Presidential Polls And The Stock Market The Results Might Surprise You
Over the past few weeks I have been hearing more and more people say things like “the market is up because Trump is narrowing the gap to win” or that “the market is proving that it likes Biden” and other similar sentiments. Lucky for us there is a lot of data we can look out to find what the market really does think about either candidate. The results might surprise you.
I first went to the king of polls and election models 538.com. As you might recall Nate Silver runs the site and last election had HRC winning with a 71.4% chance of winning and DJT with a 28.6% chance if winning. A lot of people post-election said “Your model sucked” or “you couldn’t have been more wrong” when in reality the odds are the odds and the 538 model said DJT had a 1:3.5 chance of winning. The election was very close, he lost the popular vote, and he won the electoral vote. 1:3.5 is not the same as 1:1,000,000 or even 1:100. All this to say that the polls, the betting markets, and other similar datasets can actually be quite telling.
But how does the data and the SP500 interact? We first took the polls and made a Biden-Trump spread. We looked at the spread with the SP500, we looked at each candidates individual polling series, and then we tried different offsets and measured correlations. Guess what we found? Using this data the market seems to either like both candidates equally or the market doesn’t care about either candidate equally. There was no relationship that lasted more than a few weeks at a time before the correlation blew up and went the other way.
I then went to the 538 Trump Approval and Disapproval dataset. This one just looks at how good people think Trump is doing as the president. Looking at the spread we see that in 2018 the it was negatively correlated to the approval ratings and then in the shutdown it was positively correlated and then it went negative again. In essence it proved that if you get two data series that move around you can get them to line up occasionally…in other words there was nothing there.
I then looked at each data series separately. We found very little value here. The closest thing we found to value was that if you shift the market forward by 33 days on the approval chart you can line up two different peaks. But we couldn’t get a R2 value of higher than .15 on any data manipulation we did. So there might be a tiny bit of value in that. If the country thinks the President is doing well then the market goes up a bit better than usual. I don’t know but it sounds plausible.
Next I went to the betting markets. We have two places where you can get the data. Predictit.org is the best one, if only because you can download the data in an CSV file. The other is RCP-Real Clear Politics where they average several betting sites data. We had to manually input that data and it was a pain. But alas I love data so I did it.
Anyways how did these do? They are “prediction markets” after all. Well they might do a great job of picking the Pres but what does that have to do with markets?
Turns out the spread has nothing to do with the stock market. The spread went up and then it went down, the stock market went up and then it went up some more. Either it loved one candidate and switched to loving the other candidate or it just didn’t really care.
What about the individual contracts? Well as you can see the Trump contract has a decent fit…but in reality it only starting fitting the stock market once his numbers started to improve. Before that they had a negative correlation. So it looks like once again it was statistical noise. Looking at Biden it was a good relationship, then a bad one, and lately it has been a good one once again. Interpret that as you want but I will interpret it to mean absolutely nothing. Like I did with all these data sets I offset the data to try and fit it, did different rolling correlations, did some regressions and found roughly no real relationship between the market and either candidate.
Moving on to the RCP-Real Clear Politics betting data you can guess how well any statistical scrutiny ended up between the stock market and either candidate. If you guessed that there is no signal in this noise then you are correct. Both candidates have had their good and their bad weeks where they went from positively correlated to negatively correlated and back. It is all noise.
I then made the RCP Biden-Trump spread. Unless you are hallucinating there is no relationship here worth betting on. The stock market has been going up so when the spread goes up it matches and when it goes down it does not match. Offsets did not help. Rolling correlations go up and down. There is no signal in this noise either.
Finally we looked at…just joking that was it. We went through four different data sets, did all kinds of data manipulations, played with the statistical toolbox, and found that there is currently no signal between the stock market and the current candidates poll or betting data. This could change in the future as we still have two months to go. Or maybe RenTech can figure out a way to trade this. But I found nothing. It was interesting to look at the data but it provided no real clues in relation to the stock market. So next time someone on TV says “the market is up on DJT/Bidens poll numbers” or some other similar line, just remember that they are probably wrong in the same way that they are wrong when they say “today the market went down because there were more sellers than buyers.” It means nothing.
With all that said go vote. Do me a favor and DO NOT vote straight ticket. If you love or hate the President there is a very low chance that you still feel the same way about every single person on your ballot. A lot of important policy happens at the state, county, and city level. Vote for each candidate individually. If enough people do this they might find that walking across the aisle is not that hard.
Happy Trading and Be Safe,
P.S. If you liked this then take a free two week trial of our service. If you have any questions send me an email or find me over at Twitter @DavidTaggart