Betting on the Oscars
A reflection on prediction market strategies for elections and awards.
In 2016, Americans were developing a taste for risk. The hardship of the Great Recession was starting to wear off. Bitcoin was climbing. Robinhood was bringing day trading culture to the masses. Employment was finally rebounding. People had more money in their pockets, and they wanted new and more dangerous ways to use it.
Amid all this, PredictIt rose to prominence. On paper, PredictIt is a university-affiliated, New Zealand-based project focused on gathering political research data. Almost all of that is nonsense. From a practical perspective, PredictIt is a website where people make cash bets on future world events like presidential races.
These bets use a double-auction system, which means the odds are not set by a bookie or professional oddsmaker. They fluctuate with a market of buyers and sellers. This is different from how traditional sports betting works. In some ways, it's more like the stock market.
One of the things that normal people ask when they hear about PredictIt is, “How is this legal?” The answer is sketchy. In short, PredictIt used its ‘university research’ pretense to get a ‘letter of no action’ from the Commodity Futures Trading Commission (CFTC). That letter was rescinded in 2022, and the CFTC began to take measures to shut PredictIt down. PredictIt appealed, and a judge granted them license to continue their business during the appeal process. For the moment, PredictIt is still running.
Socially, I'm one degree away from the PredictIt founders, and from what I understand they are eccentrics who made their fortunes selling electronic voting systems around the world. It makes sense that people with that background would have the connections necessary to muscle through regulation, much like Uber breaking up the taxi medallion system. Take from this story what you will.
I fell into PredictIt while working a relatively understimulating job in North Hollywood amid the early pandemic. I was a fresh college graduate, newly fully employed, and I was itching to use my money to try and prove that I was clever. After a few early successes, I ended up betting a large chunk of cash on Biden winning the presidency.
I paid 60 cents a share, for a potential payout of one dollar per share if I was right. This was a few months before the 2016 election. To me, 60 cents was a good price, and from what I was seeing the election seemed fairly certain. I told myself that if I won the bet I would quit my job and buy myself a revolver (just for fun). If I lost, I promised to eat my humble pie and stick to the S&P 500 for the foreseeable future. Biden, of course, did win, and I followed through with both the career change and the big iron.
Some time after that, I was working a better job and took a short retreat to the mountains with an old friend. We ended up snowed in, and in our isolation we both sat down with the goal of authoring something meaningful. What he wrote, I can't remember. I'm sure it was good. What I wrote was a white paper formalizing what I had learned from about a year of trading on PredictIt.
The core of my thesis was that people in political betting markets overweight black swans. ‘Black swans,’ in this case, meant things that could happen but are not very likely. Nassim Taleb talks a lot about black swans. In real life, he argues, we don't take them seriously enough. We get complacent, and we ignore the possibility of system shocks.
On PredictIt, on the other hand, traders seemed to take black swans far too seriously. Ridiculous longshots, like The Rock becoming president, would routinely trade at 4, 5, or 10 cents on the dollar. If a bet on The Rock becoming president costs 10 cents, that means a bet that he won't become president costs 90 cents. Whoever wins keeps the one dollar pot.
This means you can bet against The Rock for $900 and get $1,000 back. That's an 11.11% return, and it happens oftentimes in a matter of weeks (if the market is close to resolving itself). Then you can turn around and put that principal into a similar bet against another overvalued black swan and compound, compound, compound.
Of course, there is some risk that The Rock could win, which is why you don't want to put all your money into one single market. You need to hedge across at least a few. However, even if The Rock does have some chance of winning, it's far less than 10%, which means that across enough bets like this you end up making a lot of money. This framework became the basis of a trading operation that we called Askari Predictive. Askari, meaning ‘warrior,’ is my mother's mother's maiden name, and also the name of my production company.
We put decent time and decent money into the endeavor. The returns that we saw betting against black swans were quite strong, and the frequency with which we could compound made it all the better. The thesis from the snowed-in cabin was bearing fruit.
Unfortunately, we soon turned into a big fish in too small a pond. PredictIt has an $850 cap on positions in a given market, and liquidity for their markets is often fairly shallow, so a big bet moves the market a lot. More perniciously, they take a 10% fee on winnings and a 5% fee on withdrawals, which is a huge blow when every dollar counts. Imagine your brokerage account taking 5% of your funds when you withdraw. It just doesn't make sense. So, after a year on PredictIt, we set off looking for friendlier pastures.
We found them, and Askari Predictive is still going strong. These days, we do most of our position-taking on Polymarket, which is similar to PredictIt but powered by a Polygon, a decentralized blockchain layer. For us, Polymarket is better than PredictIt in every way. They don't have caps on their markets, and they can offer a wider range of events for speculation. The experience is clean, and the liquidity is decent. The nature of the blockchain also makes the site difficult to censor.
This year, we at Askari Predictive have branched out of politics and put together an Oscar prediction portfolio. The Oscars on Polymarket have a surprising amount of traffic. It seems that, even after the end of the zero-interest-rate phenomenon, there are plenty of people out there with money to blow betting on cultural events.
When we approach a new market, we start with research. For 2024, we've gone through many previous years of Oscar predictions and recorded the way that various publications and entertainment journalists anticipated things would go. These old prediction articles, from Variety, The Hollywood Reporter, Entertainment Weekly, and others, are available online. Based on their historical predictions, we assigned each individual journalist or publication a score representing their overall and recent accuracy. Then we took their predictions for 2024, weighted them by accuracy, and created a single aggregate forecast.
Our analysis of this year's Oscars markets found a surprising reversal of norms. In a typical online betting market, the frontrunner is underpriced and the black swan is overpriced. In the case of the 2024 Oscars, however, it actually seems like the markets are not giving enough of a chance to the underdogs.
From our forecast, the odds of the frontrunner losing to an upset in any given category seem to be around 27%. In comparison, some of the frontrunners on Polymarket are trading as high as 95 cents, which means you can take the ‘NO’ position for as low as 5 cents. Obviously, a 5 cent buy for a potential $1 return, with a 27% chance of seeing that return, is a very good bet.
For fun, and for the sake of this piece, we also polled regular people from our family and friends who consider themselves Oscar aficionados. We wanted to see how their predictions compared to those of the experts. As it turns out, the two are very close. This makes sense to me, in retrospect. After all, the people who work at these media outlets are just entertainment aficionados themselves.
If you believe in efficient markets, you might be worried that our model is wrong and the betting sites are correct. This is a reasonable fear, especially because this is the first year we've entered the Oscars fray. One thing that should provide some reassurance, though, is the fact that this 27% figure appears supported by multiple different kinds of analysis. If you look at the history of Best Picture winners going back 100 years, a little over a quarter of them were considered surprising upsets in their day.
Below is a distribution of potential outcomes for our current positions if we assume a 27% chance of an upset. Keep in mind that this distribution is treating each Oscar race as an independent variable. This is obviously not the case in real life, but is hopefully a survivable modeling conceit.
If you look at only the Oscars since 2010, the rate of upsets drops closer to a still-respectable 21%. If we dial in our model to this more conservative chance of a payout, here's what it looks like.
As you can see, this is a pretty high-risk play. It's not something that I would feel comfortable doing with more than a fifth of our portfolio, because so many of the outcomes result in a complete wipeout. However, as you can observe, the median outcome and average outcome are both in the black, even in the more conservative model.
We have a few reasons, aside from the data, to think that there are likely to be upsets. The Oscars now use ranked choice voting, and ranked choice voting produces unusual, unintuitive results because it often results in the compromise candidate winning. People in American democracy aren't used to thinking about compromise candidates because our voting systems, aside from those in Alaska, are generally highest-takes-all.
In addition, there are new elements in play that have the potential to disrupt people's mental models of what the Oscar winner should be. There are a lot of newly-inducted Academy members, and those inductees may vote in non-traditional ways. There are also diversity requirements for the nominees, which is a major complicating factor if you're trying to use nominations as evidence of true popularity.
All of these factors serve to confuse the voting process in a way that can lead to surprising results. If we're betting against the frontrunners, surprises are what we want.
In a few weeks, we’ll get to see how this strategy plays out. If you're joining us on the field, remember to keep a cool head, and always pay your taxes on your winnings. After all, that's how they got Al Capone.
Futurist Letters is an independent publication and a labor of love. It is entirely user-supported, and any patronage you provide is greatly valued. Paid subscribers have the ability to comment and browse the article archives.
If you liked this post, please consider sharing it with some friends, or on social media. You can also follow Cairo Smith on Twitter, Instagram, and Letterboxd.