Finally, Real Money

Kalshi as the first website to allow real betting in the US

There’s finally a website that lets Americans bet real money on public events. Thanks to pressure from the political appointees in the Trump administration overruling the bureaucrats at the CFTC, Kalshi allows you to risk as much cash as you want as if this is a free country or something.

After I joined the website, they gave me a custom link, where if you click on it and sign up I get $10. If you make 100 bets, you’ll also get $30 yourself. Consider it a contribution to the Substack, particularly if you’ve enjoyed it but never thought to donate. Please use the link instead of just going to the website directly so I can get paid. I’ll take whatever money I make off the Kalshi link and put it back into betting markets.

Another advantage of Kalshi is that its fees are much lower than for PredictIt. Unfortunately, the two things I feel most comfortable betting on are American electoral outcomes and geopolitical events, and Kalshi doesn’t do either of those as of yet.

Here’s a table that shows my dilemma.

Still looking for my ideal website that will let me bet real money on the things I know best. See also the Alexander Cube.

A lot of the questions at Kalshi are about stuff like inflation and GDP growth, where I don’t have any reason to think I could beat the market. Fortunately, they also ask about Supreme Court cases, the fate of proposed bills, and executive orders, so I’ve been able to bet on some things that are closer to my core interests. Here’s what my portfolio looks like so far.

The trade volumes currently seem very small. When I first went to the site, it gave a 68% chance that the Supreme Court would ban the use of race in college admissions. I spent $500 on that market, and by myself got the number up to 72%.

I have a public figures page on Metaculus, but honestly it’s hard for me to find the motivation to keep up with it given that there is no money involved. Manifold Markets has the same problem. My time is valuable enough that to induce me to keep up with a prediction website I need both money and the chance to be able to test my forecasting ability. One of these things tends not to be enough.

Last time I reviewed my portfolio on PredictIt was May 2021, when I reported I had turned $15,700 into about $39,000. Of that, I had taken out $10,000, for a portfolio value of $29,000 at the time. Since then, I’ve withdrawn another $4,000, and my portfolio is at $27,500, which means I’ve gained about 12% in the last year. I took a killing on Cuomo, who I believed would hold on to power. But my predictions for 2024, where I’ve always said it would be Trump and Biden, have held up well.

A year ago, I put Trump at a 75% chance of winning the Republican nomination in 2024, when the market put him at 26%. The market is actually now giving him a 38% chance, while I’ve decreased my forecast to 65%. I always thought the only chance that Trump could lose would be if one candidate could consolidate the anti-Trump vote. It’s looking more likely now that DeSantis is the obvious choice. A series of polls show Trump continuing to lead the pack, but DeSantis within 20 or 30 points or so and, more importantly, clearly ahead of everyone else. Part of the difficulty in stopping Trump in 2016 was there wasn’t a single alternative that everyone could get behind from the beginning. DeSantis is clearly more popular with conservative intellectuals and activists than any other candidate, and is a more likable politician than Cruz, Pence, Pompeo, or anyone else who may run, unless maybe Tucker decides to jump in. So my current Republican forecast for 2024 is Trump at 65%, DeSantis at 30%, and everyone else at 5%.

Even more incredible than the extent to which markets underestimate Trump is their underestimation of Biden. Since World War II, there have been 14 elections where a sitting president was eligible to run for another term. In 12 of those cases, the sitting president was the nominee of his party. The only exceptions were Truman in 1952 and Johnson in 1968, and these were very unusual cases in that each of them would have been running for his third term. The 1968 Democratic nominee ended up being Hubert Humphrey, who had been Johnson’s vice president. Therefore, in situations where a sitting president was eligible to run again, we have him getting the nomination 86% of the time, and the nominee of his party being either the sitting president or vice president 93% of the time. If we limit our analysis to post-1968 – on the theory that the more recent past is a better guide to the present than the distant past – a sitting president who has been eligible to run again has been the nominee of his party 10 out of 10 times.

So what do betting markets say this time? For the Democratic nomination, they give Biden a 35% chance, and Harris a 17% chance. Betfair isn’t that different from PredictIt, so it’s not an issue of limits on contracts. In other words, there is supposedly only a 50-50 chance that either the current president or vice president will be leading the Democratic ticket in 2024. This is insane. Yes, Biden is old, and Kamala is unpopular. So what? The probability of a man at Biden’s age dying is about 5% a year, but that includes those in nursing homes, people in a vegetative state, etc. There are no credible reports of Biden having serious medical issues, and no indication that he doesn’t want to be president anymore. I’m sure that if you calculated the odds of incapacitation or death for people serving as politicians at Biden’s age without any current health problems being discussed in the media, they would be quite low.

So maybe because of the age issue you adjust from the base rate a little bit, let Biden go down from the historical norm of 86% to say 75%. And then you give Kamala a 50% chance of being the nominee in the case it isn’t Biden based on the recent historical sample size of 2. That gives you an 87.5% chance that the nominee will be either Biden or Harris, and reasonable people can maybe disagree and adjust that number down to 80% or even a bit lower. But Biden and Harris combined being at 50%? There’s no way that can be justified. “Stick close to the base rate” is such a boring heuristic, as it’s much more fun to imagine some kind of Democratic implosion that sweeps aside both Biden and Harris and they all draft Oprah or The Rock to save them.

The market, by the way, made a similar mistake with Trump in 2020. I remember that months before the Republican convention, he was only given around a 75% chance of being the nominee. And some people kept expecting Biden to drop dead that year too. So base rates seem to be something markets don’t really understand yet when it comes to politics. I expect this problem to be solved in the long run.

If it’s not Biden or Harris, I’ve always thought that Stacey Abrams was the most realistic alternative. David Sacks thinks that Gavin Newsom has a shot, but try and imagine the modern Democratic Party taking the nomination away from a black woman who is the sitting vice president and giving it to a white guy instead. Abrams is the only individual who has the right characteristics to take the nomination away from Kamala, and makes sense as a high-risk/high-reward strategy for the Democrats if things continue to look bad for them in 2024.

Right now, I’d say the Democratic probabilities are 80% Biden, 15% Harris, 2% Abrams, and 3% other.

I’ll continue to use PredictIt, and I’ve also put $6,000 into Kalshi. Expect more updates on both portfolios as time goes on. Again, I encourage you to join Kalshi through this link so I can get $10 per sign up and keep gambling to ensure that I still know what I’m talking about.

Note: I’ve added a Substack heading under which you can find all my previous forecasts and writings on prediction markets.