Dem
50
GOP
50
We had a question, and then some letters, last weekend about prediction markets. So, we will pass along this news from Thomas Miller, who oversees Northwestern's data science program. In 2020, he used prediction markets to forecast the presidential election, and he correctly predicted a win for Joe Biden, getting only one state wrong (Georgia). That's better than we did, that's better than Nate Silver did, that's better than Cook Political Report did.
This year, Miller is back at it. Prior to the first debate (Trump-Biden), he thought that Donald Trump was in command of the race, and that 400 EVs were potentially within reach for him. The replacement of Biden with Kamala Harris flipped that script, such that prior to the second debate (Trump-Harris), Miller had the Harris-Walz ticket winning a close election, with 289 EVs. Now, he says that there's movement in Harris' direction that, if it holds, could presage a landslide. More specifically, he says that it's now Harris for whom 400 EVs are within reach.
We encourage caution here; you don't want to take this projection to the bank (or the betting market) quite yet. Miller's got a pretty short track record, and we are particularly leery of any system that says Donald Trump could win 400 EVs. That said, the data scientist did very well in 2020, not only on the presidential race, but also in the Senate races (he called the two Georgia races with a high degree of accuracy). So, we thought this was at least worth passing along.
And since we are on this subject anyhow, Allan Lichtman of "13 keys" fame, made his projection last week. He says eight keys (no internal contest for the nomination, no serious third party bid, economy not in recession, real per capita income growth, successful changes in national policy by the incumbent administration, no serious social unrest, no major scandals for the incumbent administration or the candidate, and a challenger who is not charismatic or a national hero) favor Harris. Lichtman says three keys (incumbent party lost seats in the midterms, the opponent is not the sitting president, the opponent is not charismatic or a national hero) favor Trump. He also says that two keys (major military failures, major military successes) are undetermined. In Lichtman's system, the incumbent party needs at least eight keys to prevail. Since Harris has them, Lichtman says she'll win.
After Lichtman made his call, we got many e-mails asking why we did not write an item about it. There are two reasons. First, we don't love his system. It's pretty squishy and subjective, and his "stellar" track record is based on a small number of close elections. Depending on exactly how you define that term ("close elections"), his success rate is basically the same as a coin flip. Also, his system does not see a difference between big wins and small wins. Sometimes, a candidate has 11 keys going for them and wins a squeaker. Sometimes, they have 8 keys and they win a blowout. That is... concerning, when it comes to evaluating the methodology.
The other reason we did not write it up is that this is the third time this cycle Lichtman has had his big reveal. He did it once for Biden, and then again for Harris 5-6 weeks ago, predicting a Democratic win both times. So, there isn't exactly any news here. Since Lichtman doesn't do squeaker vs. landslide, and since most of his keys don't change, his prediction now is the same as it was back then. Still, since we were writing about Miller anyhow, we tacked Lichtman on for those who really wanted to see something about his (latest) prediction. (Z)