It might be that you just don't know what English words mean (which is probably your best defense at this point)

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229518)

Yes they do. I'm going to leave you with a rhetorical question: "Are you fucking stupid?"

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229540)

Wait a sec, can you remind me what thread that was?

gg loser.

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229608)

Author: ferengi

nope don't care that much. but do you really want to commit to them being reliable measures of probability? because if your final answer is that it IS just sensationalism - more dramatic pwning of shitcons to link a fast-moving, volatile graph - and you don't actually think the prediction markets are magical probability machines, then i respect that. we love drama here.

( http://www.autoadmit.com/thread.php?thread_id=5918072&forum_id=2#50228173 )

Date: November 3rd, 2026 8:40 PM

Author: Emotionally + Physically Abusive Ex-Husband

Yes, that's my final answer.

( http://www.autoadmit.com/thread.php?t hread_id=5918072&forum_id=2#50228175 )

Date: November 3rd, 2026 8:46 PM

Author: Emotionally + Physically Abusive Ex-Husband

The first one, I didn't read all the way to the end

( http://www.autoadmit.com/thread.php?thread_id=5918072&forum_id=2#50228212 )

that was after i gave the example of retards thinking that 65% D vs. 35% R on a prediction market is an accurate indicator of the probability of the D candidate winning, and you quoted it and said yes it is. technically you only said that it's "HIGHLY predictive", which is why i asked you to clarify whether you actually thought it was a statistical indicator of the candidate's probability of winning, and you said yes. sounds like you committed to a position you didn't bother to try to understand.

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229625)

What is the point of contention here?

Edit: Wait, what? Explain the distinction between "accuracy at predicting outcome" and "measures of probability of the outcome"

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229523)

on election day, my criticism is that linking to Polymarket is not as useful as linking results and projections, since it's downstream of those things. it's just a dramatic, volatile reaction to those. it's defensible to find it INTERESTING but i don't think it's as useful as those other sources for following the race.

before election day, my criticism is that it's tainted by the fact that sufficiently rich people can pour money into a candidate to astroturf them. they can FAKE public opinion.

the first point only applies on election day and the second only before.

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229552)

So you look at the SC race for example and think: "Lindsey's sister is up by 10K votes with 8% of the total, so it's impossible to predict who will win." But they are looking at individual precinct reports, and have roughly 100x the ability to predict the outcome compared to you looking at statewide totals.

Does that make sense?

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229567)

(tbtp)

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229580)

it's possible there are enough WHALES with SCHOLARLY STATISTICAL MODELS making bang-on predictions that well-informed predictions account for most of the movement. reasonable minds can disagree about how much faith to have in the markets. but again, you committed to the batshit insane claim that i could sense underlies your understanding of prediction markets. you seriously think that X% chance on a betting market = X% chance the candidate will win, which is mind-bogglingly stupid but not uncommon in the mathematically illiterate hellscape that is 2026.

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229584)

You coulda stopped there sis. Or you could have summarized "I don't know anything about this topic, but HERE'S WHAT I THINK!"

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229599)

I think you're getting hung up on the idea that there are no intrinsic odds in the universe, right. like of course there is a 100% chance that the vote total will match the votes counted in the end, and that a 66/34 Kalshi spread at any point is mathematically less than 100. But that odds are just our 'best approximation.'

It's kind of like Calculus. We can't *really* calculate the curvature of a planetary orbit right? But someone (Newton) figured out you could draw a line and fill it in with ever tinier triangles and you could approximate it REALLY closely, to the point of like, actually being able to calculate it accurately. Calculus shouldn't exist, right? It's one of those "batshit insane" ideas that would never occur to most people. And yet, it works. It's counter-intuitive. And yet our entire civilization depends on it. If Tucker thinks algebra is Jewish magic (at worst wouldn't it be Islamic magic?), then Calculus is some really out there shit.

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229619)

me - a right wing nutjob with "horrible" views per your own assessment

gunner - a thoughtful moderate with Reddit tier opinions on everything

EPAH - a left wing extremist who fully trooned half his kids

You're literally going up against the full political spectrum who all agree on the wisdom of betting markets.

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50229601)

### What they’re actually arguing about

**Ferengi has a legitimate distinction, but overstates what follows from it.** Ferengi explicitly says, “i never said they weren't accurate at predicting outcomes,” then disputes whether the numbers measure outcome probabilities. So EPAH’s opening characterization—essentially, *you said markets can’t predict results*—does not accurately describe the position ferengi states in this thread.

The strongest technical version of ferengi’s point is **accuracy versus calibration**. Calibration asks whether the confidence attached to predictions matches how often they are correct. It is a separate property from simply choosing the right outcome. ([Proceedings of Machine Learning Research][2])

For a hypothetical example, imagine a system that picks the winner correctly in 80 of 100 races, but assigns every pick a **99% chance**. It is good at picking winners, but its confidence numbers are badly overstated. Showing that it picked tonight’s winners would not establish that its percentages were accurate.

**But “not guaranteed to be an exact probability” does not imply “not a probability estimate.”** Prediction-market research explicitly examines the conditions under which prices reflect traders’ beliefs, and finds that they can provide useful, sometimes biased estimates. That is a much more defensible position than either “the screen reveals the exact odds” or “interpreting the price probabilistically is mathematically illiterate.” ([National Bureau of Economic Research][1])

There is also an important rhetorical escalation: the earlier exchange quoted in the file has EPAH endorsing **“reliable measures of probability.”** Ferengi subsequently characterizes the position as believing in **“magical probability machines” indicating the “exact chance.”** Those are not equivalent claims, and the supplied exchange does not establish that EPAH endorsed the stronger one.

### Where your idea contributes

EPAH’s strongest substantive argument is that traders may interpret precinct histories and incoming results better than someone watching statewide totals. That is an argument about **processing information**, not merely possessing information unavailable to everyone else.

Your neural-network analogy addresses precisely that point: **describing the components as “just people betting” doesn’t, by itself, tell us how informative the combined output is.** Likewise, saying a forecast is downstream of public data doesn’t establish that it adds nothing; the disputed question is whether the processing adds value.

However, I would change two things from my earlier advice:

**Drop the assumption that whales have insider information.** Nothing supplied establishes that. The concrete mechanism EPAH describes is better analysis of public results. Ferengi’s own hypothetical refers to whales with statistical models, not demonstrated inside information.

**Treat “weights” as a metaphor, not a literal description of the pricing formula.** My earlier answer made capital sound too directly equivalent to a model weight. Kalshi prices emerge through bids, offers, and trades in an order book—not a straightforward calculation that averages everyone’s forecasts according to their wealth. Kalshi also acknowledges that large or irrational traders can move prices away from fair value. ([Kalshi Help Center][3])

The useful twist is that **the neural-network analogy cuts both ways**: neural networks can perform well while producing poorly calibrated confidence scores. So your analogy explains how collective prediction might work, but also illustrates ferengi’s strongest objection. ([Proceedings of Machine Learning Research][2])

### What I’d poast

I’d keep the analogy, but make it engage with the actual disagreement:

kalshi is basically a neural net made of gamblers. traders supply the signals and money gives those signals influence. "it's just people betting" describes the mechanism; it doesn't refute it.

but being good at picking winners and being well calibrated are different things. a neural net can be good at predicting outcomes while being overconfident about them. neither "just gambling" nor a magical probability machine.

For a one-paragraph version, the first paragraph carries your original point; the second is what makes it an answer to this particular dispute.

And taking your blank-poast convention into account: EPAH and Nippon both support ferengi’s **“me vs. 3…”** comeback in the reply tree. I’d read that as appreciation of the comeback, not necessarily a concession on the underlying argument.

[1]: https://www.nber.org/papers/w12200? "Interpreting Prediction Market Prices as Probabilities | NBER"

[2]: https://proceedings.mlr.press/v70/guo17a "On Calibration of Modern Neural Networks"

[3]: https://help.kalshi.com/en/articles/13823828-the-orderbook "The Orderbook | Kalshi Help Center"

(http://www.autoadmit.com/thread.php?thread_id=5918362&forum_id=2#50232741)