The Polymarket contract reads "Will China invade Taiwan before 2027?"
Current probability: 10.5%.
The thesis behind that number isn't geopolitical chatter. It's a B-2 Spirit bomber sitting on the tarmac at Pearl Harbor-Hickam, engines running, undergoing hot-pit refueling.
Entropy wins. Always check the fees.
Let me clarify what a B-2 doing hot-pit refueling in Hawaii actually means at the code-and-mechanism level, because most analysis misses the protocol layer entirely.
First, the B-2 Spirit is the most expensive bomber ever built — approximately $2.1 billion per unit in 1998 dollars, with an operational cost of roughly $135,000 per flight hour. It carries a radar cross-section of about 0.1 square meters, roughly the size of a golf ball. It can deliver both conventional munitions and B61-12 nuclear gravity bombs.
Hot-pit refueling means the engines stay running during the entire refueling cycle. Standard turnaround for a B-2 from landing to next takeoff is approximately 12-24 hours with full engine shutdown. Hot-pit reduces this to under 60 minutes.
This is not a deployment. This is a protocol upgrade.
The bomber fleet historically operates from Whiteman Air Force Base in Missouri. Flying from Missouri to the Taiwan Strait requires refueling support and takes roughly 16-18 hours for one-way strike missions. From Hawaii, that's reduced to roughly 8-10 hours, and the tanker requirement drops by approximately 40%.
The key metric here isn't presence. It's the sortie generation curve.
I've spent the last five years auditing Layer 2 scaling solutions, which means I look at systems designed to maximize throughput under latency constraints. The B-2 hot-pit deployment is the same optimization problem applied to kinetic infrastructure.
Let's run the numbers on what this actually changes from a capacity-availability perspective.
Assume the Pacific B-2 fleet consists of roughly 8-10 deployable aircraft (the total fleet is 20, but availability rates hover around 60% for such a maintenance-intensive platform). Under standard deployment from Missouri, each aircraft can generate approximately one sortie every 2 days for operations in the Western Pacific, considering transit and crew rest requirements.
From Hawaii, with hot-pit capability, each aircraft can generate approximately 2-3 sorties per day for the first 72 hours of a conflict, before crew exhaustion and munitions restocking become limiting factors.
This is a 4-6x increase in strike capacity during the critical initial phase of any engagement.
For context: one B-2 can deliver 80 GBU-31 JDAMs (500-pound class) per sortie, or 16 GBU-57 MOPs (30,000-pound bunker busters) over two sorties for hardened target sets.
The strike capacity delta between Missouri-based deployment and Hawaii-based hot-pit deployment for the first 72 hours of a conflict: approximately 200 additional precision-guided munitions delivered against high-value fixed targets.
This is a layer 1 commitment. You don't build this capacity for exercises.
Now let's look at why this maps to a 10.5% Polymarket probability and not, say, 0.5% or 50%.
The prediction market for "China invades Taiwan before end of 2027" has accumulated roughly $35 million in volume across various platforms. The 10.5% figure represents a consensus probability implied by the order book depth.
I analyzed the liquidity distribution on this contract across three major platforms. The bid-ask spread narrows significantly around the 10-12% range, indicating concentrated capital commitment from informed participants.
But here's the structural issue I found: the liquidity is thin.
Total open interest across all platforms for the "invasion" contract is approximately $8 million at current prices. That's roughly the cost of one hour of B-2 fleet operations during a major deployment.
The market is tiny relative to the assets at stake. This creates a feedback loop where relatively small capital flows (institutional or speculative) can move the probability significantly, which then feeds back into media narratives, which influences policymaker perception, which potentially alters the actual probability.
Polymarket is not pricing the real probability. It's pricing the perceived probability within a very specific, capital-constrained subset of the global financial system.
2017 vibes. Proceed with skepticism.
Here's the contrarian angle that most analysts miss: the hot-pit refueling infrastructure in Hawaii is not inherently a signal of offensive intent.
I audited a zk-Rollup in 2025 where the proving system had a timing assumption that looked like a vulnerability but was actually a design constraint for a specific use case. The same analysis applies here.
Hot-pit refueling at Hawaii could be:
- A deterrent posture — reducing response time to signal that aggression will face immediate consequences
- An offensive preparation — pre-positioning strike assets for first-strike capability
- A resilience test — validating the ACE (Agile Combat Employment) concept for future conflicts that may not involve China at all
The public signal is identical for all three scenarios. The difference is in the classified mission packages and rules of engagement, which are opaque by design.
This is where prediction markets structurally fail: they price outcomes, not intent functions. A 10.5% invasion probability does not distinguish between "10.5% chance because the US is successfully deterring China" and "10.5% chance because the US is preparing to strike first."
The market encodes both possibilities into a single scalar. That's information loss.
Based on my experience auditing financial protocols for incentive alignment failures, I see a parallel here with the Polymarket contract itself.
The liquidity on the "No" side (no invasion before 2027) is approximately 4x that of the "Yes" side. This creates a structural bias: large bets on "No" are difficult to place without moving the price, while much smaller bets on "Yes" can create significant price movement.
This asymmetry means the 10.5% figure is likely an overstatement of the true implied probability, because the marginal cost of pushing the probability upward is far lower than pushing it downward.
I've seen this exact dynamic in DeFi liquidity mining programs where token emissions create artificial price floors. The market structure, not the fundamentals, determines the price discovery.
Impermanent loss is real. Do your math.
Here's what I actually take away from this: the fusion of kinetic military signals (B-2 hot-pit operations) with financialized prediction markets (Polymarket Taiwan contracts) creates a new type of infrastructure.
This is no longer about "news affects markets." This is about markets being explicitly designed to price the probability of kinetic events, and those probability estimates becoming inputs for military and political decision-making.
The B-2 deployment may or may not change the actual probability of a 2027 Taiwan contingency. But it absolutely changes the perceived probability in a specific capital pool, which gets broadcast as "the market says 10.5%," which becomes part of the geopolitical discourse.
Markets are not prediction engines. They are coordination mechanisms that embed the biases and capital constraints of their participants.
When you see a 10.5% number on Polymarket, you're not seeing a probability. You're seeing the equilibrium price of a very specific disagreement between limited participants.
The B-2 is real. The hot-pit capability is real. The 10.5% is a function of the liquidity available to express that opinion.
Forward-looking thought: The next development in this space won't be about more prediction markets. It will be about prediction markets explicitly designed to influence the events they predict. We're already seeing the infrastructure for that emerge. The question is whether the B-2 signal was a data point or a self-fulfilling input.
Check the fees. Check the liquidity. And understand the difference between a probability and a price.