Friday, December 2, 2022

BTRTN: How We Nailed the Midterms with a Near Flawless Prediction

Tom with a wrap-up on how BTRTN did with our midterm predictions. 

The media and the pundits may have missed the big picture, but not BTRTN!

As Election Day approached, outlets across the ideological spectrum featured “expert” panels and feature articles breathlessly announcing the imminent arrival of a “red wave,” pre-ordained by the immutable “Law of First-Term Presidents.”  Joe Biden would be repudiated and the Democrats eviscerated; the Senate would be lost and the House would be an utter bloodbath.  After all, it happened to Clinton (who lost 54 seats in the House and 8 in the Senate), Obama (-63 and -7) and Trump (-42, though he managed +2 in the Senate), so of course it would happen to Biden too, given his shaky approval rating.

We, on the other hand, paid less attention to history and more to the polls.  We aggregated them analytically in our various models, tossed in a dose of judgment and suggested another conclusion:  we believed the Senate was a toss-up in which the odds leaned slightly toward the Democrats holding on by a slim majority, and, while losing the House decidedly, not by anything approaching “red wave” margins.  Our predictions were not only extremely close to the final outcome, they were “best in class,” beating those of FiveThirtyEight and Sabato.

·        Senate.  We accurately predicted 34 out of the 35 Senate races, including 7 out of the 8 battleground races.  The only prediction we missed was the very last race called, and the closest one, in Nevada, where incumbent Democratic Senator Catherine Cortez Masto defeated GOP challenger Adam Laxalt by less than a point.  Our ledger included the correct prediction that Georgia would go to a runoff.   (We will publish our prediction for that race on Monday, December 5, the day before the runoff.)

·        Governors.  We accurately predicted 35 out of the 36 gubernatorial races, including 4 out of the 5 battleground races.  The only prediction we missed here was also the very last race called, also the closest election, in Arizona, where the Democratic nominee Katie Hobbs defeated GOP challenger Kari Lake, also by less than a point.

·        House.  We predicted the GOP would take the House and pick up 21 seats; they did take control and actually picked up only (either) 8 or 9, depending upon California 13th, the only race yet to be called. This was not as close as our past House midterm prediction history, but we take (tiny) credit for predicting the GOP would win control and (a bit more) credit for predicting a relatively modest Dem loss of seats -- certainly not a red wave.  We forecasted 417 out of 434 individual House races correctly, in line with our prior performance, largely due to the GOP winning a number of “toss up” races that we had in the Democrats’ column.

These were not easy midterms to predict.  About a month before the election The New York Times declared that “the final stretch of the 2022 midterms defies predictability.”  (My emphasis added.)  Any number of factors contributed to this sense of “anything can happen” – the persistent polling issue of under sampling Trump supporters; the impact of redistricting; new voting laws that both suppressed and enhanced access to voting booths; the institutionalization of mail-in and early voting; the “Trump” factor; the potential for unexpectedly high turnout from key voting blocs like woman and young people; and just the sheer number of very close races.  From where we sat, it certainly felt daunting.

We set the stage in our first midterms update back on February 13, 2022, in a piece that has aged well.  (  We started with this observation:

“How many conversations have you had with morose Democrats recently, bemoaning the ‘near certain’ debacle in the upcoming midterms?...{But) the midterms are not as hopeless as you think, not by a long shot....Much can and will happen in the next nine months.  But there is a path to victory if events break right.  It will take some skill by the Biden Administration, and some luck.”

We pointed out that it was not “destiny” that leads to first-midterm debacles for new presidents – it is usually “performance.”  They earn it!  After all, neither Bush got crushed in their first midterms (George W. Bush actually picked up eight seats, while George H.W. Bush lost only eight), and that was largely because both Bush’s handled their first-term crises (the Gulf War and 9/11, respectively) quite well.  (Both Bush’s ran into massive problems after their first mid-terms, the elder Bush with economic woes that cost him his reelection, the younger Bush with his ill-fated war in Iraq, which commenced in March, 2003). 

We then outlined a Biden “comeback plan” that looked like this:  pray for no new Covid variants; continue robust economic growth; tame inflation at least a bit by November; leverage the appointment of a new Supreme Court justice and an adverse Roe ruling; manage the Ukraine situation successfully (this was still pre-invasion); hope that the January 6 committee keeps Trump in the headlines; figure out some way to inoculate vulnerabilities on crime, education and immigration; and pass some version of the Build Back Better legislation.

Biden and the Democrats did not get all of that.  In particular, economic growth slowed and the October  inflation report, the Dems’ last chance to show progress before the midterms, was a dud, as inflation continued to roar unabated despite five separate Fed rate increases.  But they did get most of it (and then some, with a bevy of legislative triumphs beyond the re-tooled BBB, a.k.a. the Inflation Recovery Act), giving them something to talk about as the Labor Day push began.  Had it not been for that adverse inflation report, the Dems just might have kept both chambers.  As it was, they certainly made it respectable.  The pundits might have called it a miracle, but we saw the opportunity from the start.  We also pointed out many times along the way that Biden and the Democrats were being helped, especially in the Senate, by three GOP swing state retirements, and the truly appalling slate of GOP candidates pushed by Trump to replace them, B List celebrities with tons of baggage, a dearth of political experience, and no particular on-the-stump magic.

Apart from all that, as we tracked the races and then made our actual predictions on November 7 (, we ignored the pundits, of course, and (largely) followed the polls – and the polls got it right.  As aggregators, our predictions are only as good as the collective accuracy of the polls, and, in aggregated form, they pointed to the scenario that actually unfolded – a toss-up Senate and a modest Dem loss in the House.  Of course, when the races are tight, as so many were, we have to assess other factors, including “momentum,” incumbency, money, ground game and some sense of judgment based on our experience – for instance, that some states always seem to break red or blue in the end (for example, Florida of late to the GOP, Pennsylvania to the Dems).  We also recognized early that various GOP entities were flooding the polling marketplace with polls that tended to shade red, and made appropriate adjustments.

The only other services that predicted every one of the 506 races in advance were FiveThirtyEight and Larry Sabato’s Crystal Ball.  (Many other services simply said various races were “toss ups” and left it at that.)  Neither of them envisioned a “red wave” either – no polling aggregator did except for Real Clear Politics, a right-wing group that was embarrassed by their red wave predictions. 

But both the FiveThirtyEight and the Sabato predictions were a shade “redder” than ours on most measures, and thus they both got more seats wrong in the Senate and with the Governors (though FiveThirtyEIght was a tad closer to the mark in how many seats the Democrats would lose in the House).  All in all, though, we think it is fair to say that BTRTN won the night.  Fame and fortune await!

Comments welcome, and if someone did better, please feel free to correct the record!

Here is our updated BTRTN prediction history:

If you would like to be on the Born To Run The Numbers email list notifying you of each new post, please write us at

No comments:

Post a Comment

Leave a comment