What can you really do with the prediction market suddenly becoming so popular?

Author: Shao Jiadian

In the past year, if you’ve attended a few industry events related to Web3, quantitative trading, or U.S. stocks, you’ve probably heard one term: Prediction Markets.

On one side, Kalshi obtained the DCM (Designated Contract Market) license from the U.S. CFTC (Commodity Futures Trading Commission), officially integrating “event contracts” into the federal financial regulatory system for the first time;

On the other side, Polymarket was fined $1.4 million by the CFTC, expelled U.S. users, and through a series of structural and product adjustments, continued to grow rapidly worldwide, becoming the synonym for “on-chain prediction markets.”

Amidst the buzz, I’ve been repeatedly asked a question recently: “Prediction markets are so hot now, can I do something with them too?”

In this article, I want to clarify three points about this matter:

• What exactly are prediction markets, and what is the regulatory dilemma;
• Which directions related to prediction market startups are relatively feasible, and which are high-risk zones;
• And most importantly: Are you building an information tool, or have you quietly become a trading service?

What are prediction markets? Don’t just see them as “betting platforms”

From a technical and economic perspective, prediction markets are not mysterious. Their basic logic is very simple:

• Users trade with real money on whether a future event will happen;
• Contract prices essentially reflect the market’s collective judgment of the probability of that event occurring;
• Once the event outcome is determined, contracts are settled accordingly.

Unlike traditional derivatives like futures or options: prediction markets do not trade “price trends,” but rather “whether it will happen.”

For example:

• Will the Federal Reserve raise interest rates at the next meeting?
• Can a certain candidate win the election?
• Will a specific policy be passed before a certain deadline?

These are all event-based contracts.

The real question is: who should regulate this?

The regulatory divergence around prediction markets worldwide does not fundamentally revolve around “whether it resembles gambling,” but rather: should it be incorporated into the financial regulatory system?

In reality, there are three common approaches:

1. Financial Derivatives Path

• Typical example: United States
• Regulated by the CFTC, where event contracts are viewed as a special type of derivative
• Kalshi exemplifies this route

2. Gambling / Betting Path

• Common in many European countries and some Asian jurisdictions
• Prediction markets are directly regarded as “remote gambling” or “simulated betting”
• The usual result: unlicensed platforms are banned

3. Gray Area + Multi-layered Regulatory Play

• Treading the line between financial regulation, gambling regulation, and consumer protection
• Polymarket operates precisely in this complex zone

The same product can be “financial innovation” in one country and “illegal gambling” in another.

Why did prediction markets suddenly gain popularity in the past year?

This is not a coincidence; at least four underlying forces have overlapped:

1. Regulatory engagement begins to face front, not avoid

Kalshi obtaining the DCM license is more than just “having a compliant platform.”

Its real significance lies in:

CFTC starting to seriously respond to a question: can event contracts become part of the serious financial markets?

Meanwhile, the boundary of contracts related to political events, sports, and public interest issues, and the game between CFTC and the market, is clearly escalating.

This indicates: Prediction markets have entered the “phase of rewriting legal boundaries,” not just experimenting.

2. Polymarket is not “whitewashing,” but “changing the battleground”

Many interpret Polymarket’s story as:

“Fined → Compliance → Return to the U.S.”

But from a legal perspective, a more accurate description is:

Polymarket hasn’t become Kalshi; instead, through product structuring, user targeting, and technical architecture, it places itself in a more complex but short-term feasible regulatory gap.

Its approach essentially is:

• Moving away from U.S. users;
• Not touching fiat currency, not directly engaging banks;
• Using on-chain settlement and stablecoins to weaken traditional financial regulatory links.

This is an engineering effort in regulatory boundary management, not a “success story of compliance.”

3. AI and prediction markets are naturally aligned in product logic

What is the strongest suit of large models?

• Information integration
• Probability judgment
• Scenario analysis

And the essence of prediction markets exactly is: compressing dispersed information into a “probability price.”

This is why, in the past year, we’ve seen more and more products trying:

• AI-assisted information filtering;
• AI-generated event analysis summaries;
• Cross-verification of AI + market prices.

But note:

AI participates in “information and decision-making layers,” not inherently suitable for “automatic order placement.”

4. Macro reality: predicting “events” tells a better story than predicting “prices”

In a macro environment characterized by high volatility and low certainty:

• Interest rates, regulation, elections, geopolitical issues are increasingly sources of alpha;
• Compared with predicting asset prices, predicting “whether a certain event will happen” is more intuitive for ordinary users;
• And easier to structure and discuss from a regulatory standpoint.

This is also a clear reason why prediction markets are heating up significantly in 2024–2025.

If you are an entrepreneur: which directions in prediction markets are relatively “feasible”?

To see this clearly, I’ll give you a more fundamental breakdown.

Prediction market-related entrepreneurship fundamentally involves four layers:

1. Information Layer: Data, visualization, aggregation;
2. Decision Layer: Strategies, signals, probability judgments;
3. Execution Layer: Order placing, copy trading, automated trading;
4. Settlement Layer: The market itself, settlement rules, fund flows.

Which layer you stand on determines your regulatory attribute.

1. Information Layer: Data aggregation/search/visualization (lowest risk)

This layer only does three things:

• Aggregate prediction market data;
• Visualize, filter, rank;
• Help users “understand what the market is betting on.”

Not custody of funds, not acting as an agent for orders — that’s the critical line.

Typical forms include:

• Multi-platform odds and volume aggregation;
• Popular event curves, sentiment shifts;
• Data dashboards used by media, institutions, researchers.

In most jurisdictions, this layer is closer to information services or alternative data providers.

2. Decision Layer: Arbitrage/strategy/signals tools (high demand, but need restraint)

Prediction markets inherently have price discrepancies:

• Same event, different platform prices;
• Arbitrage opportunities with traditional finance and crypto derivatives.

Therefore, developing strategy middleware, signal scanning, arbitrage alerts has genuine business demand.

The key difference: Are you “pointing out opportunities,” or “executing on behalf of others”?

The former is a tool; the latter can easily be regarded as:

• Investment advisory service
• Brokerage service
• Even a precursor to asset management

3. Execution Layer: Copy trading / automation (most profitable, but also most risky)

Copy trading is attractive not because it helps users make money, but because it’s easy for the platform to generate sustainable income. That’s why it’s often the first feature regulators scrutinize.

From a regulatory perspective:

Copy trading + automated execution are often seen as a “combination of investment advice + client execution.”

In prediction markets, which are already regulation-sensitive, this layer’s risks are further amplified.

The critical boundary here is:

Does the user still retain “final confirmation rights”?

4. Settlement Layer: Building your own prediction market (strictly regulated path)

If your goal is to:

• Create a market with settlement rules;
• Build a platform that custody funds and matches trades;

then what you face is not “Web3 entrepreneurship,” but the dual challenge of financial market infrastructure + gambling regulation.

This route is not impossible, but it always involves:

• High costs
• Long cycles
• Intense regulatory game-playing

Common and most dangerous pitfalls in prediction markets

1. Custody of funds and transfer

Whenever you involve these actions, regulatory language will almost always include two terms:

• Custody
• Money Services

Whether these steps are necessary or worthwhile is a question every team must carefully consider.

2. AI giving “buy/sell advice” + one-click trading

AI summarization and information organization are fine; but providing explicit trading advice, combined with automatic execution, in many jurisdictions, closely resembles regulated financial services.

3. Covering any topic: elections, sports, public events

Topic selection itself is a compliance issue. Some events are naturally restricted in certain countries; platform rules are also increasingly becoming “quasi-regulatory.”

4. Platform tokens + rebates + profit narratives

Prediction markets are already quite gray; adding a poorly designed token can easily push you into:

• Securities regulation
• Gambling regulation
• Prepaid card / fund pool regulation

Final honest words: This is not a “can do once you think through”

The real complexity of prediction markets does not lie in product form or technical implementation. From a regulatory perspective, the issue has never been “whether you are a prediction market,” but rather: what role you are playing, and what responsibilities you are assuming.

Many teams, when introducing their products, emphasize:

• That they are just tools;
• That they do not provide investment advice;
• That they are not responsible for outcomes.

But in reality, roles are not decided by self-declaration.

Regulatory judgment does not start from your white paper or disclaimers; it directly concerns three most fundamental questions:

• Do users change their trading behavior because of you?
• Does capital flow because of you?
• Are risks amplified or concentrated because of you?

If any one answer is “yes,” then from a regulatory point of view, your project is no longer “an external tool.”

The reason prediction markets are repeatedly controversial is precisely because they are inherently ambiguous:

• They resemble finance but are not entirely financial;
• They resemble gambling but are cloaked in information efficiency;
• They sit at the intersection of public events, price signals, and user sentiment.

This means: there is no “once-and-for-all design” that guarantees long-term success in this track. Every feature choice you make today is essentially a bet on how future regulation will define you.

So if I must give a conclusion:

Prediction markets are not impossible to do, but you must accept that: it is a track that cannot rely on ambiguity or luck to survive long-term.

The real danger is not regulation itself, but that you might, unknowingly, push yourself into a position where regulation becomes unavoidable.