The founders of prediction markets all say they're building the next Toutiao.

@Guo Yunxiao

Elsewhere has recently taken notice of a category of companies that look almost science-fictional: prediction markets.

According to our understanding, over the past two weeks, multiple USD-denominated funds have been making intensive outreach to Chinese-founded teams in this space. At present, at least three companies have entered VC radars.

Before introducing these three companies, a brief primer on prediction markets themselves.

Consider: Which will be the best model company by late January 2026? Who will win the 2026 NBA championship? When will Stranger Things release its next season? How many tweets will Elon Musk post next week? Will Taylor Swift have a baby next year?

On these platforms, you'll find countless such absurd, lottery-like questions. For any given question, as long as people put real money on its future outcome, someone will end up pocketing someone else's cash when the event finally resolves.

This is called: a prediction market.

Two overseas players, Polymarket and Kalshi, have successively secured regulatory compliance in the US market. Over the past few months, they completed two lightning-fast funding rounds apiece, with valuations increasing tenfold, now reaching the $10–15 billion range.

Polymarket was founded in 2020 by Shayne Coplan, a crypto-native NYU dropout. Polymarket started in the crypto world, got fined and booted from the US market, and ultimately embraced "offshore libertarianism" wholeheartedly.

Kalshi was founded in 2018 as a Y Combinator project. Its two founders, Tarek and Luana, brought elite pedigrees from Goldman Sachs and Citadel. They insist on calling themselves a "fintech company" and have done everything possible to demonstrate compliance to the CFTC (Commodity Futures Trading Commission).

Some view this as gambling; others see it as an "information aggregator closer to the truth."

The shift came in late 2024. While mainstream polls stubbornly showed a dead heat, Polymarket's real money had already priced in Trump's odds at over 60% — a figure so precise that prediction market data was subsequently integrated directly into Bloomberg terminals on Wall Street. This drove both platforms toward regulatory compliance and capital markets progress.

One has to admire human imagination — betting can be systematized into a business.

Back to the Chinese teams. The three we know of each have distinct personalities.

The first has virtually no presence on Chinese social media, yet maintains a Chinese-language website. The team's backgrounds span AI algorithms, quantitative trading, and international investment banking. They had previously explored financial agents, which led to the logic of "gaining market insights through trading."

The topic examples on their website also reveal a strong "embrace compliance" orientation, such as: "Will a certain brand's next-generation self-designed chip adopt N3P or 2nm process?" This company has already received investment from multiple USD funds.

The second is markedly different. Its founder, co-founder, and official accounts are all micro-influencers on X — you'll spot them in virtually any discussion about prediction markets. These young people radiate a wilder, more hacker-ish energy. They worked at major tech companies domestically and abroad in their early careers, but ultimately converged through their trials in crypto.

They carry a heavier crypto aesthetic, rejecting the centralized topic moderation mechanisms of platforms like Polymarket. They believe anyone should be able to create topics, with better recommendation algorithms solving liquidity in the long tail. Their investors include multiple well-known crypto founders, and they are currently in discussions with some USD funds.

The third team comes from a young serial entrepreneur who is himself a Polymarket user, with little crypto in his background. They named their company "Global Truth Machines" — reportedly as an homage to IBM (International Business Machines).

They believe that the PGC (professionally generated content) approach Polymarket adheres to is the superior prediction market operating model, because the long tail determines a platform's success or failure, and this precisely requires high-quality topics and refined operations. This company currently has support from multiple angel investors and has begun reaching out to early-stage funds.

What's interesting, according to our understanding, is that these teams have expressed similar views on certain occasions: the underlying architecture of prediction markets more closely resembles "Toutiao" (ByteDance's news aggregator), with one team even attempting to trademark "Mingri Toutiao" ("Tomorrow's Headlines"). One founder told me: "People have already forgotten who Toutiao disrupted. I believe prediction markets might make people forget Toutiao."

This is a story no one has told in a long time.

Prediction markets exhibit significant flywheel effects: when enough participants create sufficient liquidity, predictions converge closer to "truth," whose informational value attracts more people to use it for information, some portion of whom convert and join the market. This looks close to a "winner takes most" model — the kind of battlefield USD funds once knew best.

But here's the question: Polymarket and Kalshi's flywheels are already massive. Do Chinese-founded teams still have opportunity in global markets?

Polymarket and Kalshi's topics cluster around politics, tech, and sports, with demographics reflecting America's social baseline. But one narrative these companies share is that regional markets each have their own characteristics — Southeast Asian youth, for instance, don't care so much about US elections. The vast ocean of global pan-cultural traffic, fan sentiment, and long-tail controversies, plus the generation of young people beyond China and America behind these topics, still need a tradable vehicle.

This is why they all mention "content distribution."

A prediction market platform is essentially a funnel of "content-entertainment-trading," with topic-audience matching at its core — pushing topics to those most likely to bet. Click, convert, retain. Using the most classical product management skills to maximize platform efficiency overall. This is Chinese founders' zone of advantage.

Yet another interesting point: surrounded by AI and hardware as we are today, prediction markets feel like an anomaly of the AI entrepreneurship era — they don't seem to have much to do with AI. The frenzy around them stems more from the demonstration effect of Polymarket and Kalshi.

But in my research, I did discover an intriguing dimension.

Suppose prediction markets contain numerous trading agents — the market effectively constructs a reinforcement learning-like mechanism for them. The entire agent cluster continuously cycles through "gather information — weighted voting — reward/punishment," potentially converging infinitely on the true probability of an event occurring.

What it inputs: chaotic, noisy internet data. What it outputs: a clear, computable model of the future.

This could evolve into a kind of technological deification, a capacity to calculate randomness and foretell the future, a mathematically resurrected "Laplace's demon."

Pretty sci-fi, isn't it?

Note: Any projects, companies, or crypto assets mentioned or involved in this article do not constitute investment advice. The regulatory frameworks, market classifications, and legal environments discussed are based on regulatory systems in the US and other countries/regions, and are unrelated to legal environments in other jurisdictions.

Cover image source: Unsplash