Conversation
![David Watson 🥑](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="96" height="96"><rect fill-opacity="0"/></svg>)
Post your reply
The year is 20XX. The stock market is no more. Betting markets have oracular power. Hedge funds are teams of rogue pollster degens. Alpha is secret, covetously guarded social science methodology. Methodological consensus advances via espionage conducted by Shadowrunners
![How the Trump Whale Correctly Called the Election](data:image/svg+xml,<svg xmlns="http://www.w3.org/2000/svg" width="1200" height="600"><rect fill-opacity="0"/></svg>)
Huh; I heard about gains from the neighbor poll method at least a decade ago. Why haven't pollsters adopted it?
Not sure. Some repliers are saying that some have, others are saying neighbor method overshot Trump support significantly.
Even if not goofy per se could still be a sort of gettier case, so yeah
But even those cases it’s still interesting that he commissioned his own methodologically opinionated polling data and bet accordingly
Is "mind blowing to the favor of Trump" more correct than what we got from the polling averages?
This isn't new. This was done in 2020 as well. I suspect it had been more quietly employed even earlier. Things like this is why Rasmussen, etc; had much better polling than the established firms. They think outside the box.
Polymarket was overall an incredibly accurate signal to this election’s outcome
This is really impressive (unless he's Joseph Smithing the polls to make himself seem smarter)
I want to see what states specifically these polls predicted, to know it wasn't a random guess. This outcome of all 7 swing states going towards Trump but other states not flipping was the single likeliest outcome from Nate Silver's model. Did Theo really predict or is it luck?
as a solution to pollsters fundamentally requiring responses, and people being unwilling to respond that's.... actually really creative.
Not how I expected that sentence to end
Or you could just add a couple points to Trump's polling, to account for this rather obvious issue that seems to always affect his numbers.
this guy should be given us citizenship. this is the most american thing i can think of
Love it!
This is also the wisdom of crowds method that we applied for predicting Brexit, Trump, and Biden back in the day. Pollsters called it luck.
Well, currently it's used to predict weekly market swings. With the same striking precision
Fair to assume the neighbor method only works when those polled don't know of its existence?
Or in, other words, a more informed population about polling methods will decrease the efficacy of said method.
this sounds very similar to the Oraclum model for the 2016 election
Real skin in the game.
Also, this 
Quote
Dominic Cummings
@Dominic2306
In 2016 Ben Warner's model for Vote Leave correctly predicted our narrow win.
In 2019 Ben Warner's model predicted 364 Tory MPs (just one off - most accurate in UK).
In April 2023 (18 months ago!!) he did it again - our model of DT v Kamala: 311-227. (One of the things he
Show more
Nothing to do with high quality signals, just variance and luck. At most an expression of views
I worked with this guy for six months as a junior quant and learned more than during my PhD.
Political betting community is going to start hiring people's neighbors to get information. Sharps stay sharp!
Reminds me of an old interview question. "What would your *colleagues* say about your skills in _____ if we asked them?"
seems to give more signal than "Tell us about your ability in _____ skill"
Fascinating to see how contrarian signals can emerge from betting markets, Matt. It's a great reminder that sometimes the most valuable insights come from thinking outside the box. In my own practice, I've found that taking a step back and observing my thoughts, rather than
Show more
You should post the lottery winners who used the "prayed to god" method. The number one lesson you learn when analyzing gambling decisions is don't be results oriented, don't assume a successful gamble indicates an intelligent gambling decision, or an unsuccessful gamble
Show more
I bet this has been done by some pollsters anyway? Sounds like a well-known psychological bias to filter out in stats.
Well, would you look at me again:
Quote
Sean Ross Callaghan
@seanrcallaghan
Replying to @seanrcallaghan and @HarryDCrane
"Who do you think is going to win" polls should be explored more as win predictors. They could be weighted and binned just like state polls in forecast models, and the question synthesizes the collective knowledge of the citizenry.
Quote
Sas
@Mr_Sas0
Prediction markets have outperformed traditional polls in accuracy and insight. But instead of embracing innovation, (as usual) Europe is busy banning platforms like Polymarket. 
EU: We want Europe to be a leader in innovation!
Also EU: But only if it involves paperwork and
Show more
EU: We want Europe to be a leader in innovation!
Also EU: But only if it involves paperwork and
There were several polls which asked both "who are you voting for" and "who do you think will win the election" and the discrepancies may have notable there as well.
Maybe.
It’s funny because a few days ago I saw shitlibs making fun of him for using the neighbor method.
betting your life savings to prove that your neighbor dave from accounting is actually nostradamus
This is a well known method, and although it worked in this case, it has failed before. It all depends on what the underlying cause of systematic bias in the polls is. That varies from election to election.
Yes, and it was a binary event, so I don’t think we should read too much into it. We could be also writing the other side of this. Neighbors can capture more the environment of their area and less their own intent
q.v. Andreas Murr of Warwick University on this - crowd wisdom polling has been done before with some success. Takes out egocentric bias and draws on local knowledge.
Since the popular vote in the US presidential election appears to be pretty much 50-50, I wouldn't put too much trust on this specific method.
There is no mechanism in his polling strategy that implements the distortion of the first-past-the-post electoral college system.
surprised this is somehow alpha is this day/age - almost surprised to the point that it doesn't seem true.
if it were true, why would he reveal his methods? What is the gain in giving away the edge?
I think the real issue is that these polls just make shit up and some don’t even bother polling pipl at all. Think about it: if you could get the same results from not working as you could from actually working which would you choose?
Strikes me as a kind of thing that could work for a cycle or two before people catch on
I do the same thing with thinking about buying and selling stocks, asking myself what others are likely to do.
Stock markets just provide measurable signals to allocated capital and effort towards the most needed places. Could well be replaced by advanced betting markets.
Sadly, there will be no pollsters left in business by the end of this month.
If this is true, this actually does make the case that prediction markets add value.
Wow. Market researchers will be happy to know how the age old 'projective techniques' are rediscovered for new applications. Great example. When individuals are given a mask, they will reveal much more about themselves.
The Neighbour question has a very mixed history. If I remember correctly it predicted a 60% Trump win in Colorado in 2016.
I heard about a cognate of the “neighbor method” more than 20 years ago; how is this not being used?
The actual results were all within the expected margin of error of polling aggregates in all swing states.
These neighbor polls probably were not even as accurate as the aggregates. Bro just got lucky.
This is awesome. Such a cool technique, being also employed to study classroom cliques: kids are asked to categorize their peers into predefined categories. Kids have difficulty placing themselves but are very accurate in placing their peers.
So not only he was right, but he was right for the right reasons?
Interesting but it's still a poll, just with an innovative way of asking questions (whcih btw is not that new). Polymarket on the other hand is a fun game but makes no sense as a tool to predict elections.
Elections are way too infrequent to be worth developing good models for from a gambling perspective, you need dozens of events to make an edge tell reliably
Ngl I was astounded when I saw paddypower had mega odds for trump to win like days before the election but maybe gamblers and gambling companies are the real geniuses
So he made a profit by promoting a better way to poll. Another win for capitalism.
eh imo his comments on polymarket and interviews demonstrate he really doesn't understand polling or election modeling, he just won a 50/50 coinflip
Problem is: which specific neighbor? Doesn't the choice carry some additional bias?
We do this Product Market Fit surveys, asking: What type of person do you think would benefit most from the product? What we're really asking is: what type of person are you? people often feel more comfortable describing an ideal user rather than defining themselves directly
This was done in 2016, precisely because of Trump. If they are just doing this now in 2024, they are almost a decade late
If Theo is just a Frenchman with just profit mindset and no alterior motive to create vibes… (aka Elon Musk)… then why did he create 4 different accounts to house his trades? Why not just all from one?
I would of told a pollster that my neighbors would vote for Trump not who I voted for but I guess that still gives the right answer.
I first heard of "the neighbor method" from pollsters Richard Baris and Robert Cahaly. The fact I was tuned into such people helped me predict the EC win # for Trump. Not all the pollsters suck.
Quote
Travis Kyme
@TravisKyme
Replying to @pastorcoin and @BitcoinIsaiah
True because America will become more like Trudeau's Canada in her term. I think Trump wins by 312 though sweeping the battleground states
This is idiotic and the dude got lucky. People know who they intend to vote for, but they are guessing for many of their neighbors. I don’t buy for a second that this would be a consistently more accurate polling method.