Strategy sold BTC, Polymarket said NO: Inside the biggest inter-subjective market dispute of 2026

Introduction

Prediction markets are probably one of the most complex categories of markets to operate. They involve speculating on the outcome of events whose resolution is not always straightforward. For binary markets, such as the outcome of a football match, there is normally little room for doubt. However, for more sophisticated markets, disagreements about the outcome of an event are not uncommon, depending on how each participant interprets the situation.

Subjectivity is the main challenge facing prediction markets. First, because it is intrinsically linked to real-world events. Second, because it can become amplified when the stakes are high. And finally, because it becomes overwhelmingly dominant whenever money is involved.

Imagine that your country is playing in the World Cup final. A highly controversial situation leads to a penalty in the final moments of the match while the score is tied. Even if the referee has a clear rulebook to follow, the final decision still depends on interpretation. More importantly, your opinion of the referee's decision will inevitably depend on whether the penalty benefits your team or not.

These are known as "intersubjective markets." They represent one of the main challenges for prediction markets because, in their current form, they are incapable of handling them without disadvantaging some users. Numerous recent examples have demonstrated this, including the Zelensky Suit market, the US government shutdown market, the Khamenei death market, and the TikTok case.

In fact, we had already planned to publish an in-depth research piece on intersubjective markets, their complexity, and the potential solutions available to resolve them correctly. However, following the recent Strategy and Polymarket controversy, we decided to adapt that research and publish it today instead.

Let's dive in.

OAK Discord

Market insights, alphas, data, and discussions, all in one place.

---

Strategy sold BTC, Polymarket says "NO"

A recap of the facts

On June 1, 2026, media outlets revealed that Strategy had sold 32 BTC in order to fund dividends for its STRC preferred stock program. Within just a few days, this seemingly minor transaction became the center of one of the largest controversies in Polymarket's history.

At the center of the dispute was a relatively straightforward market: "Will Strategy sell any Bitcoin by May 31, 2026?". Roughly $400 million was traded on the market, making it one of Polymarket's largest markets of the year. The issue is that it ultimately resolved to "No", even though Strategy had in fact sold BTC during the month of May.

The problem stems from the fact that the transaction itself and the public announcement of the transaction did not occur at the same time. Strategy filed its 8-K form with the SEC on May 30, 2026, revealing that the company had sold 32 BTC between May 26 and May 31, 2026. However, the filing was only made public on June 1, 2026, after the Polymarket market had already expired.

As you can probably tell, this is a textbook example of an intersubjective market. Just hours after the filing became public, holders of YES and NO positions found themselves defending two radically different interpretations of the exact same reality.

Details of the dispute

To understand why, we need to examine the market rules. When the market was created, Polymarket stated that "the market would resolve to YES if Strategy sold any portion of its Bitcoin holdings before May 31, 2026 at 11:59 PM ET." The description also specified that "official communications from Strategy, on-chain data, and a consensus of credible sources would serve as primary resolution sources."

For YES holders, the situation therefore appeared completely straightforward. The BTC had been sold before May 31, Strategy confirmed it in an official regulatory filing, and the market explicitly stated that both company disclosures and on-chain data could be used to determine the outcome. Furthermore, the dispute window exists precisely to handle situations like this.

However, as the dispute unfolded, an additional clarification emerged. This clarification stated that "information used for resolution must be publicly available before market expiration." In other words, even if the sale itself occurred before May 31, evidence of that sale could no longer be considered if it only became public after the market had closed.

This clarification is precisely what triggered the backlash from a portion of the community. Several traders accused Polymarket of changing the interpretation of the market after hundreds of millions of dollars had already been wagered, especially since the market title referred to the sale itself rather than the public disclosure of the sale.

Conversely, supporters of the NO resolution argue that Polymarket merely applied a practice that had already been used in previous cases. In their view, accepting information that becomes public after expiration would create impossible-to-manage situations and undermine the integrity of market deadlines.

After multiple rounds of disputes and a final vote through UMA's resolution mechanism, the market ultimately resolved to NO on June 4, 2026. But beyond the final outcome, this case exposed a much deeper issue: how should a market be resolved when multiple legitimate interpretations of the rules coexist? That is exactly what defines an intersubjective market.

---

Understanding intersubjective markets

The problem of subjectivity

At first glance, the Strategy market appeared extremely simple. Did a company sell Bitcoin before a given date? The answer should be either yes or no. Yet, as we have just seen, this event produced two completely different interpretations, neither of which is inherently absurd.

What this situation reminds us is that prediction markets are not merely about outcomes. They are also, and perhaps primarily, about how those outcomes are interpreted. When a seemingly simple question allows for multiple legitimate interpretations, subjectivity enters the equation and resolution becomes difficult. This is precisely why we refer to these as intersubjective markets.

The Strategy dispute is far from an isolated case. Over the past few months, several high-profile markets have already exposed the limitations of the current model.

• The "Zelenskyy Suit" market generated approximately $200 million in volume in May 2025 and ended in a dispute over whether the Ukrainian president's outfit technically qualified as a suit. Some argued that it did, while others strongly disagreed.
• Another example was the market asking whether Khamenei would be removed before February 28. By the expiration date, no official information was available to confirm the Ayatollah's death. Interestingly, in that case Polymarket allowed an extension of the resolution period (unlike in the Strategy case) in order to wait for confirmation from a credible source.

The combination of high financial stakes and outcomes that exist in a gray area inevitably leads to frustration among losing participants. Given the growing number of cases where market resolutions contradicted what a significant portion of users considered to be reality, it is becoming increasingly clear that prediction markets still struggle to handle intersubjectivity effectively.

What makes a market subjective?

What makes a market "intersubjective" becomes fairly obvious quite quickly: it is when its resolution requires some form of human judgment, when it asks a question whose answer depends on interpretation, social consensus, or a definition that is not always universally accepted.

The Strategy case illustrates this perfectly. The real question was what exactly it meant to "sell Bitcoin before May 31." Did the transaction simply need to take place before that date? Or did the information also need to be public before the deadline? A market titled "Will Strategy announce the sale of BTC before May 31?" would probably have solved the issue, but Polymarket clearly did not anticipate this scenario.

What is even more interesting is that poorly written rules can sometimes transform a market that is supposed to be objective into a disputed one. For example, during the U.S. government shutdown at the end of 2025, Polymarket created a market allowing users to speculate on the date the shutdown would end.

Even though the funding bill that allowed the government to reopen had been passed on November 12 and virtually all media outlets considered the shutdown to be over, the market was not resolved to "YES" on that date. The reason is that the rules explicitly referred to an announcement from the Office of Personnel Management (OPM), which never came.

Intersubjective markets attract traders

Paradoxically, the most disputed markets are often the ones that attract the most liquidity. Obviously, sports betting, crypto prices, macroeconomic data, corporate earnings, weather forecasts, and political events perform very well on Polymarket and are not intersubjective markets.

However, data shows that during UMA voting periods for the resolution of certain markets, prices generally fluctuate between 20% and 40%, meaning that sophisticated traders view these volatile and ambiguous markets as profitable opportunities rather than risks to avoid.

The Strategy case is probably the most striking example to date. As debates around the resolution intensified after June 1, tens of millions of dollars continued changing hands between YES and NO supporters. The market was no longer a bet on whether Strategy had sold Bitcoin, which had officially happened, but rather a bet on how Polymarket would interpret its own rules.

This type of situation raises a fundamental question: should prediction markets prioritize a strict reading of their rules, or the reality they are supposed to predict? More importantly, can UMA’s "decentralized" resolution model still be considered the best solution for a platform as large as Polymarket?

---

UMA and dispute resolution on Polymarket

General framework

Polymarket primarily relies on the UMA protocol for dispute resolution. It is built around a two-layer oracle architecture: an Optimistic Oracle (OO) for fast and low-cost resolution, and a Data Verification Mechanism (DVM) that acts as the final arbitrator in the event of a dispute.

In practice, once an event concludes, trading stops and the market enters the resolution phase. The smart contract requests the information needed for settlement (data, price, etc.), and an Asserter (an entity recognized by Polymarket) provides it. The Asserter posts a bond to guarantee good faith and includes all the parameters necessary for settlement (identifier, timestamp, event outcome, and settlement currency).

During a two-hour window, anyone can challenge the market outcome directly through UMA’s interface by posting a bond equal to that of the Asserter (typically 750 USDC). Once a challenge is submitted, a 24 to 48-hour discussion period begins during which anyone wishing to contribute evidence or arguments may do so on UMA’s Discord server.

Once the discussion period has ended, the dispute is escalated to the Data Verification Mechanism (DVM), UMA’s arbitration layer governed through on-chain governance. UMA token holders vote to determine the correct outcome based on the protocol’s official methodologies (UMIPs). This vote generally lasts between 48 and 96 hours, and the aggregated result becomes the official truth (canonical value).

Note that if the challenger is correct, they recover their bond and receive half of the market creator’s bond as a reward. Otherwise, the Asserter receives half of the challenger’s bond.

The case of intersubjective markets

The Strategy case serves as a reminder that oracles are extremely effective when it comes to verifying objective and measurable facts, such as an asset price, an election result, or the score of a sporting event. However, they quickly reach their limits whenever a market requires human interpretation.

• The first issue lies in the way markets are written. A question that appears perfectly clear when it is created may reveal multiple legitimate interpretations once unexpected scenarios emerge. The controversies surrounding Zelensky, TikTok, the U.S. government shutdown, and more recently Strategy all illustrate the same difficulty: rules can never anticipate every possible scenario.
• A second challenge emerges even when the facts are known by everyone. Their interpretation may vary depending on the cultural, political, or legal context of the participants. In these situations, the oracle is no longer simply determining what happened, but also what it means. And that question does not always have an obvious answer.
• Governance mechanisms introduce their own limitations as well. In theory, UMA token holders are incentivized to vote for the most accurate resolution. In practice, as with any on-chain governance system, the economic interests of participants can sometimes conflict with the search for truth. The larger the amounts at stake, the more visible this tension becomes.

The industry is fully aware of these limitations. In recent years, several improvements have been introduced to strengthen resolution mechanisms: AI-assisted analysis, stricter market-writing standards, dynamic bonding systems, and a broader diversification of information sources used by oracles. These innovations help reduce disputes, but they do not solve the underlying problem.

Ultimately, intersubjective markets are built around a contradiction that is difficult to eliminate. They require a protocol to produce a binary answer from a reality that is not always binary. As long as some questions require human judgment, there will be situations where multiple interpretations can reasonably coexist. And that is exactly what the Strategy case has reminded us.

---

Our view

The Strategy case will probably go down as one of the biggest resolution failures in Polymarket’s history. Not because the market itself was particularly complex, but because the platform found itself trapped by its own mechanisms and ultimately resolved a market in direct contradiction to what actually happened, despite claiming to be a source of objective truth.

Our view on this event is straightforward. Strategy did in fact sell 32 BTC before May 31, 2026. This point is not disputed by anyone today and is even confirmed by an official document. Therefore, the market should have resolved to YES.

The main disagreement within the community concerns when that information became public. Yet this is precisely the type of situation that the dispute mechanism exists to handle: once a market closes, users have a window during which they can submit a challenge and a 24 to 48-hour period to review available information and correct potential errors.

In this specific case, information regarding Strategy’s BTC sale was published exactly during the window designed to resolve disputed outcomes. As such, the argument that "evidence published after market expiration cannot be considered" is difficult to defend. Otherwise, there would be little reason to allow a dispute period in the first place.

Of course, we are not suggesting that markets should be revised weeks or months after they expire. But when information emerges only a few hours after closure and during the designated dispute period, the objective should remain finding the outcome that most closely reflects reality. That is not what happened here.

At the same time, we understand the concerns surrounding traders who aggressively bought YES after the information became public. Nevertheless, several alternative solutions existed. Polymarket could have limited profits to positions opened before market expiration, refunded positions entered after the new information appeared, or even voided the market entirely in order to preserve its credibility.

Instead, the platform chose the only option that validated an outcome contrary to reality, rewarding those who were wrong, penalizing those who were right, and making no distinction between those who entered the trade after the announcement and those who genuinely anticipated the outcome.

And this is without even mentioning a problem we have highlighted for years: the UMA protocol itself. The so-called "decentralized governance" vote gave the impression that the main participants arranged an outcome that suited their interests rather than genuinely seeking the truth.

According to on-chain data, the ten largest UMA holders control more than 60% of the protocol’s total voting power. In other words, a small group of actors theoretically wields enormous influence over the resolution of Polymarket’s most sensitive markets. When UMA has a market capitalization of roughly $36 million while supposedly securing the resolution mechanisms of a platform processing billions of dollars in volume, it becomes difficult not to see the issue.

This is also one of the few reasons why a native POLY token for Polymarket appears relevant to us. Beyond speculation, it could create a governance and economic security mechanism that is more aligned with the actual scale of the platform. Today, one of the largest prediction markets in the world relies in part on a system whose theoretical cost of corruption appears surprisingly low relative to the financial stakes involved.

Ultimately, the real issue is not that Polymarket encountered an intersubjective market. These situations are inevitable and will continue to exist. The problem is that when confronted with a textbook example of one, the platform chose a resolution that may satisfy its internal procedures but does not reflect the reality of the event it claimed to measure. And for a prediction market, that is probably the worst possible outcome.

Become Premium

Unlock all our research and get the right insights, at the right time.