Three days before President Trump landed in Beijing, Goldman Sachs published its Trump-Xi summit preview.

The note projected that China would agree to buy more U.S. agriculture products, energy, and aircraft in exchange for avoiding further tariff escalation. The bank said the meeting "could act as a tactical catalyst for strength in the Chinese yuan and in Chinese equities."

The note was professionally hedged. Specific scenarios were named. Probabilities were implied. Catalysts were identified.

Here's what the note didn't say: every projection in it was already trading on a prediction market with deep liquidity and a specific dollar probability attached.

Kalshi was pricing Boeing aircraft announcements at 86%. Polymarket was pricing the soybean purchase at 36%. The Trump-mentions-Iran market was trading at 61%. The tariff truce extension was priced across multiple contracts. Wolfe Research, Jefferies, and Goldman were all publishing notes that aligned suspiciously closely with prediction market consensus.

This isn't an accusation of plagiarism. It's an observation about market structure.

Wall Street has crossed a line that almost nobody is reporting. Institutional research desks are now treating prediction markets as a primary data input — not an exotic alternative data source, not a sentiment indicator, but a direct numerical reference point for institutional forecasts. The summit was the moment this became impossible to ignore.

Here's what actually happened.

The Setup

Trump-Xi summit, May 14-15, 2026, Beijing. The first major U.S.-China bilateral since Trump's second-term tariff escalation. Markets across asset classes had been positioning for the outcome for weeks. The Hang Seng Tech Index had moved in anticipation. Boeing stock was up nearly 2% on Wednesday before Trump even landed. Bond markets were pricing tariff probability shifts. Currency markets were pricing yuan strength.

In normal times, the inputs to those positioning trades would have been: company guidance, government leaks, expert commentary from think tanks, and historical patterns from prior bilateral meetings.

In May 2026, the input was prediction markets.

By the time Trump's plane took off from Joint Base Andrews, Polymarket alone hosted 117 active markets on Trump-Xi outcomes. $36.7 million in trading volume across the cluster. Kalshi added dozens more, including the now-famous Boeing aircraft purchase contract at 86% probability. A separate $7.8 million flowed through markets predicting specific words Trump might use during bilateral events.

Then the research notes started landing.

What Goldman, Wolfe, and Jefferies Wrote

Goldman's note focused on what they called a "narrow but tactically important" agenda: trade, export controls, tariffs, semiconductor restrictions, rare earth exports. They expected China to commit to U.S. agriculture, energy, and aircraft purchases. They flagged the meeting as a positive catalyst for Chinese equities.

Tobin Marcus at Wolfe Research wrote: "The speculation is that Trump wants this to be the largest order ever announced, which could mean a Boeing purchase commitment in the triple-digit billions." His note advised investors to "await clarification from the company about how 'real' those numbers are."

Edison Lee at Jefferies focused on AI and semiconductor chip restrictions, predicting these topics would be discussed given the executives traveling with Trump (Micron's CEO, Meta's president).

These three analyses are professionally rigorous. They cite executive travel, recent statements, and bilateral history. They're the kind of summit previews institutional research desks have produced for decades.

What's new is that every specific outcome they projected was already a tradeable contract with a live probability attached. Boeing announcement: 86% on Kalshi. Trump mentions Iran: 61%. Trump mentions oil or gasoline: 59%. Trump mentions AI: 54%. Soybean purchases: 36%. Even the handshake duration market — 8.5 seconds — was being actively traded.

Compare any of these probabilities to the equivalent claims in the bank notes. The match isn't perfect, but the directional alignment is striking. Where research disagrees with the market, it's usually within 5-15 percentage points. Where it agrees, the language frequently mirrors the market's framing.

Table 1: "What the Markets Priced vs. What Wall Street Wrote"

The pattern: Where prediction markets had liquidity, institutional research desks tracked closely. Where the markets covered outcomes too specific or absurd for serious analysis (handshake duration, kiss probability), the desks ignored them entirely. The granularity gap is the dividing line — and it's narrowing.

This could be a coincidence. Analysts and traders are processing the same news through similar mental models. Both groups are good at their jobs.

But there's another possibility: the markets are reaching the research desks. And the research desks aren't always disclosing it.

Why This Is Happening Now

The structural conditions for prediction markets to become institutional research inputs are all in place simultaneously for the first time:

Volume has crossed the credibility threshold. Kalshi reported $13B in March 2026 volume. Polymarket added $10.6B. Combined $23B+ in a single month. Until 2024, prediction market liquidity was thin enough that institutional analysts could dismiss it as noise. At $23B/month with deep order books on specific contracts, the dismissal stops working. Liquidity is now sufficient to support the kind of price discovery institutional desks respect.

Regulatory clarity is emerging. Kalshi is CFTC-regulated. Polymarket operates internationally but has institutional partnerships in motion. The "is this even legal" objection that kept institutional analysts away from prediction markets in 2022-2023 has weakened. Bank compliance departments are increasingly comfortable with employees referencing this data.

Granularity matches institutional needs. Polymarket lists 117 markets on a single summit. That's more outcome granularity than any single research analyst could produce manually. When a research desk wants to know "what's the market-implied probability of a tariff truce extension by Q3," there's a contract priced for it. When the same desk wants to know "what's the implied probability of a soybean announcement specifically," there's a contract for that too. The granularity matches the way institutional analysts actually think about complex events.

Commercial relationships are forming openly. CNBC and Kalshi have a commercial relationship that "includes customer acquisition and a minority investment." That disclosure appears at the bottom of every CNBC article that cites Kalshi prediction market data. The line between editorial coverage and commercial partnership is getting redrawn in public. When the line moves in editorial, it has already moved in research.

The data layer is finally accessible. Until very recently, institutional research desks couldn't easily query prediction market data across venues. Dome's existence helped. Dome's acquisition by Polymarket complicated things. Independent data infrastructure — across platforms, with normalization and historical depth — is what makes this institutional integration scalable. Without it, every analyst is doing the synthesis manually. With it, the data flows into the same dashboards that already contain Bloomberg terminals, Refinitiv feeds, and FactSet pulls.

Title 2: "The Alternative Data → Primary Data Transition Curve"

The pattern: Each transition takes 5-10 years. Prediction markets just moved from "Selective Use" to "Embedded Reference" — and the Trump-Xi summit was the visible marker. The next transition (to Primary Input) is mechanical from here.

What Wall Street Used to Call "Alternative Data"

Five years ago, prediction market data was alternative data. Hedge funds with experimental quant teams might look at it for fun. A trader might mention an interesting Polymarket contract over drinks. Nobody put it in a research note.

The progression from alternative data to primary data follows a recognizable pattern. Satellite imagery used to be alternative data — until oil traders started routinely pricing inventory off it. Credit card transaction data used to be alternative data — until retail analysts started routinely citing it in earnings previews. Web scraping data used to be alternative data — until consumer goods analysts started using it as the baseline.

Each transition followed the same arc: novelty → curiosity → selective use → embedded reference → primary input. The data starts in the margins and moves to the center over five to ten years. By the end, it's so embedded that analysts cite it without explaining what it is. The audience already knows.

Prediction market data is at the "embedded reference" stage right now. Sometimes cited, sometimes not. Frequently used as the framework for what scenarios to even consider. Increasingly the underlying source of "consensus expectation" claims in research notes that don't disclose where the consensus came from.

This week's Trump-Xi coverage shows the transition in real time. CNBC openly cited prediction market percentages in its summit preview. Goldman, Wolfe, and Jefferies presented analyses that tracked closely with the markets without explicitly citing them. The CNBC piece even noted the Kalshi commercial relationship as a footnote — disclosure that itself signals the integration is mature enough to require disclosure.

In another twelve months, this is what every major institutional research note on geopolitics will reference, either openly or implicitly. The early-adopter desks already have prediction market dashboards integrated into their workflows. The late adopters are figuring out compliance frameworks. The structurally resistant are aging out.

What This Means for Everyone Else

Three implications matter regardless of where you sit in this ecosystem.

For institutional research consumers: When you read a research note now, the question is no longer "what does this analyst think." It's "what is the underlying prediction market saying, and what is the analyst adding on top of it?" Sometimes the analyst is providing genuine alpha through better interpretation. Sometimes the analyst is repackaging what the market already showed for free. Telling the difference requires reading the underlying contracts directly.

For traders and analysts: The competitive landscape has shifted. The edge that used to come from reading prediction markets is being arbitraged away as institutional desks integrate the same data. The new edge comes from cross-venue synthesis (different platforms price the same events differently), structural adjustment (the price is not the probability), and on-chain signals (volume-price mismatch, whale concentration, capital flow patterns). The raw price is becoming commodity information. The structural interpretation is becoming the differentiator.

For the prediction market industry: This is the moment that determines whether prediction markets become Bloomberg-tier financial infrastructure or remain a category-specific niche. The path through Bloomberg-tier infrastructure runs through institutional research workflows, which means it runs through data infrastructure, normalization, historical depth, on-chain integration, and the kind of API access that makes institutional embedding practical at scale. The platforms that get this right become indispensable. The ones that don't get acquired or sidelined.

The Quiet Part

Here's what nobody on Wall Street wants to say out loud:

Prediction markets are no longer a competitor to institutional analysis. They are a partial replacement for it.

When 117 markets exist on a single summit and trade with $36.7 million in volume, the question stops being "what does the analyst think will happen" and starts being "what does the analyst know that the market doesn't already price." For most outcomes most of the time, the answer is "not much." The market has already aggregated the information that analysts spend their days processing.

The analysts who add real value going forward will be the ones who understand market structure deeply enough to identify where prediction markets are systematically wrong — where liquidity is thin, where resolution rules introduce bias, where regulatory constraints prevent the most informed traders from participating. That requires understanding the data layer at a level that most current institutional analysts don't yet possess.

The institutional research function is being restructured. Not eliminated — restructured. The new function looks more like "interpretation of market signals" than "production of market opinions." That's a different job, requiring different tools, different training, and different data infrastructure.

The Trump-Xi summit was the moment this became visible. The question is what gets built on top of the realization.

That's the part Goldman didn't say. But the market did.

This is the eleventh installment in the Assymetrix Intelligence Brief series.

Previous: "The Price Is Not the Probability: Why Everything You Think You Know About Reading Prediction Markets Is Wrong"

Assymetrix is building the intelligence and synthesis layer for prediction markets. Cross-platform data infrastructure for traders, builders, researchers, and the institutional analysts whose workflows are about to change.

assymetrix.com/blog