What are the odds of Trump?
What are the odds of Trump?
Tracking the prediction markets before the big debate
Apparently there’s a presidential debate tomorrow.
Or as they say, “I’m hearing this now for the first time”…
I don’t have a dog in this fight, not really. We’ll get to politics in another ramble soon, don’t worry. Today’s ramble is going to be laser focused on how you, as an investor, business owner, or human, might observe what prediction markets are saying about these outcomes. Up to and including how to source and test strategies related to the changes in the odds of various candidates.
TL;DR: How can you track the presidential odds and apply that to trading markets?
If you ask the best investors in the world whether they bet on politics, they will usually give you some version of this refrain:
We’re not political experts, we’re experts on markets. We try to focus on good opportunities in markets, and be agnostic to political outcomes.
You will hear this even from the type of investors for whom this is most obviously not true: global macro.
Why?
Well, when you invest in Apple or Microsoft, your investment may be impacted by the outcomes of political machines, but not in any predictable way and not with sufficient impact or velocity to really be worth trading against relative to things like their market, their product, their talent etc. (You know, all the things we want investors to care about rather than punting Trump vs Harris.)
It would be kind of weird if people were betting on Microsoft vs. Apple primarily based on the outcome of a Presidential election, right?
In general, you don’t want to make investment decisions based on politics, and you don’t want to do invest in people who do.
This principle faces some cognitive dissonance when you invest in individual shares or securities sold by quasi-governmental organizations. Sometimes obviously and directly (hi Freddie Mae, PEMEX, or PDVSA) and sometimes indirectly (hi Sperbank).
This principle gets stretched even further, to the breaking point I would argue, when you get to an investment strategy called “global macro” where in a lot of case your counterparty (aka the person you bought the security from) literally is the government. As you start to bet not just on economic and financial outcomes but also on economic and financial policy.
When you hear people betting on the ‘shape of the curve’ this is what they are doing, trying to compare what’s going on in the world to what some chart says that the market for interest rates looks like. These rates then impact currency markets and risk premiums, which in turn impact local economic and financial markets. Which create outcomes, which you can also bet on. Whether it be directly on something like expected amount of inflation, or indirectly like the relative performance of bonds vs stocks vs gold etc.
If your fund has 40% invested in the debt of a small emerging market government facing an election that could lead to debt repudiation, understanding those politics suddenly becomes your fiduciary duty. (I myself haven’t been in this situation, but have multiple friends who have. Faced with a prospect of killing their pod/firm/fund on a random political outcome, they become experts pretty quick.)
Most macro investors deal with this dissonance by trying to separate the ‘politics’ from the ‘policy.’ It’s often enough to just stare at the what IS and figure out where the mispricing’s are, rather than get all worked up trying to predict the future.
This is a deep lesson from a former mentor.
Anyway, so if politics is part of investing but not something you should focus on, what should you do?!?
Well, my friend, you've come to the right place. The lesson, from the books, and Lehman, and the demi-gods of macro, looking on all of us from atop Olympus was simple:
Don’t try to predict important things you don’t understand.
Hedge them.
Spend time on your alpha and hedge out everything else.
What does that mean for little old me and you?
Well the first step to trying to hedging something is to observe and measure it. To go out and try to find something in that deep and weird external world which actually gives you a read on “how likely is this thing to happen. Some source of what some call ‘triangulation with the market.’
Aka the first step is to find the odds.
I’d recommend to start with a prediction market like Polymarket, Manifold, PredictIt. Start with the ‘wisdom of the crowds’ and try to find out what the market thinks.
Manifold currently thinks it’s a dead heat.
As does PredictIt.
These markets say it’s an even odds kind of thing.
Though notably polymarket, where the outcome of the markets is for money puts Trump 5pts higher.
Then we look for experts. Let's start with three: Nate Silver, his former organization FiveThirtyEight, and a chatbot designed to trade prediction markets. Silver gives Trump a close to 2/3 chance.
His old shop, no longer providing predictions, but rather aggregations of what the ”polls say.” Guess the old excel file was too hard to manage in his absence. Funny, that never happens!
FiveThirtyEight now looks indistinguishable from parent co ABC’s simple presentation of this information, which focuses on their polling averages. It also provides a good example of a chart where it’s okay to trim the axis.
Lastly we have a bot trained to be an superforcaster. 52-55%.
So, that's the first step. What do other folks say.
The second step is to LOOK at the data. And form some sort of opinion about where you think the actual center of gravity amongst these (potentially misaligned or contradictory) sources of information.
Actually try to get your hands on it and see if it makes sense. Does it align with your understanding of reality? Below we can see the market implied odds of a trump victory since the start of the year. We can see the peak post-Biden debate, and then a long slow march back to ~parity as the Dems rallied behind Harris.
Most people would show you these two lines on a chart and call it a day. This would be wrong. There’s so many reasons why these two lines might correlate over that period, it’s not enough information.
We need to stress test, which we can do by comparing the correlation of changes in these odds to the returns we see in the market.
When we line up all these correlations, it's pretty messy, but it tells a story that you can map through the politics and connect to the policy differences between Republicans and Democrats.
What we see, starts to tell a story. Aka it looks like the Yen and US short term bonds are both the two most negatively correlated assets in our sample. As opposed to Consumer Discretionary and Tech, the dollar, and Chinese stocks. Interesting.
Again, don't forget to LOOK at the underlying data and observe the 'structure' of those correlations before drawing any conclusions. These correlations aren't particularly stable and fluctuate between positive and negative, which should give you pause.
With that in mind, what preliminary conclusions might we get from those correlations?
First, in terms of monetary policy, the negative correlation to the yen and US short-term bonds, along with a positive correlation to the dollar, implies that Trump's election would lead to rising rates and a 'strong dollar'.
On the fiscal & growth policy side, we have positive correlations for hyper cyclical assets like tech, consumer discretionary and emerging market stocks. These would contradict the interpretation that Trump's monetary policy would be tight, suggesting instead fiscal expansion and other policies that would support stronger nominal income growth
Where does that leave us for tomorrow’s debate?
Given the relative weakness of the correlations and their unstable structure, I probably wouldn't make any bets in financial markets based on these odds. It's somewhat interesting that they're moving in the same direction as markets did after the 2016 surprise (bonds down, dollar up, stocks up), which suggests that markets are at least attempting to price this in.
However, nothing in particular stands out, and given our lack of confidence in the direction, we prefer to simply measure these moves and observe market reactions during tomorrow's debate to gauge market sentiment.
How to do this?
The next step is to track these markets as the debate unfolds. The most likely scenario is that the debate ends in a stalemate, with odds and polls not changing materially post-debate.
If they do change, it's worthwhile to compare that movement to the markets and see if it aligns with our preliminary conclusion (Trump's performance weakly correlates with rising stocks and spending, and negatively correlates with the Japanese Yen and US short-term bonds). We'll do this later this week if the results are interesting
Until then, happy hunting.
Appendix: How to pull this data in rose.ai
For those of you who are interested in following along, we recommend, you start in this notebook we made, which should pull the data from this ramble directly from polymarket into a notebook you can screeshot, export or download.
For those interested in accessing all the election data we have available from Polymarket (professionally or otherise), you can use our marketplace tab on rose.ai to directly license this data for free for download into Excel or access via our API.
Disclaimers
Interesting that Polymarket shows higher odds for Trump. I am guessing this has something to do with the fact that polymarket is a crypto platform and thus maybe skews to the right. Also, I believe Nate Silver kept ownership of his prediction model when he left 538. Seems like Disney essentially bought the website without any of the substance that people cared about.
Great post. And.....the multi-decade deterioration of US politics has the current political theater roughly on par with "professional" wrestling. i.e. it's mostly a Kayfabe show.