Predictions, Spring 2021

Intellectual honesty, and why Trump will be the 2024 nominee

I write about politics and the social sciences for a living. But a lot of people do that, and this fact raises the question as to why you should follow my work when there are so many other options out there.

I think there are many reasons, but one of the most important is that I’m willing to make public predictions with probabilities to allow people to hold me accountable. Instead of doing my own bookkeeping, I will outsource that job to PredictIt, as I already play the markets there.

I began betting on politics in fall 2018. Since that time, I’ve invested $15,700. I’ve taken out $10,000, and the current value of my portfolio is now approximately $29,000. This means that I have so far made 2.5x my original investment, which makes me feel pretty good about myself. This underestimates how well I’ve done, as PredictIt takes 10% of your winnings. That’s 10% of each successful bet you’ve made, not 10% of your net profits, making it more impressive. (i.e., imagine you win $100. They then take $10, then if you lose $100, you’re down to -$10 in net profit, and then if you win $100 again, they take another $10, and you’re only up +$80. So you’ve won $100 twice, and lost $100 once, but instead of your total payout being $100 or $90, it’s $80. You need to consistently beat the market just to break even)

This is a big relief for me. Tetlock showed that geopolitical experts are on average not any better than educated laymen at predicting events. This to me is one of the most interesting findings in the social sciences, it fit with my view that most of what we call “expertise” is fake, and encouraged me to prove to myself that I was worthy of commenting on public affairs. If it turned out I did worse than the market, there really wouldn’t be much of a justification for me continuing to write about politics, as it would mean you can just go to PredictIt and look at the averages and end up with a better understanding of the world than I can give you. Luckily, it seems that I do add some value.

Other intellectuals who I admire for their honesty like Matt Yglesias and Scott Alexander have given quantitative predictions for 2021. Yglesias uses Metaculus, which has the advantage of letting you predict a wider range of phenomena. Unfortunately, it’s less user friendly than PredictIt, and doesn’t use real money, which means it’s harder for me to find the motivation to become a regular user. So despite its flaws, I’m going to stick with PredictIt. Metaculus seems to attract a smarter crowd, but PredictIt uses real money (with limits), so it’s not clear which market is tougher to measure one’s self against.

I hope to report my returns on PredictIt twice a year, so I’ll revisit this in the fall (let’s say October). For now, I’ll review what I’ve gotten right since I’ve started playing the markets, what I’ve gotten wrong, and talk about what I’m currently betting on. This will be the normal format for these posts.

What I Got Right

I’ve been very good at predicting everything relating to Trump. By 2018, I had seen the same cycle time and time again, when people would predict Republicans would turn on him, then GOP voters would prove they liked Trump a lot more than anyone else in politics, and he’d vanquish his enemies. Then we’d start the whole cycle over again, and people will have forgotten what just happened. I personally thought Trump was finished after he made fun of McCain being captured in Vietnam, all the way back in July 2015. That was the kind of thing that traditionally got you cancelled in the Republican Party, but it didn’t happen in this case, and I learned a valuable lesson about Trump’s hold over the party that has allowed me to successfully predict events since.

I’ve previously written about how liberals care more about politics. Well, within the Republican Party, the pro-Trump faction appears to really like the guy and be dedicated to him, valuing loyalty to Trump over positions on any actual issues. The fanatical pro-Trump faction might be overcome if enough Republicans didn’t like him, but that’s not the case. It seems that there’s something like 30-40% of Republicans who love Trump, 40% who like him with less intensity, and 20% who dislike him with not much intensity (the faction that represents the Liz Cheneys of the world, Republicans who dislike him with high intensity, seems to be limited to those who work for MSNBC and The Washington Post). Theoretically, you could maybe get an anti-Trump majority or plurality, but those are terrible numbers for anyone who wants to do so. A lot can change between now and 2024, but a lot has happened between 2015 and 2021, and real world events have never seemed to impact this simple dynamic within the GOP. Liberals and anti-Trump conservatives who want the man to go away are vastly overrepresented among the chattering classes, which is why the conventional wisdom keeps being wrong.

I won the following Trump-related bets.

1) Trump would finish his first term (bought at approximately 75-80%)

2) Trump would be the 2020 nominee for president (bought at approximately 75-80%)

3) Trump would not be convicted in the first impeachment (bought at approximately 80%)

4) Trump would not be convicted in the second impeachment (bought at around 50% after January 6)

Each of these markets was really a market on whether Republicans would stick by Trump, and with few exceptions they always have.

I also predicted Biden would be the Democratic nominee in 2020 (bought at 26%, wrote about it here, screenshots included). I simply looked at the polling data, saw Biden leading for a really long time, and thought that everyone on Twitter saying he would collapse was just engaging in wishful thinking. I’d seen that in every recent case when someone had a polling lead as long as he did, they ended up being the nominee for their party. I got scared when Sanders did well in the first few primaries, and actually bought some Bernie shares at 39%, but I still held Biden on the assumption that if it wasn’t Sanders, then Biden was the only person the party could possibly rally around.

Maybe my best individual betting day was on Super Tuesday, when I bought Biden shares in states like Minnesota and Maine for around 10-20%. We didn’t have a lot of polling, and this was right after Klobuchar and Buttigieg endorsed Biden to stop Bernie, so I thought there was a good possibility that there would be a major shift towards Biden on that day.

I also predicted Biden would win the presidency, and bought Democrats having unified control of government for about 50/50, which I won but could’ve easily gone the other way.

I predicted Republicans would confirm a new Supreme Court Justice before the election and after the death of Ginsburg (about 50-60% odds IIRC). People underestimate partisanship, and Republicans have always fallen in line on judges. When there were reports Kim Jung Un was dead, it was actually 50/50 whether he would still rule North Korea at year’s end 2020, and I didn’t think some unconfirmed rumors were enough to predict the demise of a guy in his mid-30s, no matter how fat. Also thought Maduro would hang on (about 60% IIRC).

Things I Got Wrong

I bet against Kamala being the VP when YES was around 25%. I thought Biden might choose her, but the price was too high. I also had money on Klobuchar as VP nominee for much less, and maybe it would’ve happened if not for George Floyd, after which Biden specified that he was not only going to pick a woman, but a black woman.

I also put a lot of money on Stacey Abrams running for president, when it was around 15%. There were also a bunch of cases where I’d bet Democrats wouldn’t run for president, when YES was around 70-80% for most of them, like Amy Klobuchar and Seth Moulton (still can’t believe that he was delusional enough to think he had a chance as a boring generic white man no one had ever heard of or cared about).

I underestimated how many Republicans would vote to convict Trump in the second impeachment, even though I was right that he wouldn’t be convicted. I also bet on a popular vote blowout of 10 points or more for Biden at 15%, assuming that because the polls had him up by 7 or 8 points there was a decent chance of an error that would work in the direction of the Democrats. I also took 50% or lower odds of Biden winning states like Ohio, Texas, Florida, and Iowa on the assumption polls were right and markets were too optimistic towards Trump. Similar mistakes were made in Senate races like those in Iowa and Maine. Basically I trusted the polls too much in 2020, and though I was right about Biden ultimately winning and the Democrats taking Congress, it was much closer than I thought.

Predictions for the Future

Here are the things I’m betting on at PredictIt now, with how much I originally paid, what the price is now, and what I think the actual price should be.

Republican nominee for 2024: YES on Trump (Bought 26%, currently 26%, should be 75%)

This is basically following the same formula that I’ve been winning with the whole time: assume Trump will win every battle within the Republican Party. Right after January 6, I didn’t think that the Senate would convict him, but I did tell Razib that after the Twitter ban he might go away. Yet I couldn’t have been more wrong; the Republican Party has completely rallied around him and his personal obsessions, and unlike the Tea Party wave, the internal fights haven’t been about anything ideological like the size of government. It’s just been about Trump, with loyalists purging those who have shown any opposition. I disagree with those who think Trumpism is about economics, and those who think it’s about racism. It’s literally just about the man, and he has a deep emotional connection with the base, which is why we can’t expect someone to just take Trump’s most popular positions and run with them.

What about his health? The life expectancy of a 74-year-old male is 11.8 years. Trump is fat, but he has no major health problems we know about and access to unusually good health care. So I wouldn't give good odds for health related reasons stopping him from running. Some people have argued that he might not want to run, and I find this argument absolutely crazy given everything we know about the man.

Democrat nominee for 2024: YES on Biden (Bought 34%, currently 39%, should be 85%)

I don’t understand how Biden being the nominee can be as low as it is. Just take the base rate. A current president in his first term is almost always the nominee the next time around. Like with Trump, I think people are just radically overestimating how likely it is that an old man that rambles a lot is to going to die soon. I also bought NO on Kamala for 61%. The market is currently 39% Biden, 38% Harris, when I think in reality it’s something like 85% Biden, 14% Harris, and 1% other.

It doesn’t seem that PredictIt is alone in underestimating Trump and Biden. Election Betting Odds actually has both of them lower than PredictIt. If I’m right about a Trump-Biden rematch in 2024, then it’s probably because these markets are dumb, not because PredictIt limits you to $850 a contract.

King of the Republican Party, as long as he wants it

Here are some of my other bets, following the “bought-currently-should be” format. I will be trading between now and October, so it doesn’t mean I’ll stick to all these, and I’ll bet on new markets as they emerge. The accounting in October will be based on how much money I’ve gained or loss in the aggregate since today, and across my entire betting career.

What will be the corporate tax rate for 2022? Bet NO on 24.6-27.9% (bought 35%, currently 38%, should be 50%)

Gaetz on Judiciary Committee 9/1? YES (32%, 41%, 55%)

Will Biden resign in his first term? NO (75%, 78%, 96%)

Who will win the OH Republican Senate primary? YES on JD Vance (37%, 30%, 55%)

Marjorie Taylor Greene reelected to the House in 2022? YES (63%, 72%, 90%)

Will DC become a state in 2021? NO (93%, 96%, 98%)

More than 9 Supreme Court justices in 2021? NO (91%, 97%, 99%)

Harris files to run for president before end of 2022? NO (69%, 82%, 98%)

Will Liz Cheney win the 2022 House GOP nomination in WY-AL? YES (31%, 31%, 55%)

Will Adam Kinzinger win the 2022 GOP nomination in any Illinois House district? YES (42%, 26%, 55%)

Who will win the 2022 Georgia Republican Senate nomination? YES on Herschel Walker (41%, 41%, 65%)

Which party will win the 2024 US presidential election? YES on Republicans (47%, 47%, 55%)

Will Andrew Cuomo be Governor of New York at the end of the year? YES (55%, 75%, 85%)

Other Predictions

That’s it for PredictIt. Here are some predictions about American society and international relations. All these predictions are about the event in question happening by the end of 2021 unless otherwise noted.

US will see fewer than 100 deaths from political violence: 90%

US will reenter the JCPOA in some form: 75%

The Taliban will take Kabul: 55%

The US will normalize relations with Taiwan: 2%

The US will normalize relations with Syria: 3%

The US will go to war with Iran: 2%

China will go to war with Taiwan: 3%

Biden starts a war that kills at least 20 American soldiers: 5%

Saudi Arabia normalizes relations with Israel: 15%

Russia and Belarus will merge or announce some kind of intention to do so: 20%

If Republicans take the House, and Democrats win the 2024 presidential election, the Republican House refuses to certify the 2024 election in favor of the Democrat: 30% if Trump is the Republican nominee, 5% if anyone else is

Ron DeSantis will be the Republican candidate for VP in 2024: 40%

Tim Scott will be the Republican candidate for VP in 2024: 35%

I will reach 20K Twitter followers: 40%

I will reach 50K Twitter followers: 5%

US murder rate stays above 2019 level in 2021: 90%