020, 
    1. NB:  however PredictIt delayed Wisconsin b/c of a Wisconsin Supreme Court Ruling
  7. Overall gross profit was 13.8%
  8. PredictIt takes 10% of winnings, so percent winnings was 12.4%
  9. Withdrew all money (12/21/2020), PredictIt takes 5% of all money withdrawn, so profit in hand is 6.8%
  10. I'm going to pay income taxes (around 33%) on the winnings, which makes profit after income tax 4.5%
Although it all worked out in the end, during the entire time my money was invested I was worried about the possibility of losing it all.  Perhaps this was unrealistic, but given the crazy political and social atmosphere it was hard to avoid feeling that way.

2 comments:

  1. I make pocket change on Intrade and there was definitely more unrealistic money than usual to be made there several weeks after the election. More even than could be explained by the combination of people wanting to cash out and irrational loyalty. I account for that (maybe you can expect a 0-2%) and yet someone (perhaps multiple sources) was giving money away there for about a week in order to keep the results skewed towards irrational Trumpism (we were seeing around 3-8% which is quite a lot of free money.)

    ReplyDelete
    Replies
    1. Interesting thanks for sharing the observation. How do you distinguish between the "irrational loyalty" and the source(s) giving away money keep the results skewed? I would have just put those in one bucket, but was there some tell-tale difference identifying the latter?

      Delete