If a person thinks "the Democrats are too far to the left" or buys that shallow line from the mainstream corporate media, they were never going to vote for any Democrat anyway, aren't a swing voter, and aren't worth the Democrats bending over backwards to try and court at the expense of demoralizing the liberal/socialist base of the party.
Ah yes - If the People Who Don't Agree with Us Already Don't Buy Us then obviously we must retrench on our loser faction to lose in more elegant and morally pure ways to prove our moral superiority.
You know I totally get where you went with attack and in many situations I feel the same way with many Democrats (including my wife).
I think the point he/she was making was more about the crazy and dangerous behavior from POTUS making picking sides easy, not the how clearly right the Dems are.
I agree though that trying to reach everyone and not giving up on people is important and I see Liberals dismissing people almost as much as MAGA these days... Unless your arguing for something other than democracy you need to be focused on wining people over always! Even when your frustrated.
My point is that Democrats need gettable votes and not wasting resources going after the votes of people who have already dismissed Democratic Party platform and the liberal/socialist ideas it contains for fixing this nation's problems.
Democrats obviously right is a presupposition that embeds with a great deal of self-blinding self-regard.
As it happens I will vote D as I intensely loathe Trump, but I don't start out with Party Always right (insert Party name).
My remark was as this kind of idiotic reponse is fundamentally much of the intellectual snobbery of the Democrat's 'new' so-called affirmed 'base' in the college educated, and the college activist rooted culture of dismissal
Real electoral data demon
Of course a real "base" is one that is reliable, not fickle, base meaning foundation, sticking with the basic foundations no matter the vagaries or fan-service. The black Americans, they are a base in a real sense, sticking with Democrats even when Democrats moved to the Left of what polling says are median black democrats more socially conservative mores, as an example.
What that person reprsents is not a "base" - that is a fickle Faction that insists on self-regarding Fan Service - something more akin to the rabid online Fandoms that if their specific "take" and idiosyncratic desires are not driving the (Series/Show) flip to hate and rejection. That's not a base.
Of course there is further innumeracy as (a) the electoral data show rather clearly that above all in non-solid-Democratic states the Swing Voter (the proverbial Obama-to-Trump-to-Biden-to-Trump) voter is a real floating fraction - not certainly a huge percentage but in the"swing" states with tight results it is the difference beween Winning and Losing - them plus the real Bases (as in e.g. working class black democrats).
It is further mathmatically the case outside of certain narrow electoral geographiies that asserteed "base" (Fan-Service seeking faction, fickle-non-Base) is grossly insufficient to win beyond narrow districts leading to state level losses and not defense against Trump but rather putting Democrats in a structurally weaker (more Fan Serviciing for that self-declared 'base' that is nothing of the sort, unlike the black democrats) overall position to hold back MAGA / Trump.
It is a combination of intense self-regard on their part and precioius preening and affirming that is about self-Fan Service and not successful political achievement in the politicla geography that actually exists.
Working to win, not working to self-servicing Fan Service, and addressing the electorates in play.
Trump "converted" temporarily whole percentages of voter demgraphics to voting R that whenI was 20 were unthinkable to be voting R.
One looks at these movements and it renders the entire discourse of fallacious fixed "bases" that are in fact fickle factions completely nonsensical. Demographic profiles do shift, not always at once in one election or even two, but one can see it, and it's not happening because of God's Hand, it's happening from longer-term plays.
Bases are solid and parties need them to show up for every election to have a floor when building toward victory. Swing voters are fickle and a pain, but enough can be gettable to complete the ceiling that gets over 50% and win the election. The problem is that the Democratic Party keeps mistaking conservative "independents" that are really Republican voters in all but membership only for swing voters. Chasing that kind of voter by casting aside core items the base wants is a recipe for disaster.
No the cope was what Biden/Harris did when they threw the base under the bus trying to build a so called bipartisan coalition with alleged disaffected conservatives by sitting next to Liz Cheney and not cutting Netanyahu loose. Nothing like demoralizing the base with uninspiring centrist politics/policy compromises and then having the voters that was supposed to win over not showing up.
When someone says, "the Democrats are too far to the left" its a tell that they were never and are never going to vote for a Democrat. Once the Democratic Party moves close to the center, it becomes the Republican Lite Party and that kind of person is just going to vote for the real GOP not the the Lite Version.
Exist, yes. But just existing doesn't mean the Democratic Party has to put forth effort courting you if you're already of the opinion that you can never vote for a party that's too far to the left. Democrats need gettable votes from people who don't automatically flinch (or flinch on command when their preferred media of choice tells them to) when the Democratic Party brings up liberal or socialist ideas to fix problems the nation is having.
You said if a person thinks "the Democrats are too far to the left", they were never going to vote for any Democrat anyway.
I'm a Democrat - generally vote Democrat - but think the party is trending too far to the left at least on the micro level (alongside insultingly rationalizing plainly unfit candidates), and therefore I'll sometimes vote Republican.
That's fine. People treat Marshall Field's quote "the customer is always right" as an absolute. I'm a firm believer in the corollaries that "if the customer isn't right, they aren't your customer" and "if the keeping the customer drives away the rest of your customer base or detracts from your core business, then you have an obligation to fire them".
I don't like to personalize by using "our" or "your". Last time I checked, there's no brass name plate on the party. It's "the party". It belongs to those who show up to vote for its candidates and policies in the primaries and the general election.
As for the out of business part, let's see what results November 2026 and November 2028 bring.
Look, maybe it’s not about convincing folks to vote for a party that’s to far too the left, but proving that it’s worth voting for candidates who aren’t, even though they are members of a party that is perceived to be that way.
It’s about candidates beating narratives and if individual candidates beat the dominant narratives they open space for folks to be convinced to vote for them. If elections weren’t about persuasion, then how are electorates shifting over time?
That’s what your point amounts to, no need to persuade certain people, because they weren’t persuadable in the first place. The problem is the group you are claiming isn’t persuadable or worth the effort, will help determine future outcomes.
That group already determines future outcomes by reliably voting Republican no matter how often they say, "I would vote for a Democrat if they weren't so far left." They are no more or less important than base who is always loyal and swing voters who are persuadable, but a pain to deal with. It's just that its a waste for Democrats to spend money and time on unpersuadable voters when it could be both could be better spent revving up the base and knocking on doors to find getabble swing voters.
just this "But the DSA-type candidates tend to be more unapologetically proud of being on the left, and don’t necessarily seem to be team players in quite the same way" - people didn't vote for them this time because they're on the left - it's that they come across stronger than the average aging Democratic candidate who's just been around too long and who supports large corporate donors.
I think AES will be fine in the general - he is a talented candidate, and I think his brand of abrasiveness will probably be a net positive overall (fighter, etc), and I assume he will be doubling down on how much he loves capitalism.
I am not commenting on how authentic he actually is, but regardless, “authenticity” or appearance of same makes a difference.
A lot of great stuff here, the herding-on-primary-polls point especially. But a couple of objections.
The betting tip as a reward for listening is gambling advice (and bad advice) not "gambling-adjacent," whatever that means. Naming the contract, the venue, the price, and the direction isn't adjacent to gambling advice. Pairs with "probably has to be too low." Is 3% too low or not? There's no probability about that, and it doesn't have to be anything.
To distinguish a true 3% from a true 5%, you need enough comparable cases that the difference in event counts clears the noise, on the order of 400 primaries with the leader at 3¢ and a double-digit polling average. And even if you had them, there would be so many confounding differences to make the analysis worthless.
Anyone claiming to distinguish a 3% from a 5% probability should have a good model, like one of a roulette wheel, or a long track record of successfully distinguishing on comparable bets. Otherwise it's not even wrong.
And treat any claim that a longshot is underpriced with strong skepticism. A 300,000-contract Kalshi study found a clear favorite-longshot bias: cheap contracts win less often than price implies. Kalshi longshots under 30¢ have run overpriced against a 4% realized rate, and the pattern replicates on Polymarket at long horizons. The favorite-longshot bias is among the best supported generalizations in gambling.
By Sunday evening Stevens was 1.6/1.7 on Kalshi, not 3¢. The market moved to half your quoted price while the chat was being edited. Later prices are better calibrated than earlier ones.
There is no actuarial inflection point at 80. I've raised this before. Regress log q(x) on age in the SSA male table, 60 to 95: R² = 0.998, a straight line, doubling time drifting smoothly from 9.1 years at 60–65 to 6.8 at 80–85. What changes at 80 is the level, 1.6% at 65, 6.1% at 80, not the shape. On the linear scale q is convex throughout, so there's no inflection anywhere. On the log scale the one genuine inflection is late-life deceleration near 87–90, running the other way.
The real phenomenon nearby is terminal decline. Cognitive function drops sharply in the last 3–7 years of life, whenever those years fall. So post-80 steepening in age-indexed data is substantially composition: far more 80-year-olds are near death than 70-year-olds. Age predicts death, death predicts cognitive decline, and most of the apparent post-80 acceleration is that composition rather than age itself.
Also, presidents are a severely selected sample — Olshansky found 23 of 34 who died naturally outlived contemporaneous expectation. The hardest-job-in-the-world credit may run backwards.
Terrific analysis, guys, and I'm holding my breath for the Model. If I may be permitted just one standing turd-in-punchbowl gloss... Everything here is predicated, without qualification, on something akin to a genuinely free and fair election, sans interference by a seemingly motivated and ethically unconstrained Trump (& Co.).
While I understand that there's really no way to quantitatively factor in his various potential schemes (ranging from overt tactics like ICE at the polls and last-minute voter purges to covert stunts like targeted electronic vote-count manipulation), these *are* in fact real possibilities that make me very queasy in relying on any model that approaches this election "straight up."
It might be worth a qualitative qualification or disclaimer, to the effect that model and predictions are all predicated on a reasonably democratic electoral process secure from outcome-impacting rigging. And perhaps as well some sort of asterisk on conclusions about the causes of historical misses (e.g., 2016) that were based on the assumption that the vote counts were pristine and an accurate reflection of the collective will of those who voted or attempted to vote. And finally, a recognition that those conclusions necessarily found their way into subsequent and current predictive models.
There is not much you can do in modelling a rigged game if rigging is a priori off the table as a consideration. But it must be acknowledged that IF the game turns out to have been rigged, all the hard work that has gone into modelling its contours -- and explaining its surprises -- winds up corrupted.
I have always believed that the only genuine value of pre-election polling, and models based thereon, is to make it obvious if the election was corrupted.
And yet, and yet... I, and others, labored for 20+ years analyzing exactly those disparities -- which were virtually unidirectional AND strongly correlated with contest competitiveness and significance -- I even coined the term for it "the red shift" -- and the entire adult table yawned and shrugged. "Those polls must have been 'off' again!"
Is there a way to use what you generate with your model to estimate the effects on state legislative races? Not necessarily a specific prediction, but some range of influence, given the model's view of how things are changing (say, from 2024 or from 2022) to get an idea of how much we'd expect the same legislative district and candidate(s) to change?
Good point that the party out of power tends to gain in the generic ballot closer to the election. 6 of the past 8 elections have been under a Democrat president so people often mislabel the phenomenon as “Republicans gain in the generic ballot closer to the election”.
Lots and lots of good intel, and I like it. I read it word for word, twice. Pretty funny about the 9s coding throwing off the whole model --- good story. [:-)
"There are some reasons to think that everything apart from partisanship is becoming less important," Nate said. Yes, the drawing further and further apart of the electorate into right and left has to be a major issue; I've never seen anything like it in my life, though it apparently did happen before the Civil War in the 1850s. One of the possible current-changes Nate thought possible was that Republicans may decide Dems are crazy and bringing back Woke --- I immediately thought of that nice young woman basketball player, who seems very sensible and calm, who has stated that she doesn't think women should have to play against men. Well, duh!! Ya think??! Now trannies and tranny-supporters are following her from city to city demonstrating loudly against her. Of course this will be weaponized against Dems (and should be).
I was surprised at the long discussion near the end claiming that Dems are much better at fundraising. Since the Wall Street Journal just today had a story about how truly awful the situation is with the respective national committees --- the Republicans have $127 million and no debt; but the Democrats have $18 million and $16 million debt, and had to give their whole D.C. building as collateral to get that much, since they had nothing else to offer. AND the Republicans have most of the billionaires, as the story continued. AND the usual big Democratic contributors are sitting on their hands: even George Soros. It can't be he doesn't like the far-left politics swing; maybe he doesn't like the losing -- apparently the waste of their money in lost elections is bothering a lot of big past Dem contributors. So here are diametrically opposed opinions from Nate and Eli on one side and the WSJ on the other side about the effectiveness of Dem fundraising. I understand that Nate is thinking of small-amount donors, and I like the point that you weigh local state donors five times more than out-of-state donors (because money is money, but votes are what it really takes to win).
Talarico has ten times the money his opponent has, but that race is still a toss-up. I like the advice not necessarily to contribute to high-profile races: diminishing returns.
> I think people forget the longer tail of races like Iowa, for example. If you look at Iowa, it is polling as a toss-up, basically. With the fundamentals, my guess is the model will have it as “lean R” — I don’t remember offhand — but Iowa is a big deal.
"Eli, describe to me the world where that 20 percent [chance of a Republican House] happens."
- Trump orders ICE and National Guard to 'observe' key races. They intimidate and beat up people and in general disrupt the voting. Compliant election boards approve the fake results. The SCOTUS declines to intervene IF lawsuits are brought (many of those will be in republican states).
I’ve considered that too. I think ICE at polling places stimulates Dem turnout to some extent, in certain places surely (like NYC), but yeah the tactics you describe could go on.
Another scary possibility: let’s say Dan Osborn wins in NE. But let’s also say NE has Republican super-majority and Republican governor. (I don’t know offhand if they do but assuming yes.) Do they somehow undermine the legitimate outcome at the state election level? Could that scenario play out in NC, TX, OH (and MT, ID and SD if the independents had a great election night?)
1) Stevens has Schumer-gunk on her, ie, Dems still pissed about 2024 are eager to show that they will NOT cater to Chuck Schumer’s preferences any fucking more. (Whitmer, too, was a big Biden cheerleader - which probably made her Stevens endorsement a liability rather an asset).
2) Riddle me this: Even if Mike Rogers is a normie Republican, how does he simultaneously assure the MI MAGA crowd that he’s still 100% pro Trump while ALSO pulling in unhappy independents who want someone to stand up to Trump re the war and tariffs? Mutually exclusive messaging.
Eh, Whitmer is still pretty popular in Michigan. TBH I think if she had run for this seat she probably would have the primary more or less wrapped up already. If anything, I think it’s more the case that a last minute endorsement for a candidate that was/is more than likely too far buried, is more a liability for Whitmer’s future political aspirations, than it is one for Stevens’ already dim chances.
Do we know why Whitmer did NOT run? I’m not in MI so not sure of the dynamics there, but I agree she would’ve been the strongest candidate compared to these two (although El-sayed coulda still thrown down I think - ESPECIALLY if he had called out her inexplicable Biden lap-dogging in late 2023-24.)
I don’t know offhand, haven’t followed it that closely. This is just my guess off-the-cuff but I’m wondering if she has presidential aspirations in 2028 and didn’t want to be seen as abandoning a Senate seat just two years into her first term? And were she to actually win the presidency, then a special election in a swing state right after a Dem won the presidency would probably be a tough out, and losing that seat might hamstring her hypothetical first term. Obviously that’s thinking way far ahead, but maybe not for someone who fancies herself a presidential candidate?
On 1., Schumer gunk is on some great candidates too… Cooper, Peltola, Brown, Turek, etc.
Yeah he’s had some duds (Mills, and I think Stevens over McMorrow or just staying out of Michigan all together), but his record in finding quality candidates for states has actually been good and voters haven’t punished them for it.
Also, I agree about Whitmer. As an outside observer I don’t think she’s as skilled as folks give her credit for or as impactful of a politician.
Yeah in redder states Schumer does ok - like Peltola - but ARE there any other Dems in AK? If there’s a candidate with Schumer-gunk AND one without, like MI - or ME - purple-y places with more than one option, I think it’s a liability.
You always have thoughtful comments, so what do you think about Brown in OH? I feel like they woulda done better in OH to recruit a Platner-ish or Dan Osborn type instead of a 70-something establishment guy who just lost two years ago.
Honestly, yeah I agree with you about Brown and that’s true the moderate/establishment types are doing better in moderate places. I’ll say 2 things in addition. I think NC is more purple than folks realize and Ruben Gallego in AZ was someone Schumer endorsed in the primary back when he ran.
And the second thing, the bigger problem is ok maybe you can encourage strong candidates to run, but endorsing and putting the thumb on the scale from Senate leadership is bad. Encourage folks to run but don’t pick them outright. That’s the voter’s job. So in that way I don’t like any of Schumer’s endorsements.
Whitmer has a BAD Covid record. She's the one who said people couldn't shop for vegetable garden seeds in Michigan stores. She just went whole-hog Nazi control-mode during Covid, and I'll never forgive that. There are a lot of Covid chickens finally coming to roost now, like Fauci and all his sins.
In video - you discuss a possible 2pt swing when moving from RV to LV models. It seems like this is unlikely to be across the board… do we know if the 2pt average is actually driven by states that are in the bag anyway, and republican voter attendance will crater? Seems important in thinking how this difference will matter in real results.
Do we think maybe Nate’s favorite part of being on Substack is he can say/write the word “fuck”?
New York man, it's in our blood. And gets into even the incomers.
If a person thinks "the Democrats are too far to the left" or buys that shallow line from the mainstream corporate media, they were never going to vote for any Democrat anyway, aren't a swing voter, and aren't worth the Democrats bending over backwards to try and court at the expense of demoralizing the liberal/socialist base of the party.
Ah yes - If the People Who Don't Agree with Us Already Don't Buy Us then obviously we must retrench on our loser faction to lose in more elegant and morally pure ways to prove our moral superiority.
You know I totally get where you went with attack and in many situations I feel the same way with many Democrats (including my wife).
I think the point he/she was making was more about the crazy and dangerous behavior from POTUS making picking sides easy, not the how clearly right the Dems are.
I agree though that trying to reach everyone and not giving up on people is important and I see Liberals dismissing people almost as much as MAGA these days... Unless your arguing for something other than democracy you need to be focused on wining people over always! Even when your frustrated.
My point is that Democrats need gettable votes and not wasting resources going after the votes of people who have already dismissed Democratic Party platform and the liberal/socialist ideas it contains for fixing this nation's problems.
Democrats obviously right is a presupposition that embeds with a great deal of self-blinding self-regard.
As it happens I will vote D as I intensely loathe Trump, but I don't start out with Party Always right (insert Party name).
My remark was as this kind of idiotic reponse is fundamentally much of the intellectual snobbery of the Democrat's 'new' so-called affirmed 'base' in the college educated, and the college activist rooted culture of dismissal
Real electoral data demon
Of course a real "base" is one that is reliable, not fickle, base meaning foundation, sticking with the basic foundations no matter the vagaries or fan-service. The black Americans, they are a base in a real sense, sticking with Democrats even when Democrats moved to the Left of what polling says are median black democrats more socially conservative mores, as an example.
What that person reprsents is not a "base" - that is a fickle Faction that insists on self-regarding Fan Service - something more akin to the rabid online Fandoms that if their specific "take" and idiosyncratic desires are not driving the (Series/Show) flip to hate and rejection. That's not a base.
Of course there is further innumeracy as (a) the electoral data show rather clearly that above all in non-solid-Democratic states the Swing Voter (the proverbial Obama-to-Trump-to-Biden-to-Trump) voter is a real floating fraction - not certainly a huge percentage but in the"swing" states with tight results it is the difference beween Winning and Losing - them plus the real Bases (as in e.g. working class black democrats).
It is further mathmatically the case outside of certain narrow electoral geographiies that asserteed "base" (Fan-Service seeking faction, fickle-non-Base) is grossly insufficient to win beyond narrow districts leading to state level losses and not defense against Trump but rather putting Democrats in a structurally weaker (more Fan Serviciing for that self-declared 'base' that is nothing of the sort, unlike the black democrats) overall position to hold back MAGA / Trump.
It is a combination of intense self-regard on their part and precioius preening and affirming that is about self-Fan Service and not successful political achievement in the politicla geography that actually exists.
Working to win, not working to self-servicing Fan Service, and addressing the electorates in play.
Trump "converted" temporarily whole percentages of voter demgraphics to voting R that whenI was 20 were unthinkable to be voting R.
One looks at these movements and it renders the entire discourse of fallacious fixed "bases" that are in fact fickle factions completely nonsensical. Demographic profiles do shift, not always at once in one election or even two, but one can see it, and it's not happening because of God's Hand, it's happening from longer-term plays.
Bases are solid and parties need them to show up for every election to have a floor when building toward victory. Swing voters are fickle and a pain, but enough can be gettable to complete the ceiling that gets over 50% and win the election. The problem is that the Democratic Party keeps mistaking conservative "independents" that are really Republican voters in all but membership only for swing voters. Chasing that kind of voter by casting aside core items the base wants is a recipe for disaster.
You build to "victory" by having the percent to win
A Base is somehting that shows up, not fickle
Your DSA base is not a base at all but a self-regarding faction.
Your other innumerate Just So assertions are not data but cope
No the cope was what Biden/Harris did when they threw the base under the bus trying to build a so called bipartisan coalition with alleged disaffected conservatives by sitting next to Liz Cheney and not cutting Netanyahu loose. Nothing like demoralizing the base with uninspiring centrist politics/policy compromises and then having the voters that was supposed to win over not showing up.
When someone says, "the Democrats are too far to the left" its a tell that they were never and are never going to vote for a Democrat. Once the Democratic Party moves close to the center, it becomes the Republican Lite Party and that kind of person is just going to vote for the real GOP not the the Lite Version.
So apparently I don't exist in your opinion. Fascinating.
Exist, yes. But just existing doesn't mean the Democratic Party has to put forth effort courting you if you're already of the opinion that you can never vote for a party that's too far to the left. Democrats need gettable votes from people who don't automatically flinch (or flinch on command when their preferred media of choice tells them to) when the Democratic Party brings up liberal or socialist ideas to fix problems the nation is having.
You said if a person thinks "the Democrats are too far to the left", they were never going to vote for any Democrat anyway.
I'm a Democrat - generally vote Democrat - but think the party is trending too far to the left at least on the micro level (alongside insultingly rationalizing plainly unfit candidates), and therefore I'll sometimes vote Republican.
That's fine. People treat Marshall Field's quote "the customer is always right" as an absolute. I'm a firm believer in the corollaries that "if the customer isn't right, they aren't your customer" and "if the keeping the customer drives away the rest of your customer base or detracts from your core business, then you have an obligation to fire them".
Cocky talk, considering our (your?) party is very much out of business.
I don't like to personalize by using "our" or "your". Last time I checked, there's no brass name plate on the party. It's "the party". It belongs to those who show up to vote for its candidates and policies in the primaries and the general election.
As for the out of business part, let's see what results November 2026 and November 2028 bring.
Look, maybe it’s not about convincing folks to vote for a party that’s to far too the left, but proving that it’s worth voting for candidates who aren’t, even though they are members of a party that is perceived to be that way.
It’s about candidates beating narratives and if individual candidates beat the dominant narratives they open space for folks to be convinced to vote for them. If elections weren’t about persuasion, then how are electorates shifting over time?
That’s what your point amounts to, no need to persuade certain people, because they weren’t persuadable in the first place. The problem is the group you are claiming isn’t persuadable or worth the effort, will help determine future outcomes.
That group already determines future outcomes by reliably voting Republican no matter how often they say, "I would vote for a Democrat if they weren't so far left." They are no more or less important than base who is always loyal and swing voters who are persuadable, but a pain to deal with. It's just that its a waste for Democrats to spend money and time on unpersuadable voters when it could be both could be better spent revving up the base and knocking on doors to find getabble swing voters.
just this "But the DSA-type candidates tend to be more unapologetically proud of being on the left, and don’t necessarily seem to be team players in quite the same way" - people didn't vote for them this time because they're on the left - it's that they come across stronger than the average aging Democratic candidate who's just been around too long and who supports large corporate donors.
I think AES will be fine in the general - he is a talented candidate, and I think his brand of abrasiveness will probably be a net positive overall (fighter, etc), and I assume he will be doubling down on how much he loves capitalism.
I am not commenting on how authentic he actually is, but regardless, “authenticity” or appearance of same makes a difference.
A lot of great stuff here, the herding-on-primary-polls point especially. But a couple of objections.
The betting tip as a reward for listening is gambling advice (and bad advice) not "gambling-adjacent," whatever that means. Naming the contract, the venue, the price, and the direction isn't adjacent to gambling advice. Pairs with "probably has to be too low." Is 3% too low or not? There's no probability about that, and it doesn't have to be anything.
To distinguish a true 3% from a true 5%, you need enough comparable cases that the difference in event counts clears the noise, on the order of 400 primaries with the leader at 3¢ and a double-digit polling average. And even if you had them, there would be so many confounding differences to make the analysis worthless.
Anyone claiming to distinguish a 3% from a 5% probability should have a good model, like one of a roulette wheel, or a long track record of successfully distinguishing on comparable bets. Otherwise it's not even wrong.
And treat any claim that a longshot is underpriced with strong skepticism. A 300,000-contract Kalshi study found a clear favorite-longshot bias: cheap contracts win less often than price implies. Kalshi longshots under 30¢ have run overpriced against a 4% realized rate, and the pattern replicates on Polymarket at long horizons. The favorite-longshot bias is among the best supported generalizations in gambling.
By Sunday evening Stevens was 1.6/1.7 on Kalshi, not 3¢. The market moved to half your quoted price while the chat was being edited. Later prices are better calibrated than earlier ones.
There is no actuarial inflection point at 80. I've raised this before. Regress log q(x) on age in the SSA male table, 60 to 95: R² = 0.998, a straight line, doubling time drifting smoothly from 9.1 years at 60–65 to 6.8 at 80–85. What changes at 80 is the level, 1.6% at 65, 6.1% at 80, not the shape. On the linear scale q is convex throughout, so there's no inflection anywhere. On the log scale the one genuine inflection is late-life deceleration near 87–90, running the other way.
The real phenomenon nearby is terminal decline. Cognitive function drops sharply in the last 3–7 years of life, whenever those years fall. So post-80 steepening in age-indexed data is substantially composition: far more 80-year-olds are near death than 70-year-olds. Age predicts death, death predicts cognitive decline, and most of the apparent post-80 acceleration is that composition rather than age itself.
Also, presidents are a severely selected sample — Olshansky found 23 of 34 who died naturally outlived contemporaneous expectation. The hardest-job-in-the-world credit may run backwards.
Yow. The last two paragraphs are sticking in my memory. It is worth keeping in mind, re age in general and (uh-oh) in specific, politically.
And when they fade from your memory, it could be time to get your affairs in order.
I heartily agree that the country would be better if Ohio were part of PA, KY, IN and Michigan
Terrific analysis, guys, and I'm holding my breath for the Model. If I may be permitted just one standing turd-in-punchbowl gloss... Everything here is predicated, without qualification, on something akin to a genuinely free and fair election, sans interference by a seemingly motivated and ethically unconstrained Trump (& Co.).
While I understand that there's really no way to quantitatively factor in his various potential schemes (ranging from overt tactics like ICE at the polls and last-minute voter purges to covert stunts like targeted electronic vote-count manipulation), these *are* in fact real possibilities that make me very queasy in relying on any model that approaches this election "straight up."
It might be worth a qualitative qualification or disclaimer, to the effect that model and predictions are all predicated on a reasonably democratic electoral process secure from outcome-impacting rigging. And perhaps as well some sort of asterisk on conclusions about the causes of historical misses (e.g., 2016) that were based on the assumption that the vote counts were pristine and an accurate reflection of the collective will of those who voted or attempted to vote. And finally, a recognition that those conclusions necessarily found their way into subsequent and current predictive models.
There is not much you can do in modelling a rigged game if rigging is a priori off the table as a consideration. But it must be acknowledged that IF the game turns out to have been rigged, all the hard work that has gone into modelling its contours -- and explaining its surprises -- winds up corrupted.
I have always believed that the only genuine value of pre-election polling, and models based thereon, is to make it obvious if the election was corrupted.
And yet, and yet... I, and others, labored for 20+ years analyzing exactly those disparities -- which were virtually unidirectional AND strongly correlated with contest competitiveness and significance -- I even coined the term for it "the red shift" -- and the entire adult table yawned and shrugged. "Those polls must have been 'off' again!"
If anyone's interested, here's some of that work: https://codered2014.com/wp-content/uploads/2023/11/TheRealSteal-IntroAnalysisCombinedUpdated-js8_WWW-2.pdf and https://www.amazon.com/CODE-RED-Computerized-Elections-Democracy/dp/B087H83JCR/ref. Others (e.g., Election Truth Alliance) have picked up the statistical forensics torch for 2024. There's a whole lot of smoke blowing in one direction for us to blithely ignore any possibility of fire.
Is there a way to use what you generate with your model to estimate the effects on state legislative races? Not necessarily a specific prediction, but some range of influence, given the model's view of how things are changing (say, from 2024 or from 2022) to get an idea of how much we'd expect the same legislative district and candidate(s) to change?
Good point that the party out of power tends to gain in the generic ballot closer to the election. 6 of the past 8 elections have been under a Democrat president so people often mislabel the phenomenon as “Republicans gain in the generic ballot closer to the election”.
Lots and lots of good intel, and I like it. I read it word for word, twice. Pretty funny about the 9s coding throwing off the whole model --- good story. [:-)
"There are some reasons to think that everything apart from partisanship is becoming less important," Nate said. Yes, the drawing further and further apart of the electorate into right and left has to be a major issue; I've never seen anything like it in my life, though it apparently did happen before the Civil War in the 1850s. One of the possible current-changes Nate thought possible was that Republicans may decide Dems are crazy and bringing back Woke --- I immediately thought of that nice young woman basketball player, who seems very sensible and calm, who has stated that she doesn't think women should have to play against men. Well, duh!! Ya think??! Now trannies and tranny-supporters are following her from city to city demonstrating loudly against her. Of course this will be weaponized against Dems (and should be).
I was surprised at the long discussion near the end claiming that Dems are much better at fundraising. Since the Wall Street Journal just today had a story about how truly awful the situation is with the respective national committees --- the Republicans have $127 million and no debt; but the Democrats have $18 million and $16 million debt, and had to give their whole D.C. building as collateral to get that much, since they had nothing else to offer. AND the Republicans have most of the billionaires, as the story continued. AND the usual big Democratic contributors are sitting on their hands: even George Soros. It can't be he doesn't like the far-left politics swing; maybe he doesn't like the losing -- apparently the waste of their money in lost elections is bothering a lot of big past Dem contributors. So here are diametrically opposed opinions from Nate and Eli on one side and the WSJ on the other side about the effectiveness of Dem fundraising. I understand that Nate is thinking of small-amount donors, and I like the point that you weigh local state donors five times more than out-of-state donors (because money is money, but votes are what it really takes to win).
Talarico has ten times the money his opponent has, but that race is still a toss-up. I like the advice not necessarily to contribute to high-profile races: diminishing returns.
I listens to this whole conversation and no mention of my personal dark horse Iowa 😔
Not none!
> I think people forget the longer tail of races like Iowa, for example. If you look at Iowa, it is polling as a toss-up, basically. With the fundamentals, my guess is the model will have it as “lean R” — I don’t remember offhand — but Iowa is a big deal.
True I wanted to hear an actual take on Turek as a candidate in Iowa though but yes they did mention it haha
There is a mention, near the top. As you say, a mysterious situation.
"Eli, describe to me the world where that 20 percent [chance of a Republican House] happens."
- Trump orders ICE and National Guard to 'observe' key races. They intimidate and beat up people and in general disrupt the voting. Compliant election boards approve the fake results. The SCOTUS declines to intervene IF lawsuits are brought (many of those will be in republican states).
Guys, could you put odds on that?
I’ve considered that too. I think ICE at polling places stimulates Dem turnout to some extent, in certain places surely (like NYC), but yeah the tactics you describe could go on.
Another scary possibility: let’s say Dan Osborn wins in NE. But let’s also say NE has Republican super-majority and Republican governor. (I don’t know offhand if they do but assuming yes.) Do they somehow undermine the legitimate outcome at the state election level? Could that scenario play out in NC, TX, OH (and MT, ID and SD if the independents had a great election night?)
1) Stevens has Schumer-gunk on her, ie, Dems still pissed about 2024 are eager to show that they will NOT cater to Chuck Schumer’s preferences any fucking more. (Whitmer, too, was a big Biden cheerleader - which probably made her Stevens endorsement a liability rather an asset).
2) Riddle me this: Even if Mike Rogers is a normie Republican, how does he simultaneously assure the MI MAGA crowd that he’s still 100% pro Trump while ALSO pulling in unhappy independents who want someone to stand up to Trump re the war and tariffs? Mutually exclusive messaging.
Eh, Whitmer is still pretty popular in Michigan. TBH I think if she had run for this seat she probably would have the primary more or less wrapped up already. If anything, I think it’s more the case that a last minute endorsement for a candidate that was/is more than likely too far buried, is more a liability for Whitmer’s future political aspirations, than it is one for Stevens’ already dim chances.
Do we know why Whitmer did NOT run? I’m not in MI so not sure of the dynamics there, but I agree she would’ve been the strongest candidate compared to these two (although El-sayed coulda still thrown down I think - ESPECIALLY if he had called out her inexplicable Biden lap-dogging in late 2023-24.)
I don’t know offhand, haven’t followed it that closely. This is just my guess off-the-cuff but I’m wondering if she has presidential aspirations in 2028 and didn’t want to be seen as abandoning a Senate seat just two years into her first term? And were she to actually win the presidency, then a special election in a swing state right after a Dem won the presidency would probably be a tough out, and losing that seat might hamstring her hypothetical first term. Obviously that’s thinking way far ahead, but maybe not for someone who fancies herself a presidential candidate?
On 1., Schumer gunk is on some great candidates too… Cooper, Peltola, Brown, Turek, etc.
Yeah he’s had some duds (Mills, and I think Stevens over McMorrow or just staying out of Michigan all together), but his record in finding quality candidates for states has actually been good and voters haven’t punished them for it.
Also, I agree about Whitmer. As an outside observer I don’t think she’s as skilled as folks give her credit for or as impactful of a politician.
Yeah in redder states Schumer does ok - like Peltola - but ARE there any other Dems in AK? If there’s a candidate with Schumer-gunk AND one without, like MI - or ME - purple-y places with more than one option, I think it’s a liability.
You always have thoughtful comments, so what do you think about Brown in OH? I feel like they woulda done better in OH to recruit a Platner-ish or Dan Osborn type instead of a 70-something establishment guy who just lost two years ago.
Honestly, yeah I agree with you about Brown and that’s true the moderate/establishment types are doing better in moderate places. I’ll say 2 things in addition. I think NC is more purple than folks realize and Ruben Gallego in AZ was someone Schumer endorsed in the primary back when he ran.
And the second thing, the bigger problem is ok maybe you can encourage strong candidates to run, but endorsing and putting the thumb on the scale from Senate leadership is bad. Encourage folks to run but don’t pick them outright. That’s the voter’s job. So in that way I don’t like any of Schumer’s endorsements.
Whitmer has a BAD Covid record. She's the one who said people couldn't shop for vegetable garden seeds in Michigan stores. She just went whole-hog Nazi control-mode during Covid, and I'll never forgive that. There are a lot of Covid chickens finally coming to roost now, like Fauci and all his sins.
What's with the low video quality?
I am so excited to engage my worst impulses and check the new model incessantly 🥹
In video - you discuss a possible 2pt swing when moving from RV to LV models. It seems like this is unlikely to be across the board… do we know if the 2pt average is actually driven by states that are in the bag anyway, and republican voter attendance will crater? Seems important in thinking how this difference will matter in real results.