Latest / Decision Desk HQ Podcast / Decision Desk HQ 2026 Forecast Update Convo for late August, with Chris Cillizza

Transcript

Decision Desk HQ 2026 Forecast Update Convo for late August, with Chris Cillizza · Decision Desk HQ Podcast

0:00
SPEAKER_01: Hey everybody, it is Chris Eliza, so what?
0:03
SPEAKER_01: Substack, man, I hope you subscribe.
0:05
SPEAKER_01: If you don't, now's the time.
0:06
SPEAKER_01: I am joined in what is going to be a regular feature from now until the election,
0:11
SPEAKER_01: so this is exciting for me,
0:13
SPEAKER_01: by Geoffrey Skelley, who you know we do live streams together.
0:17
SPEAKER_01: We were together on Tuesday night.
0:19
SPEAKER_01: He is the Chief Elections Analyst at DDHQ.
0:24
SPEAKER_01: And then someone you don't know,
0:25
SPEAKER_01: but you're going to get to know and you need to know,
0:26
SPEAKER_01: Madison Morrell,
0:27
SPEAKER_01: who is the Data Science Fellow at Decision Desk HQ and is sort of responsible for
0:33
SPEAKER_01: the day-to-day maintenance and handling of their forecast,
0:38
SPEAKER_01: their 2026 election model.
0:41
SPEAKER_01: Why don't you guys,
0:43
SPEAKER_01: for people,
0:43
SPEAKER_01: Geoffrey and I have done one,
0:45
SPEAKER_01: when they unveiled their House-Senate forecast, Geoffrey and I did a Substack Live.
0:51
SPEAKER_01: We're going to kind of every two-ish weeks,
0:53
SPEAKER_01: we won't do another one until after Labor Day,
0:55
SPEAKER_01: but every two-ish weeks,
0:56
SPEAKER_01: we're going to check in,
0:57
SPEAKER_01: the three of us, and sort of where we are, where are we now.
1:00
SPEAKER_01: But for people who miss that and who...
1:02
SPEAKER_01: May not totally grasp the difference between polling and modeling and what this forecast shows.
1:08
SPEAKER_01: Why don't you talk about how you do what you do, and then let's talk about what it shows.
1:15
SPEAKER_00: Yeah, sure.
1:16
SPEAKER_00: So, Madison,
1:18
SPEAKER_00: if you want to give a rudimentary explanation of the forecast model,
1:23
SPEAKER_00: and then I can talk a little bit about what it shows.
1:26
SPEAKER_00: But since you're dealing with this day-to-day, hour-to-hour...
1:31
SPEAKER_02: On demand, yeah.
1:33
SPEAKER_02: So basically our model, it takes in polling, like polling is a big part of it.
1:36
SPEAKER_02: I kind of,
1:37
SPEAKER_02: I like to categorize,
1:38
SPEAKER_02: you know,
1:38
SPEAKER_02: how it works and the stuff that goes into it as like stuff that we know and
1:41
SPEAKER_02: then stuff that kind of changes.
1:43
SPEAKER_02: So the stuff that we know is like fundamentals,
1:45
SPEAKER_02: like Big one in there is like how certain county or how districts like voted in
1:50
SPEAKER_02: the 2024 election.
1:52
SPEAKER_02: If there's an incumbent, stuff like that.
1:53
SPEAKER_02: I mean, incumbency might change,
1:55
SPEAKER_02: you know,
1:55
SPEAKER_02: with primaries and such,
1:57
SPEAKER_02: but, you know, kind of like pretty more like stagnant variables.
2:00
SPEAKER_02: Right.
2:00
SPEAKER_02: And those have a lot of impact.
2:02
SPEAKER_02: Those have really a lot of impact, especially in districts who are polling.
2:06
SPEAKER_02: And it might not be, you know, like not everybody's, you know, I'm from California.
2:09
SPEAKER_02: I'm from California 10.
2:10
SPEAKER_02: Nobody's going to poll California 10.
2:12
SPEAKER_02: I'm from Connecticut.
2:13
SPEAKER_01: There's not a whole hell of a lot of general election polling.
2:17
SPEAKER_01: Yes.
2:17
SPEAKER_02: Yeah.
2:18
SPEAKER_02: Polling is great when you have it, but when you don't have it, you need something.
2:20
SPEAKER_02: So there's something that we have is historical data.
2:24
SPEAKER_02: And, you know, that kind of covers like what we like to call the fundamentals.
2:27
SPEAKER_02: There's some FEC stuff in there, stuff like that.
2:30
SPEAKER_02: And then there's the stuff that we don't know that changes every day.
2:34
SPEAKER_02: And a lot of that is polling.
2:35
SPEAKER_02: That's probably like the largest thing.
2:38
SPEAKER_02: We put quite a bit of emphasis onto that because that's kind
2:41
SPEAKER_02: of our most reliable like indicator of signal outside of, you know.
2:49
SPEAKER_02: Candidate-specific signals.
2:50
SPEAKER_02: So think for Maine,
2:52
SPEAKER_02: the model has no idea the difference between Graham Plattner and Troy Jackson,
2:55
SPEAKER_02: but the poll's And the prediction market,
2:58
SPEAKER_02: which I'll get into next, absolutely understand who those people are.
3:01
SPEAKER_02: And so then the other part of that, which is new this year, is the prediction markets.
3:05
SPEAKER_02: And those are kind of how we,
3:07
SPEAKER_02: I guess,
3:08
SPEAKER_02: get ahead of the polls a little bit because polls have their own inherent lag
3:12
SPEAKER_02: to them.
3:12
SPEAKER_02: So that's kind of,
3:13
SPEAKER_02: I think,
3:14
SPEAKER_02: like the guts of how it works or the stuff that goes into it,
3:17
SPEAKER_02: the components of the car, if you will.
3:19
SPEAKER_02: And Madison,
3:20
SPEAKER_02: can I
3:20
SPEAKER_01: just ask for sort of the broader audiences,
3:23
SPEAKER_01: like how do you decide how much the polling should matter?
3:27
SPEAKER_01: How much do you decide how what Polymarket and Calci are saying should matter?
3:33
SPEAKER_01: How much do you decide how much money the candidates have raised should matter?
3:36
SPEAKER_01: And does that Change on a daily, weekly,
3:41
SPEAKER_01: or monthly basis,
3:42
SPEAKER_01: or do you have like a set, okay,
3:44
SPEAKER_01: polling is going to matter 60%,
3:46
SPEAKER_01: 20%, 20%, and that doesn't change through the whole election.
3:52
SPEAKER_02: Yeah, so with those two, I mean, the answer is that there's no right answer, right?
3:57
SPEAKER_02: We might come out of this November 4th or whatever and be like,
4:00
SPEAKER_02: okay, so we weighted our markets way too heavily or way too lightly.
4:04
SPEAKER_02: The markets were super correct and we were super wrong, stuff like that.
4:07
SPEAKER_02: So, you know, I might have a better answer for you for that November 5th, but...
4:12
SPEAKER_02: Um, until then,
4:13
SPEAKER_02: you know,
4:13
SPEAKER_02: we kind of like, you know, we decide kind of how much weight everything should get.
4:17
SPEAKER_02: And the weight is at this point dependent on like how many poles.
4:22
SPEAKER_02: So, you know, a race with one pole is not going to get the same weight as 50 poles, right?
4:26
SPEAKER_02: That just doesn't make sense, especially if it's an internal or something like that.
4:29
SPEAKER_02: Same thing with markets.
4:30
SPEAKER_02: So we weight markets based on how much money is in them.
4:33
SPEAKER_02: Um, that, that is potentially subject to change.
4:36
SPEAKER_02: Um, but you know,
4:37
SPEAKER_02: a market where one guy has put in a hundred dollars should probably not have
4:41
SPEAKER_02: as much empirical basis as a market with like,
4:45
SPEAKER_02: I think some of them now are at like 30 million.
4:47
SPEAKER_02: So that's, that's a big part of it.
4:49
SPEAKER_02: Um, those get weighted their max weights are 60 or 45 and 25%,
4:54
SPEAKER_02: um, 45 for polling, 25 for, um, markets.
5:01
SPEAKER_02: That might be right.
5:02
SPEAKER_02: That might be wrong.
5:04
SPEAKER_02: There's a lot of discussion.
5:05
SPEAKER_02: And if you go through a methodology, we have some academic papers cited.
5:08
SPEAKER_02: But especially with the markets,
5:09
SPEAKER_02: this is a new thing for pretty much everyone that's putting out
5:12
SPEAKER_02: a forecast this year is how do you incorporate this in
5:15
SPEAKER_02: a scientifically sound manner?
5:18
SPEAKER_02: So we'll come out of this and we'll kind of see who's right.
5:20
SPEAKER_02: But we feel pretty, I would say, pretty confident in our kind of decision behind it.
5:26
SPEAKER_01: And let me one other thing before we get into the specifics of what
5:31
SPEAKER_01: the forecast says right now about House Senate, et cetera.
5:35
SPEAKER_01: Can you explain to people why,
5:38
SPEAKER_01: probably the modeler that people know the best,
5:40
SPEAKER_01: the average person is Nate Silver, right, would be my guess.
5:43
SPEAKER_01: It's like they're most familiar with him.
5:44
SPEAKER_01: He's now on Substack.
5:45
SPEAKER_01: He is a very popular Substack.
5:47
SPEAKER_01: He released his model, his forecast, I think, was it last week, Skelley?
5:51
SPEAKER_01: Sometime relatively recently.
5:53
SPEAKER_01: And it was pretty bullish on Democratic hopes.
5:57
SPEAKER_01: It was Democrats are going to win the House.
5:58
SPEAKER_01: Democrats are more likely than not to win the Senate.
6:01
SPEAKER_01: I don't need you to Picket that apart,
6:04
SPEAKER_01: but explain to people why different models show different things.
6:10
SPEAKER_01: Is it all just based on how much you weight those,
6:12
SPEAKER_01: what inputs you have and how much you weight those inputs?
6:16
SPEAKER_02: More or less, yeah.
6:17
SPEAKER_02: I think it depends on how much you care about certain inputs and how much you trust them, right?
6:21
SPEAKER_02: So for us,
6:23
SPEAKER_02: I would say our model is a little bit more bullish on Republicans this year compared
6:27
SPEAKER_02: to some of the others out there.
6:29
SPEAKER_02: And that's because we don't necessarily Yeah.
6:32
SPEAKER_02: Yeah.
6:32
SPEAKER_02: Yeah.
6:59
SPEAKER_00: For instance, Chris,
7:00
SPEAKER_00: for instance,
7:00
SPEAKER_00: you know,
7:01
SPEAKER_00: a lot of this goes back to, I mean,
7:03
SPEAKER_00: some of this can go back to the data that the forecast model is training on,
7:07
SPEAKER_00: essentially.
7:08
SPEAKER_00: You know, how far back does the data go?
7:10
SPEAKER_00: What variables are they bringing in?
7:13
SPEAKER_00: Some forecast models are simpler in terms of the inputs,
7:16
SPEAKER_00: and some are adding in additional variables.
7:19
SPEAKER_00: And so all of that can affect the probabilities that spit out.
7:24
SPEAKER_01: Skelley,
7:24
SPEAKER_01: for people who want to go deep,
7:26
SPEAKER_01: there is a way to, you guys have your methodology public, right?
7:30
SPEAKER_01: Like people can see how the decisions you're making.
7:34
SPEAKER_01: Can one of you two just post that in the live chat at some point?
7:37
SPEAKER_02: It does not go to this second,
7:38
SPEAKER_02: but for people who want to get- There's a lot of Greek characters.
7:42
SPEAKER_01: Thank you, Skelley.
7:43
SPEAKER_01: Here's the reality, folks.
7:45
SPEAKER_01: Both Madison and Geoffrey are much smarter than me and probably you.
7:48
SPEAKER_01: Like, to be totally honest, like,
7:50
SPEAKER_01: they're numbers people who know how,
7:52
SPEAKER_01: but if you want to go deep, Look,
7:56
SPEAKER_01: if people won't tell you how their model works or their polling works,
8:00
SPEAKER_01: that is red flag number one, right?
8:03
SPEAKER_01: So check that out.
8:04
SPEAKER_01: Okay, let's get,
8:05
SPEAKER_01: first of all,
8:06
SPEAKER_01: let me say to people who are on the live chat,
8:07
SPEAKER_01: if you have a specific race that you want us to talk about,
8:11
SPEAKER_01: let us know and we'll get to it.
8:13
SPEAKER_01: But let's go, can we start on the Senate?
8:16
SPEAKER_01: Because I think that's,
8:17
SPEAKER_01: why don't we start on the Senate and we can jump to the House too,
8:19
SPEAKER_01: but let's start on the Senate.
8:20
SPEAKER_01: Geoffrey,
8:21
SPEAKER_01: if you want to share screens and stuff,
8:22
SPEAKER_01: feel free to,
8:23
SPEAKER_01: Let's talk about where you guys are now and where you've been from when you launched
8:28
SPEAKER_01: the model and where we are now.
8:30
SPEAKER_00: Yeah, so let's see.
8:31
SPEAKER_00: Oh, I thought I'd already gotten this shared, but let's see.
8:35
SPEAKER_00: Okay, so for folks at home, if you go to our website for- Which is great.
8:42
SPEAKER_00: All of our election data is votes.decisiondeskhq.com.
8:48
SPEAKER_00: Our forecast is there and you can look at it.
8:50
SPEAKER_00: It's free to sign up.
8:52
SPEAKER_00: And right now we give Republicans a very slim edge.
8:57
SPEAKER_00: It is close to a coin flip for control of the Senate.
9:01
SPEAKER_00: Remember,
9:02
SPEAKER_00: you know,
9:02
SPEAKER_00: we're dealing with probabilities of things like 54 percent
9:06
SPEAKER_00: to 46 percent is not terribly different from 50 50.
9:10
SPEAKER_00: That's why I say it's close to a coin flip.
9:12
SPEAKER_00: And Republicans have had a slight edge for a while now,
9:15
SPEAKER_00: but I would say it has shrunk a bit as the overall electoral environment,
9:19
SPEAKER_00: which we mostly get a read from the generic ballot,
9:23
SPEAKER_00: which is a question that asks polling respondents,
9:26
SPEAKER_00: do you plan to vote for the Democrat or the Republican in your local U.S.
9:29
SPEAKER_00: House race?
9:30
SPEAKER_00: Remember,
9:30
SPEAKER_00: the U.S.
9:31
SPEAKER_00: House race is actually a national election because all four and 35 seats are on
9:34
SPEAKER_00: the ballot, whereas in the Senate, only about a third of Senate seats are up.
9:39
SPEAKER_00: Steve McLaughlin.
10:02
SPEAKER_00: Looking at the presidential vote in 2024,
10:05
SPEAKER_00: President Trump carried the national popular vote by about a point and a half.
10:08
SPEAKER_00: Democrats lead the generic ballot average by about six percentage points.
10:13
SPEAKER_00: So that's basically a seven and a half point swing to the left from where
10:17
SPEAKER_00: the country was in 2024.
10:19
SPEAKER_00: So, you know,
10:20
SPEAKER_00: if you sort of think of that as a back of the envelope sort of baseline for
10:24
SPEAKER_00: for what's going on in the national political environment,
10:26
SPEAKER_00: that has allowed a what is a Republican friendly map.
10:29
SPEAKER_00: I mean, this is the seats that are up.
10:31
SPEAKER_00: Democrats do not have a lot of
10:34
SPEAKER_01: I was going
10:35
SPEAKER_01: to say, the only good thing about the map for Democrats is there are 22 Republican seats
10:39
SPEAKER_01: and only 13 Democratic seats.
10:41
SPEAKER_01: But of the 22,
10:42
SPEAKER_01: we always cite this,
10:43
SPEAKER_01: but if you need it,
10:44
SPEAKER_01: of the 22 seats that Republicans are defending,
10:47
SPEAKER_01: one main was carried by Kamala Harris in 2024.
10:51
SPEAKER_01: So you're just fighting on the raw numbers look good for Democrats.
10:55
SPEAKER_01: You go race by race, it gets harder.
10:57
SPEAKER_01: Sorry, Geoffrey.
10:57
SPEAKER_00: Yeah, because then from there,
10:59
SPEAKER_00: there's North Carolina, which President Trump only narrowly carried in 2024.
11:02
SPEAKER_00: But then after Maine and North Carolina,
11:05
SPEAKER_00: to get a majority,
11:05
SPEAKER_00: Democrats have to win seats that President Trump carried by more
11:08
SPEAKER_00: than 10 percentage points.
11:09
SPEAKER_00: Ohio, which was about 11.
11:12
SPEAKER_00: And then some group
11:15
SPEAKER_00: Texas, Iowa, Alaska,
11:18
SPEAKER_00: and Florida were all in this group that were 13 to 14 points in favor
11:22
SPEAKER_00: of President Trump in 2024.
11:23
SPEAKER_00: So you have to flip those kinds of seats to get a majority.
11:27
SPEAKER_00: And so because of the national political environment and also
11:30
SPEAKER_00: the individual race polling,
11:31
SPEAKER_00: which suggests something about Democrats in a lot
11:34
SPEAKER_00: of those races have generally nominated candidates.
11:37
SPEAKER_00: Decent candidates.
11:39
SPEAKER_00: They have they're very competitive in a lot of those contests.
11:42
SPEAKER_00: And that's why the Republican edge at this point is has shrunk to close to
11:48
SPEAKER_00: a pure toss up on the set.
11:50
SPEAKER_01: And I just want to quickly echo what Geoffrey is saying and what
11:54
SPEAKER_01: the point Madison made
11:56
SPEAKER_00: earlier about polling.
11:57
SPEAKER_01: Look, if you were.
11:58
SPEAKER_01: Uh.
12:01
SPEAKER_01: Mary Peltola in Alaska.
12:03
SPEAKER_01: And there was a poll that came out that had you at 54% and Dan Sullivan,
12:06
SPEAKER_01: the incumbent Republican at 46%.
12:08
SPEAKER_00: You'd be like, all right, up eight points.
12:11
SPEAKER_00: It's all happening
12:12
SPEAKER_01: for me.
12:13
SPEAKER_01: That is not what this is showing.
12:16
SPEAKER_01: Skelley is sick, probably sick to death of me making this analogy, but I like this one.
12:20
SPEAKER_01: If you get on a plane, okay, and the pilot says, thank you so much for flying the airways.
12:27
SPEAKER_01: There's a 54% chance that this plane will land safely.
12:32
SPEAKER_01: You are like...
12:34
SPEAKER_00: Right?
12:35
SPEAKER_00: Give me out.
12:35
SPEAKER_01: There's just no way in hell that... What this means is if they run the simulation...
12:42
SPEAKER_01: A thousand times,
12:43
SPEAKER_01: 540 times Republicans retain control of the Senate,
12:48
SPEAKER_01: 460 times Democrats win the Senate.
12:51
SPEAKER_01: Like people don't see,
12:53
SPEAKER_01: I feel like we have to say that every time Skelly about probability
12:55
SPEAKER_01: because people don't get it.
12:57
SPEAKER_01: This is not Republicans have an 8% lead in their chances of winning.
13:02
SPEAKER_01: This is basically saying a 50-50-ish race, right Madison?
13:07
SPEAKER_01: Yeah.
13:07
SPEAKER_02: Yeah, pretty much.
13:08
SPEAKER_02: I think, Geoffrey, could you pivot to the seat count, actually?
13:11
SPEAKER_02: I think that that's a great indicator of how this is going,
13:15
SPEAKER_02: is that the 50-50 seat count literally has not moved,
13:19
SPEAKER_02: basically,
13:20
SPEAKER_02: since I think we've started running it internally a little bit before launch.
13:24
SPEAKER_02: And tell people what that means, Madison.
13:27
SPEAKER_02: Tell people what that means.
13:28
SPEAKER_02: Yeah, yeah.
13:29
SPEAKER_02: So basically it means that even if there's a 60% chance of Republicans winning
13:34
SPEAKER_02: the Senate, they're on average not doing it by a wide margin.
13:37
SPEAKER_02: You're not going to get a sweep.
13:37
SPEAKER_02: You're not going to get two thirds.
13:39
SPEAKER_02: You're going to get 50-50 with J.D.
13:40
SPEAKER_02: Vance as a tiebreaker.
13:42
SPEAKER_02: And this is a much prettier picture for Republicans in the Senate than it is
13:49
SPEAKER_02: for Democrats in the House because that number is a little bit more all over
13:52
SPEAKER_02: the place.
13:54
SPEAKER_02: But basically,
13:55
SPEAKER_02: it's just kind of signaling that, yes,
14:01
SPEAKER_02: the Republicans have a slight edge in probability,
14:04
SPEAKER_02: but that does not mean that they necessarily have it in the bag.
14:06
SPEAKER_02: The path to victory is a lot harder for Democrats, but it's not impossible.
14:11
SPEAKER_01: Stella, you and I have become one person,
14:14
SPEAKER_01: because I was literally just going to ask if you could show this.
14:17
SPEAKER_01: Explain to people...
14:19
SPEAKER_01: I want you to run through this.
14:20
SPEAKER_01: While you run through it, can you explain to people how many simulations you guys are running?
14:24
SPEAKER_01: Because, you know, they'll see this thing that says 102,901 simulations.
14:30
SPEAKER_01: How many simulations are being run to produce these results?
14:34
SPEAKER_01: And explain what we're looking at here, too.
14:36
SPEAKER_02: Yeah, so we run a million simulations every time the forecast gets updated,
14:41
SPEAKER_02: a million get rerun.
14:43
SPEAKER_02: So basically what this means is you can see the most probable outcome
14:47
SPEAKER_02: in this scenario for the Senate is,
14:50
SPEAKER_02: and even Senate seats obviously advance as a tiebreaker.
14:54
SPEAKER_02: But this basically means in a million simulations that take into account all
14:57
SPEAKER_02: of our probabilities,
14:59
SPEAKER_02: This is the raw number of outputs or outcomes and how many seats the Republicans
15:06
SPEAKER_02: and Democrats have.
15:07
SPEAKER_02: So you can see that Republicans plus six or something.
15:11
SPEAKER_02: Republicans plus six.
15:16
SPEAKER_02: It's possible.
15:16
SPEAKER_02: That's the current.
15:17
SPEAKER_00: To be clear, that's the current, the current makeup of the Senate right now.
15:21
SPEAKER_02: That was a good accidental poll on my end.
15:24
SPEAKER_02: So that like you can see that it's for the model, not super likely, but not super unlikely.
15:31
SPEAKER_02: It's, you know, seven, seven percent of the time.
15:34
SPEAKER_02: But, you know,
15:36
SPEAKER_02: so that just basically means that Republicans are,
15:40
SPEAKER_02: you know, keeping their keeping their current share right.
15:43
SPEAKER_02: Like they're going to lose seats pretty much.
15:45
SPEAKER_02: Right.
15:46
SPEAKER_02: And in, you know, yeah,
15:47
SPEAKER_02: it happens 70,000 times out of a million,
15:50
SPEAKER_02: but that's,
15:51
SPEAKER_02: that's definitely not zero, but it's definitely not close to a million either.
15:54
SPEAKER_01: Right.
15:54
SPEAKER_01: It's that 700,000 times.
15:56
SPEAKER_01: So right now, according to the model.
15:58
SPEAKER_01: The fat part of this curve.
15:59
SPEAKER_00: Yeah, go ahead.
15:59
SPEAKER_00: Really?
16:00
SPEAKER_00: The fat part of this is sort of the most likely range.
16:02
SPEAKER_00: It's like sort of.
16:05
SPEAKER_00: Democrats actually end up winning all those toss-up seats and maybe pulling out
16:09
SPEAKER_00: an unexpected win in a place like Florida or maybe in Nebraska,
16:13
SPEAKER_00: something weird like that.
16:14
SPEAKER_00: And Republicans roughly holding serve.
16:19
SPEAKER_00: Maybe they flip Michigan but lose North Carolina, and
16:23
SPEAKER_00: then actually Republicans end up narrowly winning all those other really close
16:26
SPEAKER_00: seats.
16:27
SPEAKER_00: So these are sort of the most likely set of outcomes.
16:29
SPEAKER_00: There are more unlikely ones, but they are not nearly as common.
16:34
SPEAKER_00: But, you know,
16:35
SPEAKER_00: that's sort of,
16:35
SPEAKER_00: I think for people's understanding of like,
16:37
SPEAKER_00: this is sort of the universe of outcomes and the most likely set of outcomes.
16:42
SPEAKER_00: And then that averages out to this 50-50.
16:45
SPEAKER_01: I mean, that's crazy that the most likely outcome,
16:48
SPEAKER_01: according to the model right now, is 50-50 with J.D.
16:51
SPEAKER_01: Vance breaking ties, which is a net game
16:53
SPEAKER_00: for Democrats, just to be clear.
16:55
SPEAKER_01: If that happens, Lisa Murkowski will become the most highly coveted.
17:01
SPEAKER_01: And we'll hear a lot about John Fetterman.
17:03
SPEAKER_01: John Fetterman and Lisa Murkowski, their phones will be ringing off the hook.
17:07
SPEAKER_01: Geoffrey,
17:07
SPEAKER_01: before we move to House,
17:09
SPEAKER_01: can you go to where you guys have the—because they model the whole Senate,
17:14
SPEAKER_01: but they also model each individual race.
17:16
SPEAKER_01: Can you go to where you have those?
17:18
SPEAKER_01: Yeah, you can either look at it on the map, which is nice.
17:24
SPEAKER_01: Obviously toss-ups and then there are toss-ups, right?
17:26
SPEAKER_01: Like North Carolina at this point, it's not really, I don't feel like a toss-up.
17:30
SPEAKER_01: Georgia is not really a toss-up according to the model, right?
17:34
SPEAKER_01: 80% probability, 79% probability.
17:36
SPEAKER_01: Let's go through the one, two, three, four.
17:38
SPEAKER_01: Let's go through the five yellows, which are the straight toss-ups.
17:41
SPEAKER_01: You guys just want to like walk through where the model says about those.
17:45
SPEAKER_01: Yeah, I was going to say, Skelly, start on the East Coast and go the other way.
17:49
SPEAKER_00: Yeah, so Maine, super tight right now.
17:51
SPEAKER_00: It's basically a coin flip in Maine, according to our forecast.
17:55
SPEAKER_00: If you click on the individual ones just to show people what this shows,
17:59
SPEAKER_00: it's like, again,
18:00
SPEAKER_00: that's not the percentage,
18:01
SPEAKER_00: although you can see their projected vote share if you want to click on it.
18:04
SPEAKER_00: And actually what's interesting on this one is- What is what
18:06
SPEAKER_01: the model suggests they will get in terms of vote share is
18:11
SPEAKER_01: that projected vote share.
18:12
SPEAKER_00: Yeah.
18:13
SPEAKER_00: So it's interesting in this case,
18:14
SPEAKER_00: actually,
18:14
SPEAKER_00: that it's a 50-50 race,
18:15
SPEAKER_00: but Collins very narrowly actually leads on the vote share thing.
18:19
SPEAKER_00: But obviously in a race this close, it's like the whole idea is that everything points to close.
18:25
SPEAKER_00: And you can scroll down and you can see actually currently what our forecast,
18:28
SPEAKER_00: like what percentage of the forecast comes from our fundamentals,
18:32
SPEAKER_00: which Madison was talking about is things like the historical partisanship of
18:37
SPEAKER_00: the state and things like that.
18:39
SPEAKER_00: The polls, because we actually don't have that much polling right now.
18:42
SPEAKER_00: And so that's just a good example.
18:43
SPEAKER_00: So just to go back to the overall map.
19:07
SPEAKER_00: Maine, 50-50.
19:09
SPEAKER_00: Ohio, basically 50-50.
19:11
SPEAKER_00: Sherrod Brown with a 50-50 shot at ousting John Husted, the appointed Republican incumbent.
19:18
SPEAKER_00: And then Texas,
19:19
SPEAKER_00: I should actually go to next because it is basically in the same,
19:23
SPEAKER_00: almost the exact same point where it's also very 50-50.
19:26
SPEAKER_00: And so these three races are basically coin flips, pure, almost pure coin flips.
19:32
SPEAKER_00: Iowa, a little bit more of an edge for closer to that likely Republican line.
19:36
SPEAKER_00: So the way we categorize these is that if no candidate has at least a 60% chance
19:41
SPEAKER_00: of winning, we call it a toss up.
19:44
SPEAKER_00: So Henson's close to that line, but she's not quite there.
19:47
SPEAKER_00: So it's still in our toss up category.
19:50
SPEAKER_00: And then in Alaska,
19:51
SPEAKER_00: it's sort of the reverse of that where Mary Peltola has a slight Slight edge,
19:56
SPEAKER_00: but slightest of edges, and it's still a toss up.
19:58
SPEAKER_00: And so those are sort of,
20:00
SPEAKER_00: you know,
20:00
SPEAKER_00: there's a good chance that it's those five seats and Michigan,
20:03
SPEAKER_00: where we do have Abdul El-Sayed is,
20:06
SPEAKER_00: you know,
20:07
SPEAKER_00: has an edge,
20:07
SPEAKER_00: but for,
20:08
SPEAKER_00: you know,
20:09
SPEAKER_00: considering we're in an environment where North Carolina is,
20:12
SPEAKER_00: And Georgia are looking pretty good for Democrats.
20:15
SPEAKER_00: The fact that Michigan, which,
20:17
SPEAKER_00: by the way,
20:17
SPEAKER_00: President Trump won by less than he won North Carolina and about the same
20:23
SPEAKER_00: as he won Georgia,
20:24
SPEAKER_00: the fact that that's not quite as comfortable says something about what all
20:28
SPEAKER_00: the conversation is about Michigan being really critical
20:30
SPEAKER_00: because Democrats control this seat and if they lose it,
20:34
SPEAKER_00: It gets a lot harder for them to win.
20:49
SPEAKER_01: I think that's true.
21:10
SPEAKER_01: It gets a hell of a lot harder for Democrats to win the majority.
21:13
SPEAKER_01: If Abdul Syed loses, it gets a hell of a lot harder.
21:16
SPEAKER_01: Have you guys updated that, I should say?
21:20
SPEAKER_02: I'm looking at yesterday's.
21:21
SPEAKER_02: I don't have today's,
21:22
SPEAKER_02: but I'm looking at yesterday's numbers because they're stale in a notebook.
21:26
SPEAKER_02: And both of those scenarios are at about 80%.
21:30
SPEAKER_02: So if Republicans win either Michigan or Maine, they have an 80% chance of taking the Senate.
21:35
SPEAKER_02: And that I think just really underscores like Democrats can't lose.
21:40
SPEAKER_02: They can't lose and they have to win.
21:42
SPEAKER_02: Republicans can kind of like do nothing and maybe be fine.
21:47
SPEAKER_00: Democrats have to win their most winnable seats or hold the seats that they
21:53
SPEAKER_00: on paper should be able to hold in an environment that's D plus six,
21:57
SPEAKER_00: as we were talking about, ballot D plus six.
22:00
SPEAKER_01: All right, let's go.
22:01
SPEAKER_01: And I will say to people,
22:03
SPEAKER_01: we're going to go through the House and then we will bring your questions,
22:08
SPEAKER_01: whether individual races, questions about the model, to Madison and Geoffrey.
22:12
SPEAKER_01: So if you have questions about individual races,
22:14
SPEAKER_01: about the model, about forecasting, how they do it, put them in.
22:16
SPEAKER_01: I promise I see them and we will go through them.
22:18
SPEAKER_01: So let's go to the House.
22:20
SPEAKER_01: So just again, Republicans are narrowly favored to hold the House and the Senate.
22:25
SPEAKER_01: And this is the House.
22:26
SPEAKER_01: Go ahead.
22:27
SPEAKER_00: Yeah.
22:27
SPEAKER_00: So here we have Democrats as at least somewhat favored, not overwhelmingly so.
22:32
SPEAKER_00: There's still definitely universal Republicans holding the House.
22:35
SPEAKER_00: But right now, Democrats have about a two and three shot of winning the House.
22:41
SPEAKER_00: And in terms of the seat count right now,
22:43
SPEAKER_00: the average seat count of, again,
22:45
SPEAKER_00: remember,
22:46
SPEAKER_00: much like with the Senate, but this time with 435 seats instead of 100 seats.
22:50
SPEAKER_00: And this number does bounce around a bit more because you're talking about a lot more seats.
22:54
SPEAKER_00: But the average outcome of those forecast simulations is
22:58
SPEAKER_00: that Democrats have 229 seats and Republicans have 206.
23:01
SPEAKER_00: So right now,
23:02
SPEAKER_00: Republicans have 220 to the Democrats 215,
23:06
SPEAKER_00: although it is worth noting that because of redistricting,
23:08
SPEAKER_00: That baseline number is maybe not the base,
23:11
SPEAKER_00: not really the actual baseline because enough states obviously redistricted.
23:14
SPEAKER_00: That's been a big part of the conversation.
23:17
SPEAKER_00: But nonetheless,
23:18
SPEAKER_00: if you're sort of operating off of where the House is right now in terms
23:22
SPEAKER_00: of its makeup, that's...
23:24
SPEAKER_00: But that would effectively translate to...
23:26
SPEAKER_01: I was going to say, it would effectively translate to like a D plus 14 or 15, right?
23:30
SPEAKER_01: Is that like based on what we expect?
23:32
SPEAKER_01: Gains from where they are right now.
23:33
SPEAKER_01: Yeah, yeah.
23:34
SPEAKER_01: Which they would be thrilled with.
23:36
SPEAKER_01: Yeah.
23:37
SPEAKER_01: Madison, question on how you do the House versus how you do the Senate.
23:41
SPEAKER_01: There's obviously less,
23:42
SPEAKER_01: you were saying one of the challenges is like if each,
23:45
SPEAKER_01: if you are from California 10, that's, is that Ellen Tauscher's old district?
23:50
SPEAKER_01: Uh, ooh.
23:51
SPEAKER_01: That's probably before your time.
23:52
SPEAKER_02: I might be, I might be too young for that.
23:54
SPEAKER_01: Ellen Tauscher, well, they changed the damn number, Skelly and I run into this all the time.
23:58
SPEAKER_01: They changed the numbers so much, it's impossible.
24:00
SPEAKER_02: They do, they do.
24:01
SPEAKER_02: I think it used to be 11, so.
24:03
SPEAKER_01: If you're from those places or you're from Connecticut and there isn't
24:06
SPEAKER_01: as much polling in like a Connecticut House race,
24:09
SPEAKER_01: how do you,
24:09
SPEAKER_01: does that,
24:10
SPEAKER_01: there's obviously more Senate polling than there is House polling,
24:13
SPEAKER_01: right, at this point.
24:14
SPEAKER_01: Yes.
24:15
SPEAKER_01: Does that,
24:15
SPEAKER_01: do you just lean on the generic ballot number in the House more when it comes to
24:19
SPEAKER_01: the kind of polling piece of the model?
24:21
SPEAKER_01: Like what other polling numbers get put in there?
24:26
SPEAKER_02: Yeah, so generic ballot's kind of it,
24:28
SPEAKER_02: but generic ballot and polling get treated very differently in the model.
24:33
SPEAKER_02: So generic ballot kind of is used mostly to adjust past presidential results
24:42
SPEAKER_02: for environment, right?
24:44
SPEAKER_02: So obviously it's not really fair to...
24:47
SPEAKER_02: Compare, let's say, Michigan's, you know, but lean towards Trump.
24:50
SPEAKER_02: It's not really fair to compare that in a R plus two environment as
24:54
SPEAKER_02: when we're currently sitting in a D plus six environment, if that is the true value.
24:58
SPEAKER_02: Right.
24:59
SPEAKER_02: So those kind of they get handled a little bit differently.
25:03
SPEAKER_02: And so I would, you know.
25:05
SPEAKER_02: The generic ballot kind of goes through all of the fun models we have,
25:11
SPEAKER_02: all the fun data science-y stuff,
25:13
SPEAKER_02: whereas that number ends up getting adjusted on polling.
25:17
SPEAKER_02: So it spits out a number,
25:19
SPEAKER_02: converted to a probability, and then that probability gets weighted against polling.
25:23
SPEAKER_02: So polling is definitely kind of untouched and just gets blended
25:27
SPEAKER_02: in versus being modeled on,
25:30
SPEAKER_02: getting technical, but Um,
25:33
SPEAKER_01: but you guys do, you guys do filter.
25:37
SPEAKER_01: You don't take any poll and just pump it into the model, right?
25:41
SPEAKER_01: Like you have a system of what's credible because we were,
25:45
SPEAKER_01: I had lunch this week with,
25:47
SPEAKER_01: with Scott Schranter, DDHQ, Kingpin, and we were talking about this at scale.
25:51
SPEAKER_01: Is it median strategies or the hell it's called?
25:54
SPEAKER_01: They did polling in Wisconsin and they basically got,
25:56
SPEAKER_01: it seems to me they got found out that
25:58
SPEAKER_00: It was BS and
25:59
SPEAKER_01: fake,
26:00
SPEAKER_01: but they were like,
26:00
SPEAKER_01: this was a social experiment and it was never meant to be polling.
26:03
SPEAKER_01: But like, there's a lot of junk out there.
26:05
SPEAKER_01: Like, how do you guys do that?
26:06
SPEAKER_01: Because I always say it's like a car.
26:09
SPEAKER_01: You put bad gas in, the car's not going to perform.
26:12
SPEAKER_01: So how do you guys make sure that there's,
26:14
SPEAKER_01: with this explosion of polling firms that none of us have ever heard of before,
26:20
SPEAKER_01: that many of which don't release any of their methodology,
26:22
SPEAKER_01: how do you guys make sure that that stuff is not going in and corrupting
26:26
SPEAKER_01: the polling chunk of the model?
26:28
SPEAKER_02: Yeah.
26:29
SPEAKER_02: So, I mean, I think this just generally speaks to our polling aggregation in general.
26:33
SPEAKER_02: But we, you know, we have our pollsters that we like.
26:37
SPEAKER_02: We do take in quite a bit.
26:39
SPEAKER_02: And I'll get to that point in a second.
26:41
SPEAKER_02: But we do take in quite a bit.
26:43
SPEAKER_02: We have pollsters.
26:44
SPEAKER_02: Obviously, when Median Strategies came out, we couldn't really nail down their methodology.
26:49
SPEAKER_02: And so we were like, let's pump the brakes on this.
26:51
SPEAKER_02: And if they publish a methodology, we'll add their stuff in.
26:54
SPEAKER_02: If they don't, we'll wait it out.
26:55
SPEAKER_02: We'll see what happens.
26:57
SPEAKER_02: And I would maintain that that was the right decision to make.
27:01
SPEAKER_02: And there's been kind of some similar errors in the polling world in the past couple
27:05
SPEAKER_02: of days as well.
27:07
SPEAKER_02: And so we cast a wide net,
27:10
SPEAKER_02: I would say,
27:10
SPEAKER_02: but we still hold a lot of scrutiny to the pollsters and their polls that we put
27:15
SPEAKER_02: in there, especially in poll averages.
27:17
SPEAKER_02: I believe it's we don't take internal polls.
27:20
SPEAKER_02: Because,
27:21
SPEAKER_02: you know,
27:21
SPEAKER_02: that's not necessarily the most scientifically accurate measure
27:25
SPEAKER_02: of how things are going.
27:27
SPEAKER_02: But also to the point of a wide net,
27:29
SPEAKER_02: I think that there's a high degree to which the law of large numbers and having
27:34
SPEAKER_02: a lot of stuff kind of works in our favor because it inevitably cancels out
27:39
SPEAKER_02: the weird ones also.
27:41
SPEAKER_01: Um, one more from me and then I'm going to get to people's questions.
27:45
SPEAKER_01: So folks,
27:46
SPEAKER_01: if you do have questions about individual races,
27:48
SPEAKER_01: about how Decision Desk HQ does their,
27:51
SPEAKER_01: uh, their 2026 election forecast, uh, put it in the chat.
27:55
SPEAKER_01: I would also urge you to go to votes.decisiondeskhq.com.
28:00
SPEAKER_01: You can look at all of this stuff that we are looking at.
28:03
SPEAKER_01: It is a great, great site.
28:06
SPEAKER_01: Um, I'm not sure if this is for one of you or both of you or for like maybe it's
28:12
SPEAKER_01: above your pay grades.
28:13
SPEAKER_01: How did you decide to include prediction markets in this?
28:20
SPEAKER_01: Why?
28:21
SPEAKER_01: Because prediction, look,
28:22
SPEAKER_01: I always say that Polymarket sponsors a fair amount of my content,
28:25
SPEAKER_01: not this, but they do sponsor me on Substack and on YouTube.
28:29
SPEAKER_01: I've taken plenty of flack for it.
28:30
SPEAKER_01: I always say to people,
28:31
SPEAKER_01: if that's the thing that drives you away,
28:32
SPEAKER_01: well, then that's the thing that drives you away.
28:34
SPEAKER_01: But why did you decide to incorporate them into the model?
28:39
SPEAKER_01: Because 10 years ago, people doing modeling, obviously that wasn't even a possibility.
28:45
SPEAKER_02: Yeah, I can tackle that one because I've been in a lot of meetings about the markets
28:49
SPEAKER_02: since I started here about six months ago.
28:52
SPEAKER_02: So, I mean,
28:54
SPEAKER_02: a lot of it is there,
28:55
SPEAKER_02: you know, there's the answer where it's like they're hot and interesting.
28:59
SPEAKER_02: But there's the other real answer of like,
29:02
SPEAKER_02: we believe at the very least that they provide signal, right?
29:05
SPEAKER_02: So, you know,
29:06
SPEAKER_02: as I kind of touched on earlier,
29:07
SPEAKER_02: polling has an inherent lag to it,
29:09
SPEAKER_02: like When, uh, when, uh,
29:12
SPEAKER_02: Senator Lindsey Graham passed away or when Graham Plattner dropped out, we pumped.
29:16
SPEAKER_00: Or when Darlene Graham says something stupid and we don't have any polls in
29:20
SPEAKER_00: the primary there to know, you know.
29:22
SPEAKER_02: Yes.
29:23
SPEAKER_02: Yeah.
29:23
SPEAKER_02: So stuff like that where,
29:25
SPEAKER_02: and it's also too,
29:26
SPEAKER_02: like the model is blind to candidates kind of inherently, um,
29:31
SPEAKER_02: And so we kind of thought the prediction markets and we believe in a lot
29:35
SPEAKER_02: of other forecasters out there believe that they kind of can like preempt
29:39
SPEAKER_02: the polling a little bit and maybe provide a more accurate signal,
29:44
SPEAKER_02: maybe not more accurate signal.
29:45
SPEAKER_02: I don't think we'll know that until after the election,
29:47
SPEAKER_02: but they at the very least can tell us something earlier
29:50
SPEAKER_02: than traditional polling can.
29:52
SPEAKER_02: And that's kind of always been a big miss with polling.
29:56
SPEAKER_02: It's good, but you have to wait for it to be good.
29:59
SPEAKER_02: And so obviously we would it's preferable for everybody that we don't wait.
30:03
SPEAKER_02: So.
30:04
SPEAKER_01: All right.
30:05
SPEAKER_01: Let me start getting into some of the questions.
30:07
SPEAKER_01: This is from Donna.
30:09
SPEAKER_01: Do you take into account what's going on in the country at the time of when
30:14
SPEAKER_01: the model comes out or do you assume that that's covered in the polls?
30:17
SPEAKER_01: Yeah.
30:19
SPEAKER_02: Um, I, I think that that's covered a little bit in, um, a lot of what we already have in there.
30:25
SPEAKER_02: So generic ballot's a good one, right?
30:27
SPEAKER_02: Obviously that's a poll, but generic ballot, you know, is our proxy for political environment.
30:32
SPEAKER_02: We also,
30:33
SPEAKER_02: you know,
30:33
SPEAKER_02: polling and markets,
30:34
SPEAKER_02: markets are a great way to kind of signal like what's going on in the country.
30:37
SPEAKER_02: If I could,
30:38
SPEAKER_02: you know, I'm,
30:39
SPEAKER_02: I'm sure some,
30:39
SPEAKER_02: some forecaster or,
30:41
SPEAKER_02: you know,
30:41
SPEAKER_02: bored teenager or something has decided to try to use like Google search trends
30:45
SPEAKER_02: to forecast, um, And I'm sure that you can get pretty close on that.
30:50
SPEAKER_02: But at the end of the day,
30:51
SPEAKER_02: it's going to be like 0.9 correlated with stuff that we already have in the model
30:54
SPEAKER_02: that is, you know, a lot easier to get.
30:58
SPEAKER_02: So directly, no, but we kind of make the assumption that it's built in there.
31:05
SPEAKER_01: Also, the other thing I would just add is the take into account what's going on in the country.
31:15
SPEAKER_01: Think of...
31:16
SPEAKER_01: How would you provide a objective measure of that sentence?
31:22
SPEAKER_01: Right.
31:22
SPEAKER_01: I mean, it's very it's like some people think things in the country are going great.
31:25
SPEAKER_01: Some people like it's what Madison and Geoffrey are in the business
31:28
SPEAKER_01: of doing is like,
31:29
SPEAKER_01: how can we systematize something that we can go back to in the generic ballot
31:33
SPEAKER_01: that Geoffrey is showing here is a way to get at that.
31:37
SPEAKER_01: Right.
31:37
SPEAKER_01: I mean, is it imperfect?
31:39
SPEAKER_01: Yes.
31:39
SPEAKER_01: But historically, it is suggested.
31:41
SPEAKER_01: I always say it kind of shows right.
31:42
SPEAKER_01: Skelley, is that fair?
31:43
SPEAKER_01: Yeah.
31:51
SPEAKER_00: Yeah, absolutely.
31:51
SPEAKER_00: I mean, I think that's the reasoning there.
31:53
SPEAKER_00: I mean, you can look at things like presidential approval
31:57
SPEAKER_00: and see President Trump's around 40 percent.
31:59
SPEAKER_00: Historically, that is a very bad place to be for a president in a midterm.
32:03
SPEAKER_00: But he's not on the ballot specifically.
32:05
SPEAKER_00: So the generic ballot at the end of the day is sort of our best indicator of the national mood.
32:13
SPEAKER_00: And again,
32:13
SPEAKER_00: knowing that in the last presidential election,
32:16
SPEAKER_00: the Republican,
32:18
SPEAKER_00: President Trump, I mean,
32:22
SPEAKER_00: you know,
32:24
SPEAKER_00: we'll see once we get closer to November and there are fewer undecideds
32:28
SPEAKER_00: in these polls and most pollsters are polling likely voters,
32:32
SPEAKER_00: you know, that the combined totals here will get closer to 100%.
32:36
SPEAKER_00: Right.
32:53
SPEAKER_00: From 2024.
32:54
SPEAKER_00: And obviously that would be a notable shift toward Democrats and give them
32:58
SPEAKER_00: an opportunity to,
33:00
SPEAKER_00: you know,
33:00
SPEAKER_00: make them favorites in the House and make it close to coin flip in the Senate,
33:03
SPEAKER_00: which is what our forecast has.
33:05
SPEAKER_00: And I think that that makes a lot of intuitive sense.
33:07
SPEAKER_00: And, you know,
33:07
SPEAKER_00: you mentioned earlier,
33:08
SPEAKER_00: so you're just saying about like sort of How the question here of,
33:13
SPEAKER_00: you know,
33:13
SPEAKER_00: how does it sort of incorporate or the polls incorporate sort of what's going on
33:17
SPEAKER_00: with the country?
33:17
SPEAKER_00: Yeah, I think.
33:18
SPEAKER_00: Yeah, absolutely.
33:19
SPEAKER_00: You've got if you kind of look,
33:22
SPEAKER_00: you know,
33:23
SPEAKER_00: I actually doesn't highlight it here on this particular way of showing it,
33:26
SPEAKER_00: but on the congressional on the generic ballot.
33:29
SPEAKER_00: Things actually got a little narrower in our average in terms
33:32
SPEAKER_00: the Democratic lead down to three and a half,
33:35
SPEAKER_00: four points actually in sort of early July,
33:39
SPEAKER_00: mid-July as I think an important thing was that gas prices went down a little bit.
33:44
SPEAKER_00: Yes.
33:45
SPEAKER_00: The Iran story moved to page two.
33:48
SPEAKER_00: Yes.
33:49
SPEAKER_00: And then it heated up again.
33:50
SPEAKER_00: And wouldn't you know it, the Democrats leading the January ballot shot back up.
33:53
SPEAKER_00: And I don't think that was a coincidence because at the same time,
33:55
SPEAKER_00: President Trump's approval bumped up just a couple of points and
33:59
SPEAKER_00: then it fell back down to 40 again.
34:00
SPEAKER_00: And I don't think there's a coincidence there because We know from polling
34:05
SPEAKER_00: that people are very worried about the state of the economy, the cost of living.
34:08
SPEAKER_00: And gas prices,
34:10
SPEAKER_00: of course,
34:10
SPEAKER_00: are the most obvious sticker shock of all because so many people are driving
34:16
SPEAKER_00: to work every day and sometimes decent drive and have to fill up pretty regularly.
34:23
SPEAKER_00: And so suddenly you see $4 gas in a lot of the country and more than $4 in a lot of the country.
34:29
SPEAKER_00: And that Is a real like easy sign that things don't seem as correct.
34:36
SPEAKER_00: And the story about Iran's conflict,
34:39
SPEAKER_00: the Iranian conflict in the Strait of Hormuz is constantly tied up in this.
34:45
SPEAKER_00: And that gets hot again.
34:46
SPEAKER_00: I think that's that's bad for the president.
34:47
SPEAKER_00: It's bad for his party.
34:49
SPEAKER_01: Yes.
34:49
SPEAKER_01: And I think to your point,
34:51
SPEAKER_01: Geoffrey,
34:52
SPEAKER_01: affirms that the generic ballot is that broad sentiment of how things are going,
34:58
SPEAKER_01: how people are feeling about things going in the country.
35:00
SPEAKER_01: Okay.
35:00
SPEAKER_01: I'm slightly hesitant to read this one because I never want
35:03
SPEAKER_01: to give like platforms we've never heard of and I don't know about like or
35:06
SPEAKER_01: that somebody works for that they're trying to promote.
35:09
SPEAKER_01: But...
35:10
SPEAKER_01: Chad Bailey says, I'm wondering if any of the models you use include the MetaCalculus platform.
35:14
SPEAKER_01: They are a prediction site and only award money for the most correct predictions.
35:19
SPEAKER_01: I have never heard of this website.
35:20
SPEAKER_01: I assume the answer is that you use Calci and Polymarket, but Madison, am I wrong about that?
35:26
SPEAKER_02: I also have never heard of this website.
35:28
SPEAKER_01: Okay.
35:29
SPEAKER_01: So we will look.
35:30
SPEAKER_01: Chad, if you work for MetaCalculus, good job.
35:32
SPEAKER_01: You probably should get a raise for PR purposes.
35:35
SPEAKER_01: But no, they do not use it currently as the answer.
35:38
SPEAKER_01: This is from Breakers.
35:39
SPEAKER_01: What makes you think John Ossoff basically has it in the bag?
35:42
SPEAKER_01: What has changed in Georgia?
35:43
SPEAKER_01: You want to pull that up, Geoffrey?
35:45
SPEAKER_01: I think it's like either a 79% or 80% probability on that.
35:49
SPEAKER_01: Look, I'll give the non-data science answer, and then I'll turn it over to them.
35:57
SPEAKER_01: Ossoff has raised a massive amount of money.
36:01
SPEAKER_01: He has been a very good and able candidate,
36:05
SPEAKER_01: which we kind of knew, but he has proven on the campaign trail.
36:08
SPEAKER_01: And Republicans nominated the worst possible option.
36:11
SPEAKER_01: Mike Collins is not only a member of Congress in a year where being a member
36:14
SPEAKER_01: of Congress isn't great as a resume item.
36:17
SPEAKER_01: He is also probably too conservative,
36:19
SPEAKER_01: particularly on social stuff,
36:20
SPEAKER_01: for Atlanta suburban women voters who are often a critical piece of a coalition.
36:26
SPEAKER_01: But do you guys want to talk about why?
36:28
SPEAKER_01: Because, and again, this does not mean John Ossoff is going to win 79% of the vote.
36:32
SPEAKER_01: This means in basically eight out of 10 of their simulations, John Ossoff finds a bunny.
36:37
SPEAKER_01: You guys want to talk about why this race in a swing state that if we,
36:41
SPEAKER_01: at the start of this election cycle,
36:42
SPEAKER_01: we would have said is,
36:43
SPEAKER_01: you know,
36:44
SPEAKER_01: he's the most endangered Democratic incumbent, why this looks like it does?
36:48
SPEAKER_02: Yeah, I can touch on that a little bit.
36:50
SPEAKER_02: So I think I think it really is what you said, though.
36:53
SPEAKER_02: He's he's a likable guy.
36:54
SPEAKER_02: And, you know, he's, you know, going for a second term.
36:57
SPEAKER_02: He's seemingly pretty popular.
36:59
SPEAKER_02: I don't I don't have his exact approval rating off the top of my head right now,
37:03
SPEAKER_02: but he's polling at about seven percent or seven point margin.
37:07
SPEAKER_02: Right.
37:08
SPEAKER_02: Which does him a lot of favors in this case.
37:12
SPEAKER_02: He has that seven point margin.
37:16
SPEAKER_02: He just looks a lot better as a candidate than Mike Collins has.
37:21
SPEAKER_02: He's a known face.
37:23
SPEAKER_02: He's not.
37:23
SPEAKER_02: You know,
37:24
SPEAKER_02: so I think and especially, too,
37:25
SPEAKER_02: I think that like Atlanta and that kind of region really,
37:28
SPEAKER_02: really it has been playing a very crucial part in Georgia politics
37:32
SPEAKER_02: and elections lately.
37:33
SPEAKER_02: And I think that that,
37:34
SPEAKER_02: you know,
37:35
SPEAKER_02: as you said,
37:35
SPEAKER_02: the the Atlanta region is not so not so keen on Mike Collins as they are on awesome.
37:40
SPEAKER_01: OK, but Madison,
37:41
SPEAKER_01: I want to follow up with something you said earlier, the model is candidate blind.
37:46
SPEAKER_01: More or less.
37:48
SPEAKER_01: So then what difference does it make whether John Ossoff is a good candidate or not?
37:55
SPEAKER_01: How is that measured?
37:57
SPEAKER_01: Because it seems to me like, I'll give a baseball analogy.
38:01
SPEAKER_01: It used to be when I was young, Which was a long time ago.
38:04
SPEAKER_01: That the way in which people were scouted,
38:07
SPEAKER_01: potential people were scouted,
38:08
SPEAKER_01: was some old dude who had been on the minor league circuit for 25 years would go
38:15
SPEAKER_01: and be like, ooh, that's Geoffrey Skelley.
38:16
SPEAKER_01: I like his look.
38:17
SPEAKER_01: He's a big kid.
38:18
SPEAKER_01: He throws the ball hard.
38:20
SPEAKER_01: Then came around the nerds,
38:22
SPEAKER_01: the sabermetric nerds, and they were like, none of this eye test matters.
38:27
SPEAKER_01: It is purely like, what is his F war?
38:30
SPEAKER_01: What is his G war?
38:30
SPEAKER_01: Like all of these stuff.
38:34
SPEAKER_01: I'm a believer in baseball of a little bit of both.
38:37
SPEAKER_01: But like when you say candidate blind, does that are we in the sabermetric end of politics?
38:44
SPEAKER_01: Is that what we're like?
38:45
SPEAKER_01: It doesn't really matter who the candidates are,
38:49
SPEAKER_01: whether John Ossoff has great hair, he does or not.
38:52
SPEAKER_01: Like, how do you how does it work?
38:54
SPEAKER_01: Explain that to people.
38:55
SPEAKER_02: Yeah.
38:55
SPEAKER_02: So and actually,
38:56
SPEAKER_02: I was just checking this right now because we do we do provide or depending on
39:00
SPEAKER_02: the candidate, you know, some some candidates overperform.
39:03
SPEAKER_02: I also was looking and the way that we measure...
39:21
SPEAKER_02: 2024 presidential results,
39:23
SPEAKER_02: if we were adjusting it and pretended that 2024 happened in a D plus 6 environment,
39:27
SPEAKER_02: Georgia would look a lot more bluer than it did in 2024, right?
39:31
SPEAKER_02: So I think that those two factors are probably the biggest parts of it.
39:35
SPEAKER_02: And then, you know, the model is candidate blind, but obviously polls and markets are not.
39:41
SPEAKER_02: And polls and markets,
39:42
SPEAKER_02: especially when they're maxed out and have a zillion dollars and one trillion polled
39:46
SPEAKER_02: in them, those are going to matter greatly.
39:48
SPEAKER_02: More than the fundamentals at the end of the day in terms of the population.
39:53
SPEAKER_01: Okay, I've got one from Grizzly96.
39:56
SPEAKER_01: With so many variables, how do you avoid quote-unquote overdetermination?
40:00
SPEAKER_01: For example, you can get a quadratic equation to go through any three points.
40:05
SPEAKER_01: So can you guys speak to that, Madison?
40:06
SPEAKER_01: I don't know if you want to speak to that,
40:08
SPEAKER_01: but how are you,
40:10
SPEAKER_01: to put it in layman's terms,
40:13
SPEAKER_01: how are you avoiding steering everything in
40:15
SPEAKER_01: the direction you want Maybe not you want the direction you think it's going to go.
40:21
SPEAKER_02: Yeah.
40:22
SPEAKER_02: So I actually would argue that our model compared to my understanding of others.
40:28
SPEAKER_02: Again, I might be I could be totally off base.
40:30
SPEAKER_02: Our model has far fewer variables than a lot of others.
40:33
SPEAKER_02: Like, for example, our model.
40:34
SPEAKER_00: It's pretty parsimonious.
40:35
SPEAKER_00: I think that's the word.
40:37
SPEAKER_00: Parsimony is the word you want.
40:38
SPEAKER_00: Yeah, yeah.
40:39
SPEAKER_00: Sure.
40:39
SPEAKER_02: Yes.
40:40
SPEAKER_02: Like our model,
40:41
SPEAKER_02: for example,
40:41
SPEAKER_02: doesn't really take into account like demographic data,
40:45
SPEAKER_02: which I know some of the other fancier ones do.
40:49
SPEAKER_02: And we kind of believe that they're not going to provide as much signal as stuff
40:57
SPEAKER_02: that we already have in there.
40:58
SPEAKER_02: And the signal they do provide is going to be highly correlated with stuff that we already have.
41:02
SPEAKER_02: So I think that that comes down to a lot of it.
41:03
SPEAKER_02: It's a very,
41:05
SPEAKER_02: I'm not going to say simplistic,
41:07
SPEAKER_02: but I'm not going to say it's definitely more complicated than
41:10
SPEAKER_02: the quadratic formula.
41:13
SPEAKER_02: There we go.
41:14
SPEAKER_02: Sorry.
41:15
SPEAKER_02: But your argument is you're not giving me that word.
41:17
SPEAKER_00: That's the word that you actually do use in
41:19
SPEAKER_02: statistical modeling to
41:20
SPEAKER_00: describe.
41:22
SPEAKER_02: Yes.
41:22
SPEAKER_02: It's parsimonious.
41:24
SPEAKER_01: And your point then is that,
41:25
SPEAKER_01: in fact, you guys are taking less inputs in than a lot of these things are.
41:31
SPEAKER_01: And so it's not just a zillion inputs and everything gets 5% equally.
41:38
SPEAKER_02: Right.
41:38
SPEAKER_02: Yeah.
41:38
SPEAKER_02: Yeah.
41:39
SPEAKER_02: Yeah.
41:39
SPEAKER_02: It's far fewer.
41:40
SPEAKER_02: Yeah.
41:43
SPEAKER_01: Cherry thoughts on Dan Osborne in Nebraska.
41:47
SPEAKER_01: So that's an interesting one for a model from a modeling perspective,
41:52
SPEAKER_01: because Dan Osborne is running as an independent in Nebraska.
41:58
SPEAKER_01: There is no Democratic nominee.
41:59
SPEAKER_01: The Democratic nominee is either drop Skelly dropped out or is about to drop out.
42:04
SPEAKER_01: It has agreed to drop out of the race.
42:07
SPEAKER_01: How do you work that?
42:09
SPEAKER_01: Because you don't have as much historical data on an independent Senate.
42:13
SPEAKER_01: I mean, I know he ran in 2024, so you've got that.
42:15
SPEAKER_01: But Madison,
42:16
SPEAKER_01: how do you—and by the way,
42:18
SPEAKER_01: for people who can see,
42:19
SPEAKER_01: it's 86% probability that Pete Ricketts, the Republican, wins.
42:22
SPEAKER_01: How do you work something like that situation?
42:25
SPEAKER_02: So— Right.
42:49
SPEAKER_02: So stuff like that where,
42:51
SPEAKER_02: you know,
42:51
SPEAKER_02: Pete Ricketts, like, he very much so deserves the Republican share of that value.
42:55
SPEAKER_02: And so we more so, you know, treat him, treat Osborne as the pseudo Democrat.
43:01
SPEAKER_02: And, you know,
43:02
SPEAKER_02: at the end of the day,
43:03
SPEAKER_02: kind of,
43:05
SPEAKER_02: we, you know, you can see 86%, I believe, yeah, this falls in the likely category.
43:09
SPEAKER_02: We don't really foresee this one flipping down.
43:12
SPEAKER_02: Blue or in Osborne's case, yellow.
43:15
SPEAKER_02: So, so long as it stays red,
43:17
SPEAKER_02: it is kind of the,
43:19
SPEAKER_01: it is interesting though,
43:21
SPEAKER_01: because Jill Stein is not a good analog for,
43:25
SPEAKER_01: Dan Osborne probably won't win,
43:26
SPEAKER_01: but he's going to do a hell of a lot better than Jill Stein did in Nebraska.
43:29
SPEAKER_01: Right.
43:29
SPEAKER_01: Go ahead, Jeffrey.
43:30
SPEAKER_01: Sorry.
43:31
SPEAKER_00: Yeah, no, no, no.
43:32
SPEAKER_00: I mean, one of the things is we don't actually have that much polling out in Nebraska.
43:36
SPEAKER_00: I was just like, when was the last poll we have?
43:38
SPEAKER_00: And it's from May.
43:41
SPEAKER_00: And so, you know,
43:43
SPEAKER_00: as Madison was saying,
43:44
SPEAKER_00: like if you were looking at like the forecast breakdown here,
43:47
SPEAKER_00: most of it right now is fundamentals and markets,
43:49
SPEAKER_00: very little from polling because we don't.
43:51
SPEAKER_00: And I think it would also...
44:14
SPEAKER_00: That would also help the Democrats' overall chances at the margins.
44:17
SPEAKER_00: That's right.
44:18
SPEAKER_00: Because the change
44:19
SPEAKER_01: there moves,
44:20
SPEAKER_01: obviously a change in one race,
44:22
SPEAKER_01: particularly a contested race, moves the overall Senate margin probability.
44:27
SPEAKER_00: Yeah, I mean, if this race moved significantly toward a more competitive,
44:32
SPEAKER_00: like 70-30 or something,
44:35
SPEAKER_00: you would see that at least somewhat reflected in the overall numbers.
44:39
SPEAKER_00: Got it.
44:41
SPEAKER_01: The overall Senate control number.
44:43
SPEAKER_00: Right, exactly.
44:45
SPEAKER_01: Now we touched on this at the top, but I think it's worth reiterating.
44:48
SPEAKER_01: First of all, I'm going to go a few more minutes here.
44:50
SPEAKER_01: I don't want to take too much time,
44:51
SPEAKER_01: but we are talking to Geoffrey Skelley,
44:54
SPEAKER_01: who is the Chief Elections Analyst at Decision Desk HQ,
44:56
SPEAKER_01: and Madison Murrell, who is the Data Science Fellow at Decision Desk HQ.
45:01
SPEAKER_01: Madison is sort of in charge of how what you are looking at,
45:05
SPEAKER_01: the DDHQ model for the election works.
45:08
SPEAKER_01: We did touch on this a little bit.
45:10
SPEAKER_01: It is worth, I think, revisiting.
45:12
SPEAKER_01: Lynn Gardner says,
45:13
SPEAKER_01: Geoffrey,
45:14
SPEAKER_01: how does this compare to what Larry Sabato does at the University of Virginia?
45:19
SPEAKER_01: Now, Geoffrey knows this very well.
45:22
SPEAKER_01: They are different individuals.
45:24
SPEAKER_01: Forecasting and handicapping are not the same thing.
45:27
SPEAKER_01: Geoffrey,
45:28
SPEAKER_01: we talked about it a little on the front on polling versus modeling,
45:30
SPEAKER_01: but you want to talk about the difference between what Sabato,
45:34
SPEAKER_01: Inside Elections,
45:35
SPEAKER_01: the Cook Political Report does, and what this Nate Silver are doing.
45:40
SPEAKER_00: Yeah, so I think one way to think about this is a...
45:47
SPEAKER_00: I'm trying to make this not too complicated.
45:49
SPEAKER_00: Essentially...
45:51
SPEAKER_00: We're dealing with the forecast model,
45:54
SPEAKER_00: sort of a spectrum of zero to a hundred, but with every number in between.
46:00
SPEAKER_00: The way that My old job at the Center for Politics.
46:06
SPEAKER_00: Which is Larry Sabato.
46:07
SPEAKER_00: Sabato's crystal ball at the EVA Center for Politics or the Cook Political Report
46:12
SPEAKER_00: or Inside Elections.
46:13
SPEAKER_00: The way they handle things is essentially they're using kind of a,
46:17
SPEAKER_00: I guess you would probably call it roughly a seven point scale
46:20
SPEAKER_00: where people One is solidly Democratic,
46:26
SPEAKER_00: seven is solidly Republican, and four is a toss-up, right?
46:31
SPEAKER_00: And then sort of gradations in between.
46:33
SPEAKER_00: And so that is very useful.
46:38
SPEAKER_00: And I certainly keep track of where ratings outlets have things
46:44
SPEAKER_00: because I – They are talking to the parties,
46:48
SPEAKER_00: they're talking to people,
46:50
SPEAKER_00: they're getting access to private polling that they're not sharing publicly
46:54
SPEAKER_00: unless they were allowed to.
46:56
SPEAKER_00: And so they're essentially getting a read of the race based on a combination
47:01
SPEAKER_00: of quantitative and qualitative information.
47:03
SPEAKER_00: And in the end, they're producing sort of a qualitative metric of lean democratic.
47:09
SPEAKER_00: Like you can apply a rough probability to like what that means, but it's not.
47:16
SPEAKER_00: You know, 68%, which is what our forecast model ends up doing.
47:22
SPEAKER_00: So you could say they're categorical and we're continuous.
47:25
SPEAKER_00: To use a mathematical thing,
47:28
SPEAKER_00: we're covering sort of 0 to 100,
47:30
SPEAKER_00: and they're telling you it's a 1, a 2, a 3, a 4, a 5, a 6, or a 7.
47:36
SPEAKER_00: And both have their uses.
47:38
SPEAKER_00: Obviously, the ratings one, I think, can be easier for people to interpret.
47:44
SPEAKER_00: And they're more familiar with it because they've been doing it for a longer time
47:46
SPEAKER_00: in terms of what's in the public eye.
47:48
SPEAKER_00: But what we're doing is producing a probability of an event happening,
47:55
SPEAKER_00: which I think can allow you to do some things that those ratings can't allow you
47:59
SPEAKER_00: to do in terms of understanding sort of the combination of outcomes.
48:04
SPEAKER_00: You know, if outcome A happens and outcome B happens, How does that affect the overall picture?
48:11
SPEAKER_00: You know, like a couple states, joint probabilities, like why?
48:15
SPEAKER_00: And in fact,
48:15
SPEAKER_00: maybe this is a good opportunity to bring up,
48:16
SPEAKER_00: you know,
48:17
SPEAKER_00: one of the things we can do with those probabilities,
48:19
SPEAKER_00: we can actually say more than just who's favored in what chamber and in
48:24
SPEAKER_00: which individual races,
48:25
SPEAKER_00: but we can also give you a probability that Democrats will win both the Senate and
48:30
SPEAKER_00: the House.
48:31
SPEAKER_00: Oh, about a 44% shot.
48:33
SPEAKER_00: Well, we also give Republicans about a 31% chance of winning both the Senate and
48:38
SPEAKER_00: the House,
48:38
SPEAKER_00: which means even though right now in our forecast,
48:41
SPEAKER_00: and this may be a little counterintuitive to people,
48:45
SPEAKER_00: Democrats are slight favorites in the House.
48:47
SPEAKER_00: Republicans are very slight favorites, very narrow favorites in the Senate.
48:51
SPEAKER_00: But actually the most likely outcome is that one party is going to win both.
48:54
SPEAKER_00: And that's because the outcomes are correlated.
48:57
SPEAKER_00: If the Democrats have a slightly better night, they're going to be of the universe of outcomes.
49:04
SPEAKER_00: And they win the Senate.
49:05
SPEAKER_00: They're definitely winning the House.
49:06
SPEAKER_00: Right.
49:10
SPEAKER_00: They're absolutely winning the Senate.
49:12
SPEAKER_01: These things don't exist in a vacuum.
49:14
SPEAKER_01: Exactly.
49:15
SPEAKER_00: The House races don't exist in a
49:17
SPEAKER_01: different and separate political environment.
49:19
SPEAKER_01: They're not independent
49:21
SPEAKER_00: of each other.
49:22
SPEAKER_00: And Madison,
49:24
SPEAKER_01: I think I know the answer to this,
49:25
SPEAKER_01: but you guys do not include a Cook Report's ratings or an inside elections race,
49:31
SPEAKER_01: anything like that in the model.
49:32
SPEAKER_02: No, no, we do not.
49:33
SPEAKER_02: Got it.
49:34
SPEAKER_00: And some forecasters do,
49:35
SPEAKER_00: you know, Nate,
49:36
SPEAKER_00: I know Nate,
49:37
SPEAKER_00: because I used to work with Nate at 538,
49:38
SPEAKER_00: like there's the deluxe version of,
49:41
SPEAKER_00: and they incorporate these ratings in that version.
49:46
SPEAKER_00: Now he doesn't use prediction markets.
49:48
SPEAKER_00: I actually have been wondering if...
49:49
SPEAKER_00: If there's a wisdom of the crowds argument about prediction markets,
49:53
SPEAKER_00: at least potentially,
49:56
SPEAKER_00: where if you're using expert ratings as sort of a conventional wisdom measure
50:02
SPEAKER_00: a sort, Yeah.
50:35
SPEAKER_01: Well, and I like Madison's point about the law of large numbers,
50:37
SPEAKER_01: which is like,
50:38
SPEAKER_01: you know,
50:38
SPEAKER_01: if you have, if you, a lot of these markets have 30, 40, $50 million in them.
50:43
SPEAKER_01: I mean, it is not like one dude betting on Jonte Porter to get one and a half rebounds.
50:50
SPEAKER_01: You know what I mean?
50:50
SPEAKER_01: Like it's, they are big and
50:53
SPEAKER_00: there is some level.
50:55
SPEAKER_00: Oh no, he didn't grab his second rebound.
50:56
SPEAKER_00: I lost my bet.
50:57
SPEAKER_00: No, exactly.
50:59
SPEAKER_01: Okay.
50:59
SPEAKER_01: We are going to do these probably, I always wonder about this.
51:03
SPEAKER_01: Bi-weekly is every two weeks, right?
51:06
SPEAKER_01: Yeah, yeah.
51:07
SPEAKER_01: Simi-weekly would be twice a week.
51:08
SPEAKER_01: I was going to say,
51:09
SPEAKER_01: I don't want to commit to twice a week,
51:10
SPEAKER_01: but bi-weekly, Madison, Jeffrey, and I will get on.
51:14
SPEAKER_01: So if you like this, we will be on after Labor Day talking about their model.
51:17
SPEAKER_01: You can go to votes.decisiondeskhq.com.
51:23
SPEAKER_01: You will see exactly what you were seeing on Geoffrey's screen at the top there.
51:26
SPEAKER_01: Votes, home vault, polling markets.
51:28
SPEAKER_01: Click on forecast if you want to see what we're talking about and you can see this stuff.
51:33
SPEAKER_01: Last and important question.
51:36
SPEAKER_01: What did you both major in in college?
51:39
SPEAKER_01: How does one become what you are?
51:43
SPEAKER_01: Skelly, you first.
51:44
SPEAKER_00: Well, I was actually a history major as an undergrad,
51:46
SPEAKER_00: but I do have a master's in political science.
51:50
SPEAKER_00: And obviously a lot of political science is based
51:58
SPEAKER_00: the quantitative
51:59
SPEAKER_00: and qualitative side,
52:00
SPEAKER_00: is looking back at things and trying to use past cases,
52:05
SPEAKER_00: even if they're not perfectly analogous,
52:07
SPEAKER_00: to understand the present,
52:08
SPEAKER_00: at least to some extent, which is one of my favorite things to do.
52:12
SPEAKER_00: So no wonder I was a history major, I guess.
52:15
SPEAKER_00: That was my background.
52:17
SPEAKER_00: And then I started working at the Center for Politics after I got out of grad school.
52:23
SPEAKER_00: And, well, I've been doing this ever since.
52:26
SPEAKER_01: And Madison.
52:27
SPEAKER_02: Yeah.
52:28
SPEAKER_02: So this is also complicated by the fact that I'm actively getting my master's.
52:31
SPEAKER_02: But my undergrad, I'll give you the full spiel.
52:34
SPEAKER_02: It's a very long title.
52:36
SPEAKER_02: I majored in political science with a concentration in methods and models and
52:41
SPEAKER_02: a minor in social data science.
52:42
SPEAKER_02: And I'm currently getting my master's in data science for public policy.
52:47
SPEAKER_01: Okay.
52:48
SPEAKER_02: That makes me feel slightly better because
52:50
SPEAKER_01: I thought you were going to be like,
52:51
SPEAKER_01: I majored in math.
52:52
SPEAKER_01: And I was always like, who the hell majors in math?
52:55
SPEAKER_01: Like the really...
52:57
SPEAKER_02: Yeah, I really wanted to major in math as a kid.
53:00
SPEAKER_02: And then I took AP Calculus and decided that it was not a path for me.
53:04
SPEAKER_02: So I commend those who do.
53:06
SPEAKER_02: We have a lot of more mathy nerds than one of the other people that very heavily...
53:12
SPEAKER_02: Actually contributed to this model more than I did, has a PhD in physics.
53:16
SPEAKER_02: That is a level of math that I will never, ever in my life understand.
53:19
SPEAKER_02: I'm so happy that he does.
53:21
SPEAKER_01: My high school counselor recommended against AP Calculus,
53:24
SPEAKER_01: which I think in his defense was smart for me.
53:27
SPEAKER_01: I majored in English,
53:28
SPEAKER_01: which is why I ask the questions and they don't let me even in the room
53:33
SPEAKER_01: where they are calculating the model.
53:35
SPEAKER_01: It's for the best.
53:36
SPEAKER_01: For the best.
53:38
SPEAKER_01: Madison Murrell, Jeffrey Skelley, again, go to votes.decisiondeskhq.com.
53:46
SPEAKER_01: You can see the forecast.
53:47
SPEAKER_01: You can see vote count.
53:49
SPEAKER_01: It's a wonderful place.
53:51
SPEAKER_01: We are, I'm a giant fan.
53:53
SPEAKER_01: So like, But it's a great website if you're interested in this sort of stuff.
53:58
SPEAKER_01: You can look at the model.
53:59
SPEAKER_01: You can find out about the methodology of the model.
54:01
SPEAKER_01: You can look at each race.
54:03
SPEAKER_01: How much is fundamentals?
54:04
SPEAKER_01: How much is polling?
54:05
SPEAKER_01: How much are the prediction markets?
54:08
SPEAKER_01: We're going to do this every couple of weeks.
54:10
SPEAKER_01: I find it fascinating.
54:11
SPEAKER_01: I think it is a critical piece of understanding sort of where we are and
54:14
SPEAKER_01: where we're going, most importantly.
54:16
SPEAKER_01: Madison, Jeffrey, thank you.
54:17
SPEAKER_01: We'll see you in a couple of weeks.
54:20
SPEAKER_00: Thanks, Chris.
54:21
SPEAKER_01: Bye, y'all.
54:22
SPEAKER_01: Thanks, everybody.
54:23
SPEAKER_02: Take care.
54:24
SPEAKER_00: And thank you, everyone.
54:25
SPEAKER_01: Yep.