According to PolyBeats, on August 19th Beijing time, an individual achieved a 120% return within a mere 89 seconds by predicting "whether Russia could occupy a certain area on the battlefield".

This person spent $4,000 buying "Russia will seize Konstantynivka this month" on a platform called a prediction market. The probability of this event occurring was only 43% at the time. 89 seconds later, this probability rose to 92%.

On this platform where probability is price, if the event ultimately occurs, the probability will be fixed at 100%, and the position's value will rise accordingly. This mysterious figure completed placing the order, experiencing a surge, and taking profits off the table in under two minutes: in an era where various trading models are emerging endlessly, such a lucky trade is not uncommon.

The suspicious thing about this person is that Konstantynivka was his 14th such operation.

14 Predictions on Military Developments All Correct

The account's transaction records, as concluded by PolyBeats statistics, indicate that this person's account made 14 prediction trades related to the Russia-Ukraine war, all of which were correct, with a total realized profit of approximately $17,500.

They did not cover the same city, nor the same timeframe: from Maliivka, Pokrovka, to Huliaipole and Konstantynivka; some were "will Russia seize" and others were "will Ukraine re-enter". But the operations were almost like copies from the same template.

Even more bizarrely, this account not only got all 14 predictions right, but for 11 of these trades, the probability surged to near 100% within 5 minutes after the buy order.

Guessing the development of the war situation correctly 14 times in a row and buying before the market probability soars is difficult. But if what one knows ahead of time is not the battlefield situation, but the map update time, things become much simpler.

Prediction Market Settlement Source: A Breeding Ground for Insiders

Almost all Russia-Ukraine occupation market titles on Polymarket follow one template:

"Will Russia seize [Region Name] by [Date]?"

While the question seems straightforward, what actually determines the answer is not when the gunfire starts.

Polymarket rules point out that only when the Russia-Ukraine battlefield map produced by ISW — the Institute for the Study of War — marks the specific area as under Russian control, and meets the duration requirement specified in the rules, will the market settle to "Yes" — the corresponding probability will settle at 100%.

In other words, even a soldier standing in an occupied area cannot act as an insider to profit stably and long-term in such a market: even if the occupation actually occurs, as long as ISW, the settlement source, does not update the map, this market will also settle to "No".

In such a market, only one kind of person can achieve a perfect profit record like the account mentioned above: they don't need to know the frontline situation, understand Russia-Ukraine geopolitics and military strategy, or even follow any international current affairs news. They only need to know in advance which area ISW is about to mark as under Russian control, and when that update will take effect publicly.

This group of people is the technical staff responsible for updating the war status at ISW. They only need to go to the corresponding market on the prediction market to build a long position moments before clicking the mouse to confirm updating the ISW map, and they can complete a handsome profit within a minute.

Sounds like a conspiracy theory? Last year, ISW was already embroiled in a similar farce.

Employee Alters Data on Open-Source Intelligence Platform, Earns Hundreds of Times Return

In November 2025, prediction markets had already demonstrated in advance how this mechanism could spiral out of control.

At that time, the market question was "Will Russia seize Myrnohrad by November 15th?". The rules were almost identical to those described earlier — as long as the ISW map showed the Russian military controlled any part of the specified street intersection, the result could be judged as "Yes".

The biggest difference from the current rules was that the rules at the time did not have a "waiting period": as soon as the map changed, the market's settlement conditions could be met.

And on the evening of the 15th, with only a few hours left until the deadline, the probability for "Yes" was already less than 3% — with no geopolitical news reporting the occupation of Myrnohrad, everyone had already assumed this market would ultimately settle to "No".

However, at that moment, the ISW map suddenly showed the area as occupied; before most people could even react, the probability for "Yes" was instantly pushed to 100%.

In the following days, just as traders questioned that this map marking had no substantive evidence, ISW publicly admitted that between the night of November 15th and 16th, an "unauthorized, unapproved" edit occurred on the interactive map; this edit was removed before the normal workflow began on the 16th and did not represent its official battlefield assessment.

The map was withdrawn, but the market was not. Accounts that predicted "Yes" when the probability was approaching zero made a fortune through this "unauthorized edit". Among them, 0x69A9 achieved an excess return of 10,888.68% by buying when the probability was only 0.9%.

After the Myrnohrad incident, ISW stated that its map is continuously revised on weekdays and should not be understood as real-time battlefield conditions; it later also added status prompts like "Editing" and "Finalized". Related prediction markets later added stricter persistence requirements: the map status needs to persist across the next full update cycle to become a settlement basis.

As of now, this 14-time winning streak oracle has not publicly revealed their name, nor have they been charged by any institution; ISW has also not admitted to having employees profiting from map updates.

But the timeline of the 14-time winning streak is already laid out there: almost every time, it placed a bet when a city was considered "not yet determined" by the market; almost every time, it waited to sell when the market was close to certainty; almost every time, it captured the price difference between "not yet known by anyone" and "known by everyone".

In the past, an ordinary ISW employee responsible for updating the map might have been just an anonymous cog in the vast open-source intelligence system. What they held in their hands was merely a mark about to change color, an update about to be released, and a few seconds before clicking confirm.

This information originally could not be traded, and perhaps no one cared.

But when a map can determine the final settlement of a market, those few seconds before the mouse click gain a price. A person doesn't need to leak top-secret military information, manipulate the battlefield, or even change any facts; they just need to know a mark is about to appear sooner than everyone else to turn the authority granted by their position into one transaction after another of nearly guaranteed profit.

Innovation opens up a market where everyone can predict the future, but it also opens another door: when the settlement source can be seen in advance, altered in advance, or traded in advance, you and I can all become insiders.