Skip to main content
ABC News
The Debate Exposed Bloomberg’s Downside — But It Was There All Along

On Wednesday, I was working on a story about how our primary forecast was handling former New York Mayor Michael Bloomberg when other things got in the way. Truth be told, I’m glad I got distracted, because a mildly bearish case on Bloomberg — roughly how I’d describe our model’s current position on him — is a lot easier to explain after Bloomberg’s difficult debate performance last night.

That’s not to say the model has handled Bloomberg perfectly. His odds of winning a majority or plurality of pledged delegates1 were probably too low initially because Bloomberg is an unusual case. But those odds have also improved a lot in recent days. As of Thursday afternoon — before any post-debate polling was in — he had a 9 percent chance of winning a majority of pledged delegates and an 18 percent chance of getting a plurality of delegates in our forecast.

At the same time, the hype about Bloomberg — a candidate who had yet to compete in any states, to participate in any debates, or to face sustained scrutiny from the media and other candidates — had probably gotten out of hand. Prediction markets have his chances cut almost in half as a result of the debate, from about 30 percent before the debate on Wednesday to only around 15 percent as of early Thursday afternoon. That’s an awfully big correction after a single debate for which we don’t yet have any polling. It may reflect the fact that these markets — and by extension the conventional wisdom generally — had overestimated Bloomberg’s chances to begin with.

So let me advance a few interrelated propositions about Bloomberg:

  1. Bloomberg is unusual because of his money — but also his late start. That makes him hard to predict.

There has understandably been a lot of attention paid to the unprecedented amount of money that Bloomberg has spent on his campaign, especially on television advertising. And even if the evidence is mostly on the side of self-funded candidates not performing well, Bloomberg is spending so much money that he tests the boundaries of any potential equation involving campaign spending.

But Bloomberg is also unusual for his very late start to his campaign, which he launched on Nov. 24, as well as for his skipping the first four states. (He was actually eligible to receive votes in Iowa, but he didn’t campaign there.) In general, starting late has been a bearish indicator for presidential candidates — think about actor and former Sen. Fred Thompson, who flamed out after a lot of hype in 2008, or about former Massachusetts Gov. Deval Patrick, who also entered the race in November but who quit last week after getting 0.4 percent of the vote in New Hampshire. “Not competing” in early states has also been bearish — think about former New York Mayor Rudy Giuliani skipping Iowa in 2008 and how that turned out. Often, “not competing” has just been an excuse that candidates offer to lower expectations in states that they figure they’re going to lose.

Somehow, though, the combination of these factors allowed Bloomberg to continue gaining in the polls despite not contending in the early states. Both his paid media and his “earned” media — cable news and social media love talking about him (see point No. 2 below) — kept him in the conversation even when he wasn’t on the ballot. So when former Vice President Joe Biden cratered in the polls after Iowa, the biggest gainer was Bloomberg — not one of the candidates who had actually performed well in Iowa, namely former South Bend, Indiana, Mayor Pete Buttigieg and Sen. Bernie Sanders.

  1. Bloomberg’s recent polling surge is at least partially driven by news coverage. That opens him up to a “discovery, scrutiny, decline” cycle.

Bloomberg had risen slowly but somewhat steadily in the polls since his campaign launch, climbing from 3.6 percent in our national polling average on Dec. 12 to 8.8 percent on Feb. 3. That isn’t bad — a 5.2 percentage-point gain in 64 days — although it was short of the pace he’d need to be seriously competitive on Super Tuesday. If you had extrapolated out Bloomberg’s rate of increase — decidedly not a safe assumption! (see point No. 3) — he would have reached 11.2 percent in the polls by Super Tuesday, short of the usually 15 percent threshold that Democrats require a candidate to clear in order to receive state or district delegates.

Instead, Bloomberg had an abrupt, nonlinear surge in our polling average, climbing from 8.8 percent on Feb. 3 to 15.4 percent on Feb. 13, just 10 days later. He has since somewhat stalled out, for what it’s worth, having risen only to 16.1 percent as of Thursday afternoon.

This increase also happened to coincide with a big spike in news coverage of Bloomberg. I looked at how often candidates’ names appeared3 in headlines at Memeorandum, a site that aggregates which political stories are gaining the most traction, and found that from the Iowa caucuses on Feb. 3 through Thursday afternoon (Feb. 20), Bloomberg was the subject of 80 headlines at Memorandum, slightly trailing Sanders (84) but well ahead of Biden (53), Buttigieg (32), Sen. Amy Klobuchar (19) and Sen. Elizabeth Warren (15).

Now, not all of these headlines have been positive for Bloomberg, especially in recent days. But that’s sort of the point. It’s not uncommon for candidates to undergo what political scientists Lynn Vavreck and John Sides call a “discovery, scrutiny, decline” pattern in the polls, where an initial spark triggers a surge in media attention and a rise in the polls, but storylines turn more negative as the candidate gets more scrutiny and their actual performance doesn’t match the newfound hype. Candidates such as businessman Herman Cain and former Speaker of the House Newt Gingrich underwent this cycle in 2012. Sen. Kamala Harris and former Rep. Beto O’Rourke did so this year.

  1. Be wary of assuming there’s momentum in polls. Bloomberg may keep rising, or he may have already peaked.

Let’s pause for breath here. There’s no evidence yet that Bloomberg is about to enter the decline phase of a discovery-scrutiny-decline cycle. And even if he does, he may have a higher floor than other candidates because he’s running so many ads.

Still, it’s a huge mistake to assume that just because a candidate is rising in polls now, he or she will continue to do so. Empirically, polls come much closer to what statisticians call a “random walk” than to, say, an object in flight that gains or loses momentum. That is to say, if a candidate rises from say 9 percent to 16 percent, that candidate is about as likely to revert back to basically where they were before (to, say, 11 percent) as they are to continue rising (to, say, 21 percent). Or their numbers could flatten out.

There are some qualifications to this. Candidates can gain momentum from winning states, and candidates who fall in the polls are at risk of dropping out. But to a first approximation, political observers vastly overrate the importance of momentum in the polls. There are plenty of examples of that this year, involving not only Harris and O’Rourke but also Warren, whose summerlong rise in the polls abruptly turned into a decline in November, and Buttigieg, who has had several rises and falls in national polls.

  1. Bloomberg needs to improve his position by Super Tuesday to be a front-runner for the nomination. That’s possible, but it’s not a great bet.

But suppose Bloomberg doesn’t decline in the polls. Instead, he holds steady. After the reviews his debate got last night, his campaign might be happy to take that.

The problem is that merely holding steady in the polls through Super Tuesday would result in Bloomberg treading water in the delegate count. Here, as of early Thursday, are our model’s projected delegate counts after Super Tuesday, along with a high (90th percentile) and low (10th percentile) range for each candidate. Note, again, that these numbers don’t yet reflect any post-debate polling.

Sanders leads in projected delegates after Super Tuesday

Projected delegate averages and those delegate totals as a percentage of delegates awarded through Super Tuesday, according to FiveThirtyEight’s forecast as of Feb. 20

Projected delegates after Super Tuesday
Candidate Average Pct. Low Pct High Pct.
Sanders 590 39% 313 21% 856 57%
Biden 296 20 15 1 584 39
Bloomberg 287 19 56 4 521 35
Buttigieg 138 9 27 2 324 22
Warren 115 8 9 1 292 19
Klobuchar 59 4 10 1 118 8
Steyer 14 1 0 0 29 2
Gabbard 0 0 0 0 0 0

“Average” reflects the mean of 10,000 simulations. The “low” and “high” columns reflect the 10th and 90th percentile outcomes for each candidate, respectively.

If everyone exactly hits their averages, then after Super Tuesday, Sanders would have won about 40 percent of the delegates that had been awarded to that point, Biden and Bloomberg would have about 20 percent each, and the rest would be scattered between the remaining candidates (mostly Buttigieg and Warren).

Of course, we probably won’t wind up with those exact results. Candidates may rise and fall in the polls as a result of last night’s debate, or as a result of the voting outcomes in Nevada and South Carolina. One or more candidates could even drop out by then.

Still, 38 percent of all pledged delegates will have been handed out after Super Tuesday, so having only about a fifth of the delegates awarded to that point would make it nearly impossible for Bloomberg to get a majority of delegates later on. And it would be highly difficult for Bloomberg to even get a plurality with Sanders having about twice as many delegates as the former mayor. His chances at that point would probably depend on a contested convention.

But Bloomberg’s 90th percentile scenario, in which he winds up with about 35 percent of the delegates who had been handed out through Super Tuesday, would leave him in a much better position. He’d certainly have a good shot at winning the plurality of pledged delegates, for instance. A majority would require some more work, though it could be plausible depending on which of his opponents dropped out and how much he gained in the polls as a result of his Super Tuesday performance.

And pending that post-debate polling, those upside scenarios are still on the table. By definition, there’s a 1 in 10 chance that Bloomberg hits or improves on his 90th percentile forecast. Far more unlikely things have happened.

But the downside cases are equally likely. And as the debate exposed, if Bloomberg has some unique strengths as a candidate — his money, a smart team behind him, and a slightly Trumpian ability to command media attention — he also has some unique weaknesses. These include: his lack of polish as debater and public speaker, his past as a Republican, his status as a billionaire in the age of Sanders and Warren, his lack of practice as a candidate because of his campaign’s late start, New York’s use of the stop-and-frisk policy during his time as mayor and his relationship to black voters, his age (78), and the lewd comments he has allegedly made toward and about women. On top of that, we don’t know anything yet about how Bloomberg’s support in polls will translate into actual votes; as compared with most other Democrats, his voters are more likely to say they haven’t firmly committed to a candidate.

These are some big liabilities, and ones the media should have paid more attention to amidst the hype surrounding Bloomberg’s rise but for now we’ll just have to wait and see what the post-debate polls say.


  1. Remember our model does not forecast the outcome of the nomination per se because the model doesn’t try to predict what would happen in the event of a contested convention.

  2. That is, a week after his campaign officially began — enough to see voters’ initial reaction to it in polls.

  3. As of 1 p.m. each day.

Nate Silver is the founder and editor in chief of FiveThirtyEight.