The prediction-market category went from a curiosity to a serious asset class in about twenty-four months. The numbers driving that shift are now public and unambiguous, and yet most institutional allocation committees still don't have a position, most major brokerages don't have a launch plan, and most media operators don't have a strategy. That gap closes inside the next two years. This is what the data says, who's already moving, and what the operators we work with are using to build their thesis.
A 6× annual growth rate is the kind of number that warrants attention on its own. The reason it warrants strategic attention specifically — not just curiosity — is that this growth is happening on infrastructure that visibly cannot meet the demand. The platforms clearing the volume today were not built to be the global venues of an asset class. They were built as single jurisdiction venues that got overwhelmed. That mismatch creates the window.
Where the volume actually came from
The clean way to understand the last 24 months is to break down what drove each leg of the curve.
The first leg, late 2023 through early 2024, was driven almost entirely by political-event volume on Polymarket — election markets, congressional markets, Supreme Court markets. This was retail flow, primarily US-based despite the geo-block, finding its way to the venue through VPNs and crypto-native onboarding.
The second leg, mid-2024 through mid-2025, was the Kalshi institutional ramp. Once the CFTC's no-action posture stabilized, professional desks started quoting on Kalshi at meaningful size. Macro funds in particular discovered that Fed-decision contracts and CPI contracts were liquid enough to use as hedges for short-duration rate exposures.
The third leg, late 2025 to today, is what changed the framing. Volume diversified beyond political and macro. Sports outcomes, commodity prints, geopolitical events, and crypto milestones all started clearing nine-figure weekly volume. This is the moment a category stops being a "political-betting site" and starts being a multi-vertical contract exchange.
| Vertical | Q4 2024 share | Q1 2026 share | Growth multiple |
|---|---|---|---|
| Political / electoral | 62% | 21% | ≈ 2.0× |
| Macro / rates / CPI | 11% | 26% | ≈ 14.1× |
| Sports / culture | 7% | 22% | ≈ 18.7× |
| Crypto / asset milestones | 9% | 16% | ≈ 10.6× |
| Geopolitical / climate | 11% | 15% | ≈ 8.1× |
The multiplier on the right is the part to dwell on. Political volume roughly doubled. Every other vertical grew between 8× and 19× in the same window. That's not a category that needs another political event to keep growing — that's a category whose growth is being held back mostly by available product surface and available regulated distribution.
Why brokerages are next, not later
The XP × Kalshi distribution agreement, signed in March 2026, is the single most useful data point we have for what comes next. The agreement isn't novel as commerce; it's novel as a signal. Kalshi explicitly chose distribution-by-partnership over expansion-by-licence, which means they expect to compete more for institutional flow than for retail.
That is good news for any brokerage that's been wondering whether it should compete with Kalshi. It almost certainly shouldn't.
What it should do — what XP is doing — is host the product on infrastructure that someone else operates. The brokerage owns the client relationship, the local language, the local regulatory standing, and the trust. The infrastructure operator owns settlement, custody, liquidity, and contract design. The economics split predictably:
- The brokerage takes the operator fee on every trade — typically 0.5% to 2.0% of notional, depending on jurisdiction and risk appetite.
- The infrastructure layer takes a smaller, capped fee for hosting, matching, and settlement.
- The trader gets a venue with the local language, local regulatory comfort, and the same liquidity they'd find on a global venue.
This is exactly the same shape as the merchant-payments stack. Stripe hosts payment infrastructure; the merchant takes the customer relationship and the margin. Nobody asks a coffee shop to write its own card-processing code. The same will be true of prediction markets for the same reasons: the infrastructure problem is hard, the distribution problem is local, and they shouldn't be solved by the same team.
The regulatory shift you can already see
The narrative in 2022 was that regulators were hostile to prediction markets. That is no longer accurate. In Q1 2026, three significant shifts are visible:
- United States. Kalshi's no-action position is now extended through several rule-making cycles. The CFTC has effectively signaled that event contracts on a CFTC-registered exchange are inside the perimeter, which removes the existential overhang that killed InTrade.
- European Union. ESMA's discussion paper on event-based derivatives, published December 2025, treats binary outcome contracts as a recognised category for the first time. The path to a regulated EU venue is no longer theoretical; it's a multi-year procedural question.
- Brazil. The CVM and the Ministry of Finance are jointly consulting on a domestic framework, with B3 publicly committing to exploring the asset class. The XP × Kalshi deal pressures this consultation forward.
What you don't have yet is universal clarity. What you do have is the removal of the worst-case scenarios. That is the regulatory environment in which institutions start writing checks.
Will the EU finalise an event-contract regulatory framework before Q4 2026?
Who is already positioning
A handful of operator categories have moved already. Watching them is the cheapest way to understand who'll move next.
Crypto-native exchanges. Kraken and Bitget both signaled work on prediction-market integrations in late 2025. The thesis is obvious: their existing user base trades volatility instruments, and a binary contract is the simplest possible volatility product. The execution challenge is regulatory — most crypto exchanges aren't licensed to host outcome contracts in their primary jurisdictions.
Local brokerages outside the US. XP is the named example, but we're aware of at least four other Latin American brokerages running internal feasibility on the same model. The economics are clearer for them than for any US comparable, because the local-distribution moat is harder to dislodge.
Sports books and consumer-fintech apps. A binary contract on a sports outcome is functionally identical to a parlay leg, but with better unit economics for the operator and lower vig for the trader. We expect at least one major sports book to ship a prediction-market SKU before the end of 2026.
Media operators with engaged audiences. This is the category most underestimated by traditional finance. A newsletter with 50k engaged subscribers, a Twitch streamer with a community of 200k, or a specialist publication with a vertical audience can host markets specific to their niche. Conversion is much higher than against a general retail audience because the audience already has views, and the customer-acquisition cost is effectively zero — the audience is already aggregated. We've seen creator-led venues hit five-figure weekly volume within their first ten days live, on audiences a fraction of the size that a traditional retail brokerage would consider worth targeting.
What it costs to wait
The instinct, when a category is moving fast, is to wait until the "winners" emerge. With prediction markets, that instinct is miscalibrated. The protocol thesis says there is no single winner at the venue layer — there will be hundreds of operator brands, each serving a specific audience, all running on shared infrastructure.
Waiting until 2027 to start thinking about a launch means missing the formative period in which the operator standards are set, the distribution patterns are codified, and the early operator brands lock in audience attribution. Waiting also assumes that the cost of launching stays where it is. We don't expect it to.
What we expect to break before it scales further
A category growing 6× a year doesn't grow that way smoothly. Three known stress points are visible from where we sit, and operators planning to enter should be aware of them because each is also an opening.
Stress point one: settlement bandwidth. The optimistic-oracle model that resolves most on-chain markets today was sized for a fraction of the contract count we now expect. UMA's voter base, the challenge windows, and the dispute-bond capital are all going to be load-tested by a category clearing 50,000+ markets per week. Expect either a hardening of the existing rails or a fragmentation into multiple resolution networks specialised by vertical. Operators that inherit a managed settlement layer rather than running their own will navigate this transition without engineering work.
Stress point two: fragmented liquidity across operator brands. The protocol thesis says hundreds of operator brands will host the same underlying markets. Without shared liquidity, that fragmentation makes every individual book thinner than the legacy single-venue books. With shared liquidity (the model we operate), order flow on any operator brand contributes to and draws from a unified book. Operators that don't integrate into a shared-liquidity arrangement are going to feel a structural disadvantage in spreads inside 18 months.
Stress point three: regulatory whiplash in 2027. The current trajectory in the EU and Brazil is constructive, but it's not linear. The most likely path includes at least one regulatory pull-back — probably driven by a high-profile resolution dispute or a leveraged loss on a retail account — that triggers a tightening cycle. Operators with clean compliance posture survive and grow during those cycles; operators leaning on grey-area frameworks lose distribution overnight. Build assuming a tightening event will happen.
None of these are fatal. They're the kind of stress points every asset class encounters in its formative years. The operators that do well are the ones that read them as inevitable and design their posture accordingly.
What the operators we work with are doing right now
The operator playbook that's emerging across the brokerages, media companies, and institutional desks we work with has four moves:
- Define the audience and the markets. Not "we want to host prediction markets" — "we want to host political and macro markets for our 80k retail clients in São Paulo." The narrower the audience definition, the faster everything else gets.
- Pick the regulatory posture. The choice is between launching under an existing offshore framework (faster, fewer constraints) versus working toward a domestic regulated venue (slower, much larger TAM). Most operators we speak with are doing both — soft launch under an offshore framework while the domestic license matures.
- Decide on the infrastructure layer early. This is where the build vs license analysis matters. The deltas between six months and six weeks to launch are entirely in this decision.
- Treat liquidity as a Day-One feature, not a Month-Six problem. We've written separately on why shared liquidity decides who wins the next wave. The short version: operators who try to bootstrap liquidity from cold lose 60–80% of their launch audience inside 30 days.
The category will look very different in 18 months. The operators positioning now will be the ones whose names are on the leaderboard when it does.