Prior to cancelling my paid sub, I’d like to share my thoughts despite the potential “just leave” comments that will inevitably come.
I’ve consistently been vocal on here that Trump was comfortably in the lead, that the mainstream polls were once again wrong, that Seltzer’s poll was going to miss horribly, and that Nate’s model is flawed. Because of that, I was ridiculed and called names, which is unbecoming in this type of a newsletter.
The reason I’m sharing my thoughts is the earnest hope that it will lead to future improvements in the Silver Bulletin.
For, despite being a critic of Nate’s, I’ve been following him for quite a while, enjoying the past 538 articles on politics, sports and society.
I’ve been a paid sub and even purchased his book.
All this to lay the groundwork and show I’m not a troll.
Other groundwork to lay is my background. I share this to blunt whatever stereotypes critics will make of anyone who is a Trump supporter, which I am one of.
Raised in Canada, lived in Southern California for 25 years. Bachelor’s in Poli Sci and Psych. MBA. Worked in the corporate world. For the past 20 years, founded and run a humble small business. Taiwanese-American.
I had high hopes for the Silver Bulletin, but as time went on and we approached Election Day, it was clear to me that what I was reading wasn’t what I was hoping for.
All the signs were there for a massive Trump win. Yes, Nate acknowledged the possibility as much, but his algorithm really didn’t as shown by his 50/50 result after tens of thousands of simulations.
The fatal flaw, from my humble opinion, is his reliance on polls. If the polls are wrong, his model will be wrong. I realize he incorporates other factors, but the primary driver are the polls.
And this cycle, again, the “right wing” polls were all extremely accurate.
The news media and university polls missed…some by quite a bit. Again.
But honestly, you didn’t need these polls to know Trump was a heavy favorite.
All the key indicators pointed to Trump having the massive advantage:
-Wrong track was at 28%. As Harry Enten pointed out, no president wins with those numbers.
-Party ID by Pew and Gallup showed a net swing from D+4 to R+3. That’s 7 points. Party ID has a strong correlation with who wins.
-Economy / cost of living, and immigration were consistently the top issues, which Harris ran behind on.
-Harris simply would never do the same or better than Biden in the Rust Belt. Scranton Joe vs far left California Kamala. That alone should have disqualified her from consideration to run.
-Harris could never detach herself from the horrible numbers of Biden. She was trying to convince voters she will bring change, but could never answer the fundamental question of why she didn’t enact change as VP.
Then as the polls rolled out, we consistently saw Trump overperforming his 2020 numbers by 4 to 10 points.
We saw these in swing states but also in blue states.
We saw him tie the national average, which signaled a massive EC victory.
We saw him up with almost every demo: blacks, hispanics, Catholics, Jews, Arabs/Muslims, union workers, young men.
Add up all the above, and it was clear Trump should have been a 2-1 favorite at least, which coincidentally was about what the betting markets trended towards in the final days.
Yet despite all of the above, those on the left were oblivious and kept trotting out “advantages” Harris had which we now know were simply non-existent. Yet many on the left here, on blue X, commenters on Daily Kos or Slate, were convinced Harris was trending to a massive EC victory with states like FL and TX going blue.
Ann Seltzer’s historically disastrous poll that missed by an astonishing 16 points only added to their false bravado.
Because of the above, I saw an arbitrage opportunity to profit, and placed multiple bets on Kalshi.
I went 5 for 5 betting on: Trump to win, GOP to sweep Presidency / House / Senate, Trump to win the popular vote, Trump to win about 312, McCormick to win the Senate.
Those profitable bets were helped by data from Rich Baris at Big Data Poll, host of Inside the Numbers, and Robert Barnes, professional political and sports better. The info they provided were far more accurate and dispassionate than Nate’s. Eg. They didn’t give hopium that Trump would win states like VA, NH, etc.
Contrast that to Nate’s insistence that FL could NEVER go +8 for Trump, in his infamous scuffle with Keith Rabois. Nate is lucky Keith didn’t take him up on his $100,000 wager. Because not only did Trump win by 8, he won by a staggering 13.2%, flipping Miami-Dade solid red, exceeding the 10-12 that Keith predicted.
Yet there’s no acknowledgement of being wrong from Nate, on this or other aspects.
To end my much-too-long diatribe, I’m leaving some suggestions that could help improve the Bulletin:
-Take a long, hard look at how you weight polls. I kept saying, and no one had a good retort, that it was disastrous to give a poll like Quinnipiac an overweight of 1.36. Quinnipiac missed 2020 by an average of 7, and 100% in favor of Dems. They should have been disqualified from that alone, but instead Nate actually values them higher. This cycle, Quinnipiac was right only 40% of the time, worse than a coin flip.
Same goes for polls like Washington Post, which was right only 33% of the time.
Conversely, the right wing polls were right:
Atlas Intel 100%
Big Data Poll 90%
PollFair 86%
Rasmussen 86%
Quantus Insights 86%
Trafalgar 71%
Insider Advantage 71%
Left wing pundits like Litchmann, Sabato, etc were also disastrously wrong.
-I hate the 13 keys as well, but there’s something to the meta-themes that polls can’t capture. Again, I know Nate tries to incorporate some of this, but perhaps look to reduce the importance of polls and increase the importance of other factors.
-Dissect where his left wing lens may have blinded him to what was right in front of his eyes. None of us are non-partisan, but being too partisan can make us see anything we want in the data.
For those commenters who ridiculed me, called me nasty names, etc., I don’t really hold any animus. But I do hope this gave you a dose of humility, to learn that sometimes when you’re in a bubble, you can’t see that you’re in a bubble.
Despite cancelling my paid sub, I’ll still follow Nate, and I do wish him well. There’s absolutely a place for a newsletter like his that can add value to the public discourse.
Dude, the polls will likely have been off by about 2 1/2 points, which is less than the typical disparity going back to the 1960s. In other words, they were pretty accurate by historical standards. And if you don’t consult poll data, you’re relying on vibes.
EDIT: And meant to add, I thought avoiding a reliance on vibes was pretty much the whole point of subscribing to Nate's blog!
I'm not saying you're wrong broadly speaking, but the polls were off by considerably more than that in some very notable instances (e.g., Iowa (courtesy of the gold standard of polling, I'm told), or the other handful of states that were outside of the confidence interval in the model). There's also the fact that polls have underestimated Trump's performance three elections in a row, to the point that most seemed to take it as a given that the polls would skew leftward (see betting markets). I'm not saying any of that is definitive, and I generally defer to Nate's more reasoned insights. But I take the comment to be arguing for some weighting of non-polling factors and accounting for bias. Nate already does this to some degree, and I don't think that the commenter's criticisms are too ridiculous.
>...the polls were off by considerably more than that in some very notable instances...<
There are always outliers. I bet some polls thought Hoover was going to win in 1932. I assumed nearly all Silver readers were far more interested in aggregates.
Anyway, the polling consensus seemed to have converged on Harris +1.2 or so on the eve of the election. A Trump PV victory of roughly two points (or a Harris victory of roughly four points) was consistent with history, based on what the polls were telling us. I'm not going to claim I knew *in which direction* the polls were going to err: I thought it was a coin toss, as each campaign had elements they could point to as good signs.
I'm not looking to argue with you about whether or not the aggregates were good or bad this year. That doesn't seem useful or interesting. And as I said, I wasn't even disputing your general point. I was making a defense of the commenter's critiques, which seemed reasonable. But I always think it's helpful to do that, even if it doesn't immediately seem like there's much there.
It may not be useful to you, but it's useful to me in responding to the commenter's critique, which seems to accusing Nate of being badly off, because (I quote) "All the signs were there for a massive Trump win."
This seems deeply mistaken, because all the signs weren't like that at all. (And that's even assuming what we saw was a "massive" Trump win, which itself seems highly questionable: Trump's going to have one of the smallest PV margins ever for a second term winner, and he looks like he will have *lost* House seats!)
Anyway, my point is: contra the OP, the polling we had made this a challenging race to call.
It was massive by most metrics. First PV win by a GOP since 2004. He won all the swing states, increased his margins in FL and TX, and cut the D lead in blue states. Kamala did not do better than Biden in a single county. That's astounding.
Since the beginning of the Trump era (2016), the polls have been wrong for the presidential election years, but fairly accurate for the midterm elections (2018 and 2022). This might be because low-propensity voters (who are hard to reach in polls) are excited by Trump but not excited enough to vote in the midterm elections without Trump on the ballot. Or maybe because of recent events in the midterm years (white supremacist rallies in 2017-18, Donna in 2022) that got the Democratic base energized and excited to vote.
The more education a voter has the more likely they are to vote. As the two parties trade bases that means that Republicans--who used to win the college educated vote--now do better in general elections while the Democrats do better in midterms and special elections. That's the exact opposite of the status quo from a decade ago.
I agree, directionally. Playing devil's advocate though, it seems not super neat and obvious as a blanket statement. For example, many have explained last week's result as, at least in part, a problem of democratic voters not showing up (e.g., "Trump got X more votes, sure, but Kamala got Y less votes, and that's the REAL story"). There may be something to that, but it would be inconsistent with a world in which dems are now the party of high-turnout voters. Who knows lol
One hypothesis: hardcore D partisans turned out. Less devoted Democrats were discouraged by the economy, illegal immigration and crime and stayed home.
Agreed. The polls were quite accurate, well within their margin of error. Can't expect much more. Sure, Ann Selzer's polls was widely off at the end. However, in reflection, the pollsters can have a drink and celebrate. They did a better job than I was expecting.
A simple yardstick would be to look at party ID which swung from D+4 to R+3. That alone could have been a far more reliable indicator that Trump was the heavy, heavy favorite. Biden needed +4 PV win to barely eke out the EC. If I just looked at this, and ignored all the poll, I likely would have been far more accurate in deciding Trump should be a 2-1 or 3-1 favorite. Sometimes too much data is too much data.
I said GIGO before the election. You can normalize and manipulate the data all you wish, but garbage in, garbage out. Polling is broken due to low response rates from most of the electorate (and subsequent over-response of college educated Democrats) and the unwillingness of cheap public polls to take measures to combat this. There was a time when poll aggregation solved the problem, but that ended after 2008, in my opinion. The RCP average had Harris winning the popular vote, mostly because of a bunch of polls that misread the electorate. The individual states got closer, but point at each state and you'll find a few pollsters that produced wildly out of whack results time and time again that influenced the averages. Those CNN polls that dropped the last week showing +5 and +6 leads in the Rust Belt for Harris...400 LV. Crazy. Nate's averages were even more biased to Harris than the RCP ones, mainly because of his weightings.
Some better analysis than this model's inputs is required to come up with a reliable picture.
Look dude, you seem to be genuinely trying have a discussion in good faith, so I will try and do the same. The internet was just chock full of a 100,000 takes just like this for why “Trump is the clear favorite” and “Harris in a landslide” and you know what? Many of them were hot garbage from both sides, echo chamber bullshit, but there were quite a few well done, clearly articulated, and well reasoned arguments why both candidates should win. And now your guy won and you can crow about it if you want but it doesn’t make your take any more special. Hindsight is 20/20 and no one really knew a damn thing for sure until Election Day came. I came to the sliver blog not because he’s Nostradamus, but because it was a refuse from the false certainty arrogance that is ubiquitous in nearly every other area of the internet. It’s tiring. 8 billion people in the world and they are so god damn cocky. I just wanted one place where there was an acknowledgement of the uncertainty.
Also, America’s peculiar form of democracy has absolutely skewed our perceptions. 2% in the tipping point state is simply a close a fkn election, it just happens to be enough for a huge electoral victory and a clean sweep of congress. Yes, on many fronts the democrats got their ass handed to them, which is a combination of a global backlash and the Democratic having their head thoroughly stuck up their own ass. But at the end of the day, the model predicted nearly even split in the tipping point state (which it correctly predicted to be PA) and it was off by 2%, which was well within 1 standard deviation of the mean.
I get it, there are big time republican blowout vibes right now, and there is good reason. In nearly every region and every demographic voters made their displeasure with the Democratic Party known. I’m not here to repudiate the fact that this was an absolute romp. The democratic brand is dog 💩 right now and they need to do a complete top to bottom house cleaning. It’s just that, the margin of the tipping point state was 2%, so from an electoral college perspective, it was very close and the model properly prepared us for this.
> All the signs were there for a massive Trump win.
I'm not sure what you are talking about, but in any case, does that mean you realize you were wrong now, because there was no "massive Trump win"? This was a 2 point squeaker, that could just as easily have gone the other way.
“Trump over performed virtually everywhere. The polls didn’t capture that.”
When you say “overperformed” what I think you mean is “had better numbers than in 2020.” And the polls absolutely captured that. Biden won the PV by 4.5 points in 2020. Most national polls showed essentially a tied race this go around. So that means the polls captured a 4.5 point shift to Trump.
You want to get all up in your feelings because Trump is going to win the popular vote by about 1 point? OK, so the polls captured a 4.5 point swing instead of a 5.5 point swing. But don’t sit here and tell me the polls predicted something else.
I looked at the 17 most recent national polls included in the RCP average. While I think you and I can both agree in polling there isn't a statistically significant difference (with a standard MOE of 3, those numbers mean very little) between Harris +1, tied, and Trump +1, for the sake of playing your game let's say there is.
-5 polls showed Trump winning the popular vote
-5 polls showed the race tied
-7 polls showed Harris winning the popular vote
So 10/17 polls showed either Trump winning or the race tied, when the final PV margin is going to be Trump +1. That's quite good! Don't sit here and get butt hurt and cry because the data isn't supporting the conclusion you want to have.
The RCP average still had Harris up from what I recall, and here Silver had Harris heavily favored to win the popular vote. Trump beat his polling for the third consecutive election.
Again, currently Trump is ahead in the popular vote by about two points, not one. Additionally look at the number of counties where he did better compared to 2020 versus the counties where Harris improved on Biden.
Squeaker really? Trump won popular vote which not even Nate was predicting. NY was won by democrats by less than Trump.win TX and FL. NJ was <4. Complete sweep of swing states. This was a mandate and such a polarized environment This was a landslide. I think mid terms will be interesting. What if republicans keep.hold of the house in 2026?
That Trump won the popular vote was definitely somewhere in the tail of the distribution (as opposed to the center). But exceeding expectations doesn't make something a landslide lol, it makes it an 'upset'.
When all is said and done, Trump is going to have won the popular vote by about 2 points.
By contrast, Biden won it by 4.5 points, Obama by 4 points in 2012, and 7 points in 2008. Clinton won by 5.5 and 8.5, HW by 7, and Reagan won the popular vote by 10 points in 1980, and 18 points (!!) in '84.
If we compare these results to the past 50 years of elections, it definitely looks like a squeaker, albeit an unexpected one. It only looks like a 'landslide' if we compare to exactly Bush Jr. or Trump's first campaign – neither of which are particularly high bars.
Trump was like half a percentage point popular vote worth of votes away from turning NJ, VA, NH, and MN, Maine, and NM red and creating a blood bath map 360 EVs
They're inventing some world where the vote counts in all states other than the 6 they mentioned stay the same, and then adding up the margins in those 6 states to get some total that I suppose comes to 0.5% of the *national* electorate.
TL;DR: It's really bad math, and using that same logic we could say that Kamala wins the election with just 0.2% more of the popular vote – so long as that 0.2% was entirely in the blue wall (mostly PA)
Yeah but is Kamala the one who had a 7 point national swing?
I can't believe none of you are seeing the electoral vulnerability in those states in the future lol. I guess the idea is too scary to process right now. Fair enough.
My point here is to show that for every person who copes by saying Kamala was a hair away in the rust belt, don't forget trump ended up close in a whole bunch of not very swingy states, swinging many of them by double digit margins
"Compared to this one rock falling over, these two rocks falling over was def a landslide!"
Also Trump just won the PV by less than half as much as Biden, maybe a third as much, and got 6 more electoral votes. Just... what are you talking about?
Maybe you should be more than a little alarmed at New Jersey nearly becoming a swing state this cycle instead of clinging onto the idea that this election wasn’t a foreboding sign for democrats. Trump got thrashed in the public sphere for years and still managed to decisively beat Kamala; that should say a lot.
Neither election was a mandate election. Both elections repudiated the performance of the incumbent president. Both elections had historically narrow popular vote margins. Both elections had surprisingly strong performances in states that were historically out of reach. Both elections narrowly shifted control of the Senate toward the winning presidential party but lost ground in the House. Both elections ended with about 310 electoral votes (the difference mostly because of the changed number of electoral votes in battleground states after the 2020 census), and all seven 2024 battleground states were won by the winning candidate in both elections.
These two elections were almost perfect mirror images of each other, and either you must conclude that both were consequentially large victories by one party or the other, or you must conclude that both are narrow victories magnified by an electoral system that exaggerates the swings in political influence from small changes in election results. Any other interpretation is evidence of partisan wishfulfillment analysis.
For the record, I have consistently maintained that Biden did not win any form of governing mandate in 2020. The message voters sent in 2020 was only that they were not ready for another four years of Trump after suffering through the first 10 months of COVID-19 and the civil unrest following the murder of George Floyd.
President Biden massively misread the 2020 results and foolishly allowed the progressive wing of his party to pursue an extremely ambitious legislative agenda as if he had received a true electoral rout and mandate. Your comments and what we are already seeing in the first week after the election suggest that President-elect Trump is about to make the same mistaken interpretation of his election results and also overplay his political hand.
This will result in an ever-so-predictable backlash in the 2026 midterms, with the most favorable Senate map of the three Senate cycles and a House with only a very narrow Republican margin. This is a clear recipe for swinging both houses of Congress, much like the 2006 midterm cycle.
The big question is whether the Democratic Party can find a fresh new face with a fresh new vision for the party, like Obama provided in 2008. My two best outsider candidates to fulfill this role of a fresh new voice for the party are Mark Cuban and Jamie Dimon, both highly respected, politically savvy, high-profile public figures with no allegiance or baggage from legacy Democratic politics and the freedom to redefine the image of the party. Both are extremely successful businessmen with strong centrist credentials and the independence to craft a new vision for the party much as Bill Clinton did in 1992 and Obama did in 2008.
A mandate for what though? In Biden's case it was clearly a mandate for a return to the old status quo. The critical point is that there was a mismatch between what the public wanted and what the Biden admin wanted.
That's why this question is meaningless. Are the results a mandate for mass deportations? I would argue that the answer is yes, absolutely. A mandate for more troops in Syria? No, the exact opposite.
To call this a "squeaker" is a bit much. By today's standards, this was a solid win. Not a blowout, but decisive.
With so much of the electorate permanently committed to one of the two parties, you're never going to see things like Reagan '84. Obama '08 is about as big as it gets.
Like the man said, democracy is the proposition that the people know what they want and deserve to get it good and hard.
It's often used but I think it's harmful to use "today's standards" to characterize what are historically very close races as a "solid win" or (as if often done but not what you did here, a "landslide"). We are in an era of nearly 50/50 split of the voting public of the two parties. If a <= 3% shift in the votes would have changed the outcome, then it was a squeaker.
I didn't pick the number 3% out of thin air. In 1980 pollster Pat Caddell was asked why he said the election was a flat footed tie and Reagan won by 3%. He explained: The Carter polls showed Carter down by 3%. But it had a margin of error of 3%. So they assumed they were tied when, with hindsight, they were probably down by 6%. And, back then when people really decide at the last moment and we didn't expect pollsters to exactly predict that, bad news the weekend before the election moved the voters another 3%. Resulting in what Reagan would call "a mandate" with only a 9% margin of victory.
Like in 2020, this was a very close election with a clear winner. But the fundamentals of the voting public of the two parties being split around the 50%/50% has not changed. Could it be the start of a trend: Trump lost by 4.5% four years ago and won by 1% this time... maybe. But given the seasaw nature of the last several elections, probably not.
It was massive by most metrics. First PV win by a GOP since 2004. He won all the swing states, increased his margins in FL and TX, and cut the D lead in blue states. Kamala did not do better than Biden in a single county. That's astounding.
2024 was almost an exact mirror image of 2020. Neither election was a mandate election. Both elections repudiated the performance of the incumbent president. Both elections had historically narrow popular vote margins. Both elections had surprisingly strong performances in states that were historically out of reach. Both elections narrowly shifted control of the Senate toward the winning presidential party but lost ground in the House. Both elections ended with about 310 electoral votes (the difference mostly because of the changed number of electoral votes in battleground states after the 2020 census), and all seven 2024 battleground states were won by the winning candidate in both elections.
These two elections were almost perfect mirror images of each other, and either you must conclude that both were consequentially large victories by one party or the other, or you must conclude that both are narrow victories magnified by an electoral system that exaggerates the swings in political influence from small changes in election results. Any other interpretation is evidence of partisan wishfulfillment analysis.
For the record, I have consistently maintained that Biden did not win any form of governing mandate in 2020. The message voters sent in 2020 was only that they were not ready for another four years of Trump after suffering through the first 10 months of COVID-19 and the civil unrest following the murder of George Floyd.
President Biden massively misread the 2020 results and foolishly allowed the progressive wing of his party to pursue an extremely ambitious legislative agenda as if he had received a true electoral rout and mandate. Your comments and what we are already seeing in the first week after the election suggest that President-elect Trump is about to make the exact same mistaken interpretation of his election results and also overplay his political hand.
This will result in an ever-so-predictable backlash in the 2026 midterms, with the most favorable Senate map of the three Senate cycles and a House with only a very narrow Republican margin. This is a clear recipe for swinging both houses of Congress, much like the 2006 midterm cycle.
The big question is whether the Democratic Party can find a fresh new face with a fresh new vision for the party, like Obama provided in 2008. My two best candidates to fulfill this role are Mark Cuban or Jamie Dimon, both highly respected, politically savvy, high-profile public figures with no allegiance or baggage from legacy Democratic politics and the freedom to redefine the image of the party. Both are extremely successful businessmen with strong centrist credentials.
Even if Trump were overwhelmingly popular odds are that the Dems would take the House in 2026. That's just the normal pattern for the first midterm after a presidential election--see Clinton, Obama, Trump and Biden.
In the past, a party flipping the WH by winning both the electoral and popular votes also has long coattails and pick up lots of House seats. And then the midterms are a regression to the mean.
But so far this decade that is not happening. The GOP picked up 5 seats in 2020. And then another 5 in 2022. And this time they will be right around net zero. We're around a stability point. So I think there would have to be an unusually strong pro or anti GOP wave for 2026 to see much of a swich either way. Which means the Democrats chances of flipping the House in 2026 is probably less than, or no better than, 50/50. (Maybe 50/50 since the party out of power may be more likely to show up at midterms. But there just not as many swing districts as in prior decades.)
This is the best take I’ve seen anywhere on the internet about the last two elections. (and I don’t just say that because you’re predicting a Dem win in 2026!). Thanks for sharing.
Didn’t Harris do better than Biden in the counties around Atlanta, in western North Carolina, on the northwest tip of Michigan’s lower peninsula, and several other regions?
This was a 2004 style election. A clear win, but nothing like “massive”. There hasn’t been a “massive” win since 1984. This is smaller than Obama’s 2008 win, which was the only large win in decades.
The prior election is mostly irrelevant here. I say "mostly" because you do have to account for factors like polarization; victory margins on both sides are much smaller today than they used to be.
But you don't get bonus points for improving on your previous showing. And you certainly shouldn't assume those same trends will continue. Trump was the face of the pandemic in 2020. Biden and Harris were the face of inflation in 2024. The Republican nominee in '28 will be the face of whatever is top of mind that year -- good or bad.
Maaaaybe. Elections are about more than one thing--the Trump team is convinced that their best ad was the one attacking Harris for supporting sex change operations for illegal immigrants in federal custody. I would not be surprised if the Dems are out of the executive until they pull a Sista Souljah and put some distance between themselves and trans activists: "No boys in girls' sports".
I disagree with your second half. No chance the low prop voters turn out for a RINO like Nikki. In fact, I'm doubtful the low prop voters will turn out for anyone besides Trump. That's a worry the GOP will need to address sooner rather than alter.
I wouldn't call Haley a RINO - she is a proud graduate of a segregation academy, and is definitely MAGA adjacent. If Haley and people like her leave the Rs the GOP will be in a world of hurt.
That is accurate. But Nate need to lower the ratings for left wing polls of morning consult, Washington post, quinipiac the same way he does for right wing polls of trafalgar, Rasmussen. He was actually saved a little by right wing pollsters.
But they were more right in 2020. Quinipiac, Washington post and morning consult were more wrong in 2020 and now in 2024. So.shouldnt they be weighted down going forward as well?
He gave WaPo an A+ and Quinnipiac an A-. Both polls were trash in 2020 and 2024. That alone makes me question Nate's capabilities. If I didn't tell you the names of the polls but just showed you how much they were off by, and how often they favored Dems, you'd say toss them out.
Do you have any justification for why he added 1 point to the popular vote estimate his own model spit out in order to "sync them up" with her performance in the battleground states?
It was an entirely unscientific endeavor, taking a stab that national polls were on the aggregate underestimating Kamala because of disproportionately better battleground polling when the simpler answer could have been that the Pop vote EC bias disappeared (a point he conceded in follow up)
None of this is entirely mathematical, scientific or objective. To some degree his own model does not rely only on polling. He himself states there's some fundamentals integrated into it
The simple average of polls at RCP has performed better than Silver's weighted model for three presidential elections in a row I believe. They're poll based too--that's not the issue.
I can read your comments (better than you can, apparently). You said lots of people would tell you to just leave. I didn’t want you to be wrong about that as you are about so much else. You’re welcome.
This was a clear trump win, but you guys need to stop convincing yourself it was some "massive" win. Reps are acting the same right now as Dems did in 2020 and I don't think it will end well for them.
It was massive by most metrics. First PV win by a GOP since 2004. He won all the swing states, increased his margins in FL and TX, and cut the D lead in blue states. Kamala did not do better than Biden in a single county. That's astounding. But you're right, 2026 could easily flip the House back to D.
TLDR, but did you explain somewhere in there how polls or the model were supposed to be "wrong" while getting the outcome right in the fat part of the distribution/calling the most likely state map?
Prior to cancelling my paid sub, I’d like to share my thoughts despite the potential “just leave” comments that will inevitably come.
I’ve consistently been vocal on here that Trump was comfortably in the lead, that the mainstream polls were once again wrong, that Seltzer’s poll was going to miss horribly, and that Nate’s model is flawed. Because of that, I was ridiculed and called names, which is unbecoming in this type of a newsletter.
The reason I’m sharing my thoughts is the earnest hope that it will lead to future improvements in the Silver Bulletin.
For, despite being a critic of Nate’s, I’ve been following him for quite a while, enjoying the past 538 articles on politics, sports and society.
I’ve been a paid sub and even purchased his book.
All this to lay the groundwork and show I’m not a troll.
Other groundwork to lay is my background. I share this to blunt whatever stereotypes critics will make of anyone who is a Trump supporter, which I am one of.
Raised in Canada, lived in Southern California for 25 years. Bachelor’s in Poli Sci and Psych. MBA. Worked in the corporate world. For the past 20 years, founded and run a humble small business. Taiwanese-American.
I had high hopes for the Silver Bulletin, but as time went on and we approached Election Day, it was clear to me that what I was reading wasn’t what I was hoping for.
All the signs were there for a massive Trump win. Yes, Nate acknowledged the possibility as much, but his algorithm really didn’t as shown by his 50/50 result after tens of thousands of simulations.
The fatal flaw, from my humble opinion, is his reliance on polls. If the polls are wrong, his model will be wrong. I realize he incorporates other factors, but the primary driver are the polls.
And this cycle, again, the “right wing” polls were all extremely accurate.
The news media and university polls missed…some by quite a bit. Again.
But honestly, you didn’t need these polls to know Trump was a heavy favorite.
All the key indicators pointed to Trump having the massive advantage:
-Wrong track was at 28%. As Harry Enten pointed out, no president wins with those numbers.
-Party ID by Pew and Gallup showed a net swing from D+4 to R+3. That’s 7 points. Party ID has a strong correlation with who wins.
-Economy / cost of living, and immigration were consistently the top issues, which Harris ran behind on.
-Harris simply would never do the same or better than Biden in the Rust Belt. Scranton Joe vs far left California Kamala. That alone should have disqualified her from consideration to run.
-Harris could never detach herself from the horrible numbers of Biden. She was trying to convince voters she will bring change, but could never answer the fundamental question of why she didn’t enact change as VP.
Then as the polls rolled out, we consistently saw Trump overperforming his 2020 numbers by 4 to 10 points.
We saw these in swing states but also in blue states.
We saw him tie the national average, which signaled a massive EC victory.
We saw him up with almost every demo: blacks, hispanics, Catholics, Jews, Arabs/Muslims, union workers, young men.
Add up all the above, and it was clear Trump should have been a 2-1 favorite at least, which coincidentally was about what the betting markets trended towards in the final days.
Yet despite all of the above, those on the left were oblivious and kept trotting out “advantages” Harris had which we now know were simply non-existent. Yet many on the left here, on blue X, commenters on Daily Kos or Slate, were convinced Harris was trending to a massive EC victory with states like FL and TX going blue.
Ann Seltzer’s historically disastrous poll that missed by an astonishing 16 points only added to their false bravado.
Because of the above, I saw an arbitrage opportunity to profit, and placed multiple bets on Kalshi.
I went 5 for 5 betting on: Trump to win, GOP to sweep Presidency / House / Senate, Trump to win the popular vote, Trump to win about 312, McCormick to win the Senate.
Those profitable bets were helped by data from Rich Baris at Big Data Poll, host of Inside the Numbers, and Robert Barnes, professional political and sports better. The info they provided were far more accurate and dispassionate than Nate’s. Eg. They didn’t give hopium that Trump would win states like VA, NH, etc.
Contrast that to Nate’s insistence that FL could NEVER go +8 for Trump, in his infamous scuffle with Keith Rabois. Nate is lucky Keith didn’t take him up on his $100,000 wager. Because not only did Trump win by 8, he won by a staggering 13.2%, flipping Miami-Dade solid red, exceeding the 10-12 that Keith predicted.
Yet there’s no acknowledgement of being wrong from Nate, on this or other aspects.
To end my much-too-long diatribe, I’m leaving some suggestions that could help improve the Bulletin:
-Take a long, hard look at how you weight polls. I kept saying, and no one had a good retort, that it was disastrous to give a poll like Quinnipiac an overweight of 1.36. Quinnipiac missed 2020 by an average of 7, and 100% in favor of Dems. They should have been disqualified from that alone, but instead Nate actually values them higher. This cycle, Quinnipiac was right only 40% of the time, worse than a coin flip.
Same goes for polls like Washington Post, which was right only 33% of the time.
Conversely, the right wing polls were right:
Atlas Intel 100%
Big Data Poll 90%
PollFair 86%
Rasmussen 86%
Quantus Insights 86%
Trafalgar 71%
Insider Advantage 71%
Left wing pundits like Litchmann, Sabato, etc were also disastrously wrong.
-I hate the 13 keys as well, but there’s something to the meta-themes that polls can’t capture. Again, I know Nate tries to incorporate some of this, but perhaps look to reduce the importance of polls and increase the importance of other factors.
-Dissect where his left wing lens may have blinded him to what was right in front of his eyes. None of us are non-partisan, but being too partisan can make us see anything we want in the data.
For those commenters who ridiculed me, called me nasty names, etc., I don’t really hold any animus. But I do hope this gave you a dose of humility, to learn that sometimes when you’re in a bubble, you can’t see that you’re in a bubble.
Despite cancelling my paid sub, I’ll still follow Nate, and I do wish him well. There’s absolutely a place for a newsletter like his that can add value to the public discourse.
Dude, the polls will likely have been off by about 2 1/2 points, which is less than the typical disparity going back to the 1960s. In other words, they were pretty accurate by historical standards. And if you don’t consult poll data, you’re relying on vibes.
EDIT: And meant to add, I thought avoiding a reliance on vibes was pretty much the whole point of subscribing to Nate's blog!
I'm not saying you're wrong broadly speaking, but the polls were off by considerably more than that in some very notable instances (e.g., Iowa (courtesy of the gold standard of polling, I'm told), or the other handful of states that were outside of the confidence interval in the model). There's also the fact that polls have underestimated Trump's performance three elections in a row, to the point that most seemed to take it as a given that the polls would skew leftward (see betting markets). I'm not saying any of that is definitive, and I generally defer to Nate's more reasoned insights. But I take the comment to be arguing for some weighting of non-polling factors and accounting for bias. Nate already does this to some degree, and I don't think that the commenter's criticisms are too ridiculous.
>...the polls were off by considerably more than that in some very notable instances...<
There are always outliers. I bet some polls thought Hoover was going to win in 1932. I assumed nearly all Silver readers were far more interested in aggregates.
Anyway, the polling consensus seemed to have converged on Harris +1.2 or so on the eve of the election. A Trump PV victory of roughly two points (or a Harris victory of roughly four points) was consistent with history, based on what the polls were telling us. I'm not going to claim I knew *in which direction* the polls were going to err: I thought it was a coin toss, as each campaign had elements they could point to as good signs.
I'm not looking to argue with you about whether or not the aggregates were good or bad this year. That doesn't seem useful or interesting. And as I said, I wasn't even disputing your general point. I was making a defense of the commenter's critiques, which seemed reasonable. But I always think it's helpful to do that, even if it doesn't immediately seem like there's much there.
>That doesn't seem useful or interesting.<
It may not be useful to you, but it's useful to me in responding to the commenter's critique, which seems to accusing Nate of being badly off, because (I quote) "All the signs were there for a massive Trump win."
This seems deeply mistaken, because all the signs weren't like that at all. (And that's even assuming what we saw was a "massive" Trump win, which itself seems highly questionable: Trump's going to have one of the smallest PV margins ever for a second term winner, and he looks like he will have *lost* House seats!)
Anyway, my point is: contra the OP, the polling we had made this a challenging race to call.
I think this is the third presidential election in a row where RCP's simple average of polls did better than Silver's weighted model.
It was massive by most metrics. First PV win by a GOP since 2004. He won all the swing states, increased his margins in FL and TX, and cut the D lead in blue states. Kamala did not do better than Biden in a single county. That's astounding.
OK thx
Since the beginning of the Trump era (2016), the polls have been wrong for the presidential election years, but fairly accurate for the midterm elections (2018 and 2022). This might be because low-propensity voters (who are hard to reach in polls) are excited by Trump but not excited enough to vote in the midterm elections without Trump on the ballot. Or maybe because of recent events in the midterm years (white supremacist rallies in 2017-18, Donna in 2022) that got the Democratic base energized and excited to vote.
The more education a voter has the more likely they are to vote. As the two parties trade bases that means that Republicans--who used to win the college educated vote--now do better in general elections while the Democrats do better in midterms and special elections. That's the exact opposite of the status quo from a decade ago.
I agree, directionally. Playing devil's advocate though, it seems not super neat and obvious as a blanket statement. For example, many have explained last week's result as, at least in part, a problem of democratic voters not showing up (e.g., "Trump got X more votes, sure, but Kamala got Y less votes, and that's the REAL story"). There may be something to that, but it would be inconsistent with a world in which dems are now the party of high-turnout voters. Who knows lol
One hypothesis: hardcore D partisans turned out. Less devoted Democrats were discouraged by the economy, illegal immigration and crime and stayed home.
Agreed. The polls were quite accurate, well within their margin of error. Can't expect much more. Sure, Ann Selzer's polls was widely off at the end. However, in reflection, the pollsters can have a drink and celebrate. They did a better job than I was expecting.
A simple yardstick would be to look at party ID which swung from D+4 to R+3. That alone could have been a far more reliable indicator that Trump was the heavy, heavy favorite. Biden needed +4 PV win to barely eke out the EC. If I just looked at this, and ignored all the poll, I likely would have been far more accurate in deciding Trump should be a 2-1 or 3-1 favorite. Sometimes too much data is too much data.
I said GIGO before the election. You can normalize and manipulate the data all you wish, but garbage in, garbage out. Polling is broken due to low response rates from most of the electorate (and subsequent over-response of college educated Democrats) and the unwillingness of cheap public polls to take measures to combat this. There was a time when poll aggregation solved the problem, but that ended after 2008, in my opinion. The RCP average had Harris winning the popular vote, mostly because of a bunch of polls that misread the electorate. The individual states got closer, but point at each state and you'll find a few pollsters that produced wildly out of whack results time and time again that influenced the averages. Those CNN polls that dropped the last week showing +5 and +6 leads in the Rust Belt for Harris...400 LV. Crazy. Nate's averages were even more biased to Harris than the RCP ones, mainly because of his weightings.
Some better analysis than this model's inputs is required to come up with a reliable picture.
Look dude, you seem to be genuinely trying have a discussion in good faith, so I will try and do the same. The internet was just chock full of a 100,000 takes just like this for why “Trump is the clear favorite” and “Harris in a landslide” and you know what? Many of them were hot garbage from both sides, echo chamber bullshit, but there were quite a few well done, clearly articulated, and well reasoned arguments why both candidates should win. And now your guy won and you can crow about it if you want but it doesn’t make your take any more special. Hindsight is 20/20 and no one really knew a damn thing for sure until Election Day came. I came to the sliver blog not because he’s Nostradamus, but because it was a refuse from the false certainty arrogance that is ubiquitous in nearly every other area of the internet. It’s tiring. 8 billion people in the world and they are so god damn cocky. I just wanted one place where there was an acknowledgement of the uncertainty.
Also, America’s peculiar form of democracy has absolutely skewed our perceptions. 2% in the tipping point state is simply a close a fkn election, it just happens to be enough for a huge electoral victory and a clean sweep of congress. Yes, on many fronts the democrats got their ass handed to them, which is a combination of a global backlash and the Democratic having their head thoroughly stuck up their own ass. But at the end of the day, the model predicted nearly even split in the tipping point state (which it correctly predicted to be PA) and it was off by 2%, which was well within 1 standard deviation of the mean.
I get it, there are big time republican blowout vibes right now, and there is good reason. In nearly every region and every demographic voters made their displeasure with the Democratic Party known. I’m not here to repudiate the fact that this was an absolute romp. The democratic brand is dog 💩 right now and they need to do a complete top to bottom house cleaning. It’s just that, the margin of the tipping point state was 2%, so from an electoral college perspective, it was very close and the model properly prepared us for this.
I think the early vote guys had a pretty good indication that Trump was ahead.
> All the signs were there for a massive Trump win.
I'm not sure what you are talking about, but in any case, does that mean you realize you were wrong now, because there was no "massive Trump win"? This was a 2 point squeaker, that could just as easily have gone the other way.
Look at relative shifts by precinct. Trump overperformed virtually everywhere. The polls didn't capture that.
“Trump over performed virtually everywhere. The polls didn’t capture that.”
When you say “overperformed” what I think you mean is “had better numbers than in 2020.” And the polls absolutely captured that. Biden won the PV by 4.5 points in 2020. Most national polls showed essentially a tied race this go around. So that means the polls captured a 4.5 point shift to Trump.
You want to get all up in your feelings because Trump is going to win the popular vote by about 1 point? OK, so the polls captured a 4.5 point swing instead of a 5.5 point swing. But don’t sit here and tell me the polls predicted something else.
How many polls predicted that Trump would win the popular vote?
So far as Trump's margin he's currently up by two points in the popular vote, and in the electoral count he's at 312.
I looked at the 17 most recent national polls included in the RCP average. While I think you and I can both agree in polling there isn't a statistically significant difference (with a standard MOE of 3, those numbers mean very little) between Harris +1, tied, and Trump +1, for the sake of playing your game let's say there is.
-5 polls showed Trump winning the popular vote
-5 polls showed the race tied
-7 polls showed Harris winning the popular vote
So 10/17 polls showed either Trump winning or the race tied, when the final PV margin is going to be Trump +1. That's quite good! Don't sit here and get butt hurt and cry because the data isn't supporting the conclusion you want to have.
The RCP average still had Harris up from what I recall, and here Silver had Harris heavily favored to win the popular vote. Trump beat his polling for the third consecutive election.
Again, currently Trump is ahead in the popular vote by about two points, not one. Additionally look at the number of counties where he did better compared to 2020 versus the counties where Harris improved on Biden.
Squeaker really? Trump won popular vote which not even Nate was predicting. NY was won by democrats by less than Trump.win TX and FL. NJ was <4. Complete sweep of swing states. This was a mandate and such a polarized environment This was a landslide. I think mid terms will be interesting. What if republicans keep.hold of the house in 2026?
That Trump won the popular vote was definitely somewhere in the tail of the distribution (as opposed to the center). But exceeding expectations doesn't make something a landslide lol, it makes it an 'upset'.
When all is said and done, Trump is going to have won the popular vote by about 2 points.
By contrast, Biden won it by 4.5 points, Obama by 4 points in 2012, and 7 points in 2008. Clinton won by 5.5 and 8.5, HW by 7, and Reagan won the popular vote by 10 points in 1980, and 18 points (!!) in '84.
If we compare these results to the past 50 years of elections, it definitely looks like a squeaker, albeit an unexpected one. It only looks like a 'landslide' if we compare to exactly Bush Jr. or Trump's first campaign – neither of which are particularly high bars.
Nate had something like a 23% chance Trump would win the PV. That’s definitely not in the tail.
Fair enough
My thoughts exactly
2% is not a landslide, and all winning politicians claim to have a mandate.
It is impressive to cite a loss in a blue state as evidence of a landslide victory though.
Look at the margins in NJ and VA.
Look at the Tories.
Look at the actual vote share for the Liberals in the UK. Only possible in a parliamentary system.
CJ never knows when to take the L.
I've never seen someone on the W side whine as much as you.
I suppose you do it because you are still confused.
Yup - 1.6% in the Blue Wall flips the result.
Trump was like half a percentage point popular vote worth of votes away from turning NJ, VA, NH, and MN, Maine, and NM red and creating a blood bath map 360 EVs
Kamala got cooked lol
What?
Are you cherry picking a percentage of the popular vote and putting it in specific states, and pretending that means something?
The 1.6% I mentioned was the 255k votes in those states divided by the total vote in those states.
Sorry if I wasn't clear.
It does mean something because there was a greater than expected shift in many of the blue states than the overall trend
No.
That doesn't mean anything because you don't get to magically move those people to other states.
In 2020 Trump lost by about 44k votes in three states. He has definitely outperformed that in this more recent election.
Did someone suggest otherwise?
What are you talking about? He lost all those states by at least 5 points. Harris would have swept the swing states with a 3% shift.
They're inventing some world where the vote counts in all states other than the 6 they mentioned stay the same, and then adding up the margins in those 6 states to get some total that I suppose comes to 0.5% of the *national* electorate.
TL;DR: It's really bad math, and using that same logic we could say that Kamala wins the election with just 0.2% more of the popular vote – so long as that 0.2% was entirely in the blue wall (mostly PA)
Yeah but is Kamala the one who had a 7 point national swing?
I can't believe none of you are seeing the electoral vulnerability in those states in the future lol. I guess the idea is too scary to process right now. Fair enough.
My point here is to show that for every person who copes by saying Kamala was a hair away in the rust belt, don't forget trump ended up close in a whole bunch of not very swingy states, swinging many of them by double digit margins
Compared to Joe's victory in 2020, this was def. a landslide.
"Compared to this one rock falling over, these two rocks falling over was def a landslide!"
Also Trump just won the PV by less than half as much as Biden, maybe a third as much, and got 6 more electoral votes. Just... what are you talking about?
Maybe you should be more than a little alarmed at New Jersey nearly becoming a swing state this cycle instead of clinging onto the idea that this election wasn’t a foreboding sign for democrats. Trump got thrashed in the public sphere for years and still managed to decisively beat Kamala; that should say a lot.
2024 was almost an exact mirror image of 2020.
Neither election was a mandate election. Both elections repudiated the performance of the incumbent president. Both elections had historically narrow popular vote margins. Both elections had surprisingly strong performances in states that were historically out of reach. Both elections narrowly shifted control of the Senate toward the winning presidential party but lost ground in the House. Both elections ended with about 310 electoral votes (the difference mostly because of the changed number of electoral votes in battleground states after the 2020 census), and all seven 2024 battleground states were won by the winning candidate in both elections.
These two elections were almost perfect mirror images of each other, and either you must conclude that both were consequentially large victories by one party or the other, or you must conclude that both are narrow victories magnified by an electoral system that exaggerates the swings in political influence from small changes in election results. Any other interpretation is evidence of partisan wishfulfillment analysis.
For the record, I have consistently maintained that Biden did not win any form of governing mandate in 2020. The message voters sent in 2020 was only that they were not ready for another four years of Trump after suffering through the first 10 months of COVID-19 and the civil unrest following the murder of George Floyd.
President Biden massively misread the 2020 results and foolishly allowed the progressive wing of his party to pursue an extremely ambitious legislative agenda as if he had received a true electoral rout and mandate. Your comments and what we are already seeing in the first week after the election suggest that President-elect Trump is about to make the same mistaken interpretation of his election results and also overplay his political hand.
This will result in an ever-so-predictable backlash in the 2026 midterms, with the most favorable Senate map of the three Senate cycles and a House with only a very narrow Republican margin. This is a clear recipe for swinging both houses of Congress, much like the 2006 midterm cycle.
The big question is whether the Democratic Party can find a fresh new face with a fresh new vision for the party, like Obama provided in 2008. My two best outsider candidates to fulfill this role of a fresh new voice for the party are Mark Cuban and Jamie Dimon, both highly respected, politically savvy, high-profile public figures with no allegiance or baggage from legacy Democratic politics and the freedom to redefine the image of the party. Both are extremely successful businessmen with strong centrist credentials and the independence to craft a new vision for the party much as Bill Clinton did in 1992 and Obama did in 2008.
"Neither election was a mandate election."
A mandate for what though? In Biden's case it was clearly a mandate for a return to the old status quo. The critical point is that there was a mismatch between what the public wanted and what the Biden admin wanted.
That's why this question is meaningless. Are the results a mandate for mass deportations? I would argue that the answer is yes, absolutely. A mandate for more troops in Syria? No, the exact opposite.
To call this a "squeaker" is a bit much. By today's standards, this was a solid win. Not a blowout, but decisive.
With so much of the electorate permanently committed to one of the two parties, you're never going to see things like Reagan '84. Obama '08 is about as big as it gets.
Like the man said, democracy is the proposition that the people know what they want and deserve to get it good and hard.
Hi Evan,
It's often used but I think it's harmful to use "today's standards" to characterize what are historically very close races as a "solid win" or (as if often done but not what you did here, a "landslide"). We are in an era of nearly 50/50 split of the voting public of the two parties. If a <= 3% shift in the votes would have changed the outcome, then it was a squeaker.
I didn't pick the number 3% out of thin air. In 1980 pollster Pat Caddell was asked why he said the election was a flat footed tie and Reagan won by 3%. He explained: The Carter polls showed Carter down by 3%. But it had a margin of error of 3%. So they assumed they were tied when, with hindsight, they were probably down by 6%. And, back then when people really decide at the last moment and we didn't expect pollsters to exactly predict that, bad news the weekend before the election moved the voters another 3%. Resulting in what Reagan would call "a mandate" with only a 9% margin of victory.
Like in 2020, this was a very close election with a clear winner. But the fundamentals of the voting public of the two parties being split around the 50%/50% has not changed. Could it be the start of a trend: Trump lost by 4.5% four years ago and won by 1% this time... maybe. But given the seasaw nature of the last several elections, probably not.
It was massive by most metrics. First PV win by a GOP since 2004. He won all the swing states, increased his margins in FL and TX, and cut the D lead in blue states. Kamala did not do better than Biden in a single county. That's astounding.
2024 was almost an exact mirror image of 2020. Neither election was a mandate election. Both elections repudiated the performance of the incumbent president. Both elections had historically narrow popular vote margins. Both elections had surprisingly strong performances in states that were historically out of reach. Both elections narrowly shifted control of the Senate toward the winning presidential party but lost ground in the House. Both elections ended with about 310 electoral votes (the difference mostly because of the changed number of electoral votes in battleground states after the 2020 census), and all seven 2024 battleground states were won by the winning candidate in both elections.
These two elections were almost perfect mirror images of each other, and either you must conclude that both were consequentially large victories by one party or the other, or you must conclude that both are narrow victories magnified by an electoral system that exaggerates the swings in political influence from small changes in election results. Any other interpretation is evidence of partisan wishfulfillment analysis.
For the record, I have consistently maintained that Biden did not win any form of governing mandate in 2020. The message voters sent in 2020 was only that they were not ready for another four years of Trump after suffering through the first 10 months of COVID-19 and the civil unrest following the murder of George Floyd.
President Biden massively misread the 2020 results and foolishly allowed the progressive wing of his party to pursue an extremely ambitious legislative agenda as if he had received a true electoral rout and mandate. Your comments and what we are already seeing in the first week after the election suggest that President-elect Trump is about to make the exact same mistaken interpretation of his election results and also overplay his political hand.
This will result in an ever-so-predictable backlash in the 2026 midterms, with the most favorable Senate map of the three Senate cycles and a House with only a very narrow Republican margin. This is a clear recipe for swinging both houses of Congress, much like the 2006 midterm cycle.
The big question is whether the Democratic Party can find a fresh new face with a fresh new vision for the party, like Obama provided in 2008. My two best candidates to fulfill this role are Mark Cuban or Jamie Dimon, both highly respected, politically savvy, high-profile public figures with no allegiance or baggage from legacy Democratic politics and the freedom to redefine the image of the party. Both are extremely successful businessmen with strong centrist credentials.
Even if Trump were overwhelmingly popular odds are that the Dems would take the House in 2026. That's just the normal pattern for the first midterm after a presidential election--see Clinton, Obama, Trump and Biden.
One more point.
In the past, a party flipping the WH by winning both the electoral and popular votes also has long coattails and pick up lots of House seats. And then the midterms are a regression to the mean.
But so far this decade that is not happening. The GOP picked up 5 seats in 2020. And then another 5 in 2022. And this time they will be right around net zero. We're around a stability point. So I think there would have to be an unusually strong pro or anti GOP wave for 2026 to see much of a swich either way. Which means the Democrats chances of flipping the House in 2026 is probably less than, or no better than, 50/50. (Maybe 50/50 since the party out of power may be more likely to show up at midterms. But there just not as many swing districts as in prior decades.)
It depends. We'll see what happens then. I think economic performance will be more meaningful than anything else. Or if we are in a war or something.
The only exception was Bush after 9/11. Americans just like divided government.
This is the best take I’ve seen anywhere on the internet about the last two elections. (and I don’t just say that because you’re predicting a Dem win in 2026!). Thanks for sharing.
Didn’t Harris do better than Biden in the counties around Atlanta, in western North Carolina, on the northwest tip of Michigan’s lower peninsula, and several other regions?
This was a 2004 style election. A clear win, but nothing like “massive”. There hasn’t been a “massive” win since 1984. This is smaller than Obama’s 2008 win, which was the only large win in decades.
I think she did better in Dane in Wisconsin.
I was looking for instances of that, and that was one I saw.
R+6 national popular vote shift from the prior election is indeed a blowout given a Republican has won the popular vote exactly once since 1988.
Even Bush 2004 despite a better pop vote margin was much closer; Kerry was 2% away from winning the EC in Ohio.
This is the most decisive Republican victory in decades. MAGA platform with a less divisive leader is capable of a 400 EV rout in 2028.
For reference Nikki Haley was beating Biden by 15 in the national popular vote during the primary season.
So a football team that wins by 1 points has a blowout victory if they lost the previous game by enough.
Pretty funny.
Right now it's two points and Trump above 50%. And in terms of the electoral count Trump got to 312.
And it still isn't a popular vote blowout.
It's big enough to impress guys like Harry Enten. And it's big enough to kick off a civil war in the Democratic party.
The prior election is mostly irrelevant here. I say "mostly" because you do have to account for factors like polarization; victory margins on both sides are much smaller today than they used to be.
But you don't get bonus points for improving on your previous showing. And you certainly shouldn't assume those same trends will continue. Trump was the face of the pandemic in 2020. Biden and Harris were the face of inflation in 2024. The Republican nominee in '28 will be the face of whatever is top of mind that year -- good or bad.
You do if your party has had one popular mandate in 3 decades
If you or the party don't think that's an alarming sign then this will be more like 1980 than 2004
Maaaaybe. Elections are about more than one thing--the Trump team is convinced that their best ad was the one attacking Harris for supporting sex change operations for illegal immigrants in federal custody. I would not be surprised if the Dems are out of the executive until they pull a Sista Souljah and put some distance between themselves and trans activists: "No boys in girls' sports".
I disagree with your second half. No chance the low prop voters turn out for a RINO like Nikki. In fact, I'm doubtful the low prop voters will turn out for anyone besides Trump. That's a worry the GOP will need to address sooner rather than alter.
Woot! Something we agree on mostly.
I wouldn't call Haley a RINO - she is a proud graduate of a segregation academy, and is definitely MAGA adjacent. If Haley and people like her leave the Rs the GOP will be in a world of hurt.
The 2028 Republican primary will be interesting.
Haley, DeSantis, Vance, Kemp?, a couple senators and/or Cabinet folks (Cruz/Rubio/Noem/etc), Lombardo?, Huckabee Sanders?
And of course Tucker Carlson.
If you don’t like Silver Bulletin being poll-based, then just leave.
That is accurate. But Nate need to lower the ratings for left wing polls of morning consult, Washington post, quinipiac the same way he does for right wing polls of trafalgar, Rasmussen. He was actually saved a little by right wing pollsters.
His model adds weight to pollsters who are more accurate already, and he compensates for house effects.
Of course Rasmussen and Trafalgar were further off in 2022, so that balances things a little.
This is the third presidential election in a row where just averaging polls (RCP) performs better than Silver's weighted model.
It's like how stock market ETFs perform better than 75% of actively managed funds.
You don't seem to understand that RCP is an actively managed polling average.
But they were more right in 2020. Quinipiac, Washington post and morning consult were more wrong in 2020 and now in 2024. So.shouldnt they be weighted down going forward as well?
They were, are, and will be in the future assuming Nate uses the same basic model.
https://www.natesilver.net/p/pollster-ratings-silver-bulletin
He gave WaPo an A+ and Quinnipiac an A-. Both polls were trash in 2020 and 2024. That alone makes me question Nate's capabilities. If I didn't tell you the names of the polls but just showed you how much they were off by, and how often they favored Dems, you'd say toss them out.
He didn't do that. As I said, he gave overweights to Quinnipiac and WaPo, two of the worst pollsters by far.
You don't even know how to read his table, do you?
Add up the influence values for Atlas.
What difference does it make? RCP just averaging all the polls did better.
Do you have any justification for why he added 1 point to the popular vote estimate his own model spit out in order to "sync them up" with her performance in the battleground states?
It was an entirely unscientific endeavor, taking a stab that national polls were on the aggregate underestimating Kamala because of disproportionately better battleground polling when the simpler answer could have been that the Pop vote EC bias disappeared (a point he conceded in follow up)
None of this is entirely mathematical, scientific or objective. To some degree his own model does not rely only on polling. He himself states there's some fundamentals integrated into it
The simple average of polls at RCP has performed better than Silver's weighted model for three presidential elections in a row I believe. They're poll based too--that's not the issue.
Was it poll based when he sneakily adjusted pop vote
Margin to +2 based off strong battleground polling relative to weaker national polling?
When his own model showed +0.9?
I like data driven approaches as much as the next guy but don’t pretend his bias didn’t get in the way
Like I said, I am. Guess you didn't read?
I can read your comments (better than you can, apparently). You said lots of people would tell you to just leave. I didn’t want you to be wrong about that as you are about so much else. You’re welcome.
I was right a lot more than I was wrong, which again you'd have known if you read the post.
This was a clear trump win, but you guys need to stop convincing yourself it was some "massive" win. Reps are acting the same right now as Dems did in 2020 and I don't think it will end well for them.
It was massive by most metrics. First PV win by a GOP since 2004. He won all the swing states, increased his margins in FL and TX, and cut the D lead in blue states. Kamala did not do better than Biden in a single county. That's astounding. But you're right, 2026 could easily flip the House back to D.
Massive if you grade on a curve.
Most of your previous comments on this website have been antagonistic and rude towards other subscribers.
Not sure why you're writing this long manifesto to convince others that you are not. Just weird.
Guess JD Vance and me are just weird.
Certainly true for JD Vance :)
That's President Vance to you after January 2029 :)
Polling averages were off by what, 2%? Silver kept pointing out that such a shift either way would take all seven swing states with it.
That's such an obvious statement, and offers no value.
So how was he way off when a tiny shift either way would determine the election? Margin of error is a very real thing in polling and polling averages.
You clearly don't understand the point of Nate's model.
Not a surprise really.
If it was so obvious, why did no one in 2016 offer it except Nate?
Yall don’t understand Monte Carlo simulations, but convince yourself that you do.
Simulations are only as good as their assumptions and inputs. Garbage in, garbage out.
TLDR, but did you explain somewhere in there how polls or the model were supposed to be "wrong" while getting the outcome right in the fat part of the distribution/calling the most likely state map?