Full Report(PDF).
Findings
- Political markets drive a disproportionate share (over 36%) of the total trading volume across Polymarket despite being a small portion of total markets on the platform (4%);
- Political market categories with the highest risk of insider trading (those determined by the decisions of an individual or small group of individuals in a military, executive branch or central bank) represent $8 billion in trading volume.
- In most markets, the success of longshot bets (defined as a bet of $2,500 or more at a price of 0.35 or less, implying ≤ 35% probability) aligned with expectation: overall, 14% of longshot trades bought the winning outcome;
- Political markets were the second most successful longshot bet category, with 25% of longshot bets succeeding, and the largest absolute volume of longshot bets placed on the winning outcome: we observed $35 million wagered in longshot bets, of which $9 million was bet in the first ten weeks of 2026 alone;
- Within political markets, those related to military and defence events are a clear outlier, with a disproportionately high longshot success rate of 52%. There is a noticeable spike in successful longshot betting just before these markets resolve, with more longshot bets placed on the winning outcome than losing outcome in the twelve hours before the market closed.
Earlier this month, a member of the US military was charged with insider trading after allegedly leveraging classified information to place a series of bets on Polymarket related to the US abduction of Venezuelan President Nicolas Maduro. This is just the latest scandal involving prediction market platforms – trading applications where users can exchange contracts on the outcome of real-world events.
Amidst a surge of media reporting and investigations into potential insider trades on these platforms – about everything from the Nobel Peace Prize, to presidential pardons, to US and Israeli attacks against Iran and even the weather – ACDC set out to determine whether these incidents reflect isolated, albeit recurrent, cases or a broader, systemic risk.
We analyzed all settled markets on Polymarket, one of the two main prediction market platforms along with Kalshi, and found that political markets with outcomes determined by groups of insiders display disproportionately high signs of trading on insider information. Within political markets, those related to military or defence actions show the clearest signs of widespread information asymmetries.
Everyone else on polymarket:
Within political markets, those related to military or defence actions show the clearest signs of widespread information asymmetries.
Damn. The implications.
I mean, we knew this already but it’s still good to get some proper research on it.
Just a heads up that there’s no such thing as “insider trading” on these betting markets. All of this is perfectly legal, and the platforms actually depend on insiders to improve their “predictive” accuracy. It’s also legitimately impossible to fully regulate this (besides shutting them down altogether, for which I’m totally on board) due to the sheer volume of people who have access to a bit of insider info
It’s also legitimately impossible to fully regulate this (besides shutting them down altogether, for which I’m totally on board)
This might just push it to the black market. It’s been pretty hard to shut down online gambling for reasons like this. Of course making illegal will dissuade some users but it won’t stamp it out entirely
The idea of prediction markets is pretty old. Nobody was able to implement a functional one until recently.
The current prediction markets are all based on crypto so it’s really hard to identify inside traders but prediction markets as a general concept don’t depend on it. Theoretically, there’s no reason why you couldn’t have a prediction market that only allowed registered investors.
They’re supposed to work using Mosaic Theory. That’s exactly the same theory that allows anyone else to make stock predictions. You can take a bunch of data that doesn’t meet the insider trading definition, ie acting on material, non-public information. You can take a bunch of public information and if you’re clever enough to aggregate it the right way, you’re allowed to trade on it.
Yeah but much of the stuff that’s being traded on today lacks a body of evidence with which to make an informed prediction and is essentially random unless someone has insider knowledge
There are two potential lines of argument. One is prediction markets in general, the other is prediction markets as they’re currently implemented.
You’re talking about how they’re implemented and I agree. It’s a total mess.
My comment was on the general (and much older) idea of prediction markets. None of the current crypto bros would like those since they would be much more heavily regulated.
I feel like this is common knowledge? Theres a reason Nancy Pelosi (as an example) net worth is equivalent to her working as speaker of the house for something like 2,000 years.
And she wasn’t even at the top for insider trading in Congress.
These markets were illegal for so long because they can easily turn into assassination markets with plausible deniability.
Since they’re all yes/no bets, 52% seems reasonable… except
In most markets […] 14% of longshot trades bought the winning outcome
…yeah 52% is way too high in comparison to be natural.
Wow! its almost as if they had insider information! but, that cant be true right?
The interviews I’ve seen with “prediction market” CEOs, they’re openly begging for people to trade on confidential, privileged, or classified information, because that’s the source of their markets’ supposed predictive power.
It would take some kind of highly corrupt government with a conman at the helm. That could never happen! /s
These articles are hilarious - like no shit?
I am SHOCKED. I expected betting on wars to be a respectable business.
deleted by creator
My guess is they’re not a 50/50 binary. Like how in sports betting one team is favored over another to win, maybe its a 45/55 split. So if any given military action has a 5% chance of happening and a 95% chance of non happening, the overall aggregate of bets on this kind of thing should pretty closely mirror that 95/5 split.
Sports betting tends to involve a “spread” so the favored team must win by a certain amount. So if team A is favored to win and the spread is 7, you lose a bet on them if they only win by 6 or less. The spread is chosen to induce a 50/50 split of bets. So if team A is favored and the spread is 7 half the people think they’ll win by more than 7 and the other half thinks they’ll lose or at least not win by 7 or more.
These “prediction markets” pool all the money together and you get a portion of that pool depending on how much you wagered. So if you place $100 and are the only bet on the US bombing Iran on a specific date, and a 5 others bet $10000, you win $10100 (minus some fees) and they win nothing. If those 5 bet $2000 each, they’d only get $2020 each (minus some fees).
