At SGV, we’ve seen a handful of upstart prediction market players introducing novel mechanisms and add ons to the already bleeding edge market that is only growing in the past few years. Before getting into the innovations, let’s start with the incumbents.
The two incumbents are Polymarket and Kalshi, which represent different approaches. Polymarket has an advantage in volume, given they offer no trading fees, but Kalshi is seeing growth in top markets as well, especially sports. First, we'll dive deeper into them, and then into more nascent projects.
Data comp of last 12 months of Polymarket vs. Kalshi:
Valuation: $1B vs. $2B
Est. Revenue (on volume): $240M vs. $168M
YoY Growth: 3× vs. 30×
Polymarket
Polymarket is a decentralized, crypto-based prediction market launched in 2020. It runs on blockchain order book (using USDC stablecoin for wagers) and lets users trade on event outcomes (e.g. elections, COVID case counts, crypto events).
Per the seed investment memo from 1confirmation (here), the main appeal of Polymarket was the market resolution mechanism it had, and the product ethos of being as user friendly as possible, simple things like allowing USDC as the main unit of account allowed them to devour the market.
Polymarket operates globally (though geofenced from U.S. users after regulatory scrutiny) and has gained popularity for markets on real-world news. However, Polymarket is not a fully open marketplace, users cannot create markets directly. The site’s team curates and creates markets, taking community suggestions but ultimately deciding what gets listed. This gating of market creation is intentional: it ensures markets are well-defined and resolvable, and helps Polymarket avoid scams or opaque bets. (It’s also partly a legal precaution, in early 2022 Polymarket was fined $1.4M by the CFTC for offering unregistered event-based swaps, and was forced to wind down certain markets. Since then they’ve been careful to comply and not list problematic markets.)
How resolutions work: Polymarket uses a decentralized oracle mechanism for settling markets. Specifically, outcomes are determined by the UMA Optimistic Oracle, a smart contract system where a proposer submits the result and posts a bond, and if no one disputes it within a challenge window, that outcome is accepted. If it’s disputed, a resolution vote occurs via UMA’s token holders. In theory this allows trust-minimized resolution, though in practice it’s seen some drama. For example, in 2025 a Polymarket question about a Trump-related “Ukraine mineral deal” saw a malicious oracle attack, a user with a large UMA stake manipulated the vote to prematurely settle the market incorrectly, a $7M debacle that highlighted oracle governance risks. Polymarket has been working to strengthen its oracle and dispute processes to prevent such incidents. On the whole, though, most markets resolve smoothly according to clear criteria set upfront. Reality is, even if the Polymarket team had an internal resolution system, problems with the markets are along defining the criteria for resolution, especially on subjective manners.
Data shows retention curves are only getting better and better as the product grows. This is a net indicator that the product experience is achieving better levels.
User experience and community: Polymarket’s interface feels like a prediction trading app, showing odds that update in real time as people buy and sell shares. One notable addition has been social features. In 2023 Polymarket introduced comment sections on markets, and these have become extremely active on high-interest questions. Users discuss their reasoning, drop news links, or trash-talk and meme about the event. Interestingly, the site displays each commenter’s position size, injecting accountability. This social layer has driven engagement and time-on-site, suggesting that prediction markets double as discussion forums for those invested in the outcome. It’s a convergence of trading and social media that may become standard in this space.
Kalshi
Kalshi takes a very different route: it’s a fully regulated U.S. prediction market exchange. Kalshi became the first event-contract platform designated as a CFTC-regulated exchange (Designated Contract Market) in 2021. This means Kalshi can legally offer binary event contracts in the U.S., but only under the confines of what regulators approve. Every market on Kalshi must fit CFTC guidelines. As a result, Kalshi’s markets skew toward economic and political events: Fed interest rate decisions, inflation rates, job numbers, election results, etc. They also got approval and started trading in sports related markets and have seen great traction. That said, you won’t find celebrity gossip or crypto meme bets on Kalshi. Kalshi has to get regulatory approval for new categorie
Market creation is entirely in-house, Kalshi’s team (with regulatory oversight) lists the contracts, and users trade them. This centralized listing approach is necessary for compliance (no user-generated markets), and Kalshi even has a rulebook filed with the CFTC detailing how contracts are defined and the position limits, etc
How resolutions work: Because it’s regulated, Kalshi has very clear resolution criteria for each market. Every market page includes a “Rules Summary” that spells out exactly what is being predicted, the timeframe, and the verification source for the outcome. For example, a weather market will cite the official NOAA weather station data as the source, a financial market will cite the government’s published statistics, an election market might cite AP or an official certification. The outcome is determined by that source at the end date, there’s no community voting on outcome; it’s an objective resolution by reference to authoritative data. In short, Kalshi acts much like a traditional futures exchange: strict rules, formal processes, and if there’s ever a dispute, presumably Kalshi and the CFTC have mechanisms to adjudicate (thus far, major disputes have been rare). All contracts settle in cash (USD) with the typical binary $1 payout for a correct bet.
“Robinhood users traded around $1 billion worth of event contracts on Kalshi markets during the second quarter of the year, the company’s chief financial officer said, likely generating around $10 million in revenue for the business.
In an earnings call following the stock trading giant’s Q2 results Wednesday, Jason Warnick spoke of the company’s success in integrating prediction markets into its offering. The business does not host its own exchange but instead allows its customers access to Kalshi’s markets, charging an additional fee for them to do so.” (source)
Differences in userbase and activity
Kalshi and Polymarket cater to somewhat different audiences. Polymarket, being crypto-based and (until recently) unofficial, attracted a more global, crypto-savvy crowd; many users treat it as both entertainment and investing, and its volumes have spiked during big news cycles.
For example, ahead of the 2024 election Polymarket hit a record ~$90 million in open interest across its markets, reflecting hundreds of millions in total bets over time, a sign of “blisteringly popular” interest in political betting on crypto platforms. Kalshi, meanwhile, as a regulated platform, has had to grow more slowly within U.S. legal constraints.
One notable development is Kalshi’s partnership with Elon Musk’s new AI company, xAI. In mid-2025, Kalshi announced a deal to integrate xAI’s “Grok” AI chatbot into the platform to help analyze real-world events for traders. The idea is that Grok can summarize news, sift data, and provide insights in real-time to Kalshi users as they bet on things like Fed decisions or elections.
Interestingly, Musk’s X had also partnered with Polymarket (as the “official prediction market” for Twitter/X) earlier in the year, effectively bringing assistance to both an unregulated and a regulated market. This suggests a future where AI and prediction markets intersect, with AI helping traders digest information overflow. (It’s also a tacit nod that Musk sees prediction markets as valuable “truth-seeking” tools, worth integrating into social media and AI efforts.)
Market Gating
Both platforms currently centralize the market creation process, Polymarket explicitly says markets are created by its team (with user input), and Kalshi as an exchange must list allowed contracts only.
The core reasons are quality control and legal compliance. An open marketplace where anyone can create a prediction market is powerful, but it runs into issues: How do you ensure the question is well-defined and will have an unambiguous answer? Who will provide the data or decision on the outcome? What if someone creates a market on an illegal or unethical topic (e.g. assassination markets or insider info)? For Polymarket, which operates in a legal gray zone, it’s safer to have internal curation so that every market has a clear resolution source and doesn’t inadvertently break laws or moral norms. They also manage the oracle and dispute process, so having fewer, well-structured markets makes that manageable.
For Kalshi, the gating is mandated, under U.S. law, only approved contracts can be offered, and adding new categories requires regulatory review.
In summary, while crypto ethos favors permissionless systems, in practice current prediction market leaders have found it necessary to curate markets to maintain integrity (and stay out of trouble). Later in this piece we’ll discuss how future models might enable more open market creation without sacrificing trustworthy resolution.
Categories
Both Kalshi and Polymarket are prediction markets that emphasize the user experience on non-sports events, assuming that it will be a growing category, with political events being the most influential, but Sportsbooks was and still is the most profitable category for pms markets all round, and Kalshi seems to be ahead of Polymarket his this realm.
This piece will ignore sports for the most part, as it deems a separate piece altogether.
Beyond Polymarket, a new wave of prediction market platforms has been emerging, particularly on modern blockchain networks. These projects aim to solve pain points of earlier PMs and tap into niches that remain underserved.
Limitless (on Base): Launched in 2023, Limitless is a crypto-native prediction market focused on short-term asset prices. In essence, it lets users bet on very near-term moves in crypto markets: “Will Bitcoin close above
today?” or even hourly resolution markets. These are akin to ultra-short-duration options (so-called binary options in traditional markets), but packaged in a user-friendly way. Within a few months of launch, it facilitated over $250 million in volume on Base. Limitless emphasizes speed and simplicity: they even introduced a mobile-first trading UI so people worldwide can “seamlessly wager on their favorite asset’s performance in the next hour or day”. This mobile UX, plus fast resolution and settlement (winners paid out every hour/day), aims to make trading feel more like a game you can play anytime.
Onit (on Base): Onit brands itself as “the distributed prediction market protocol” that allows you to deploy fully customizable markets for any question, and it integrates into chat apps and social contexts. Onit’s big idea is to bring prediction markets “right into the group chat”. Imagine you and friends are debating something in a chat, with Onit, a bot could let you instantly create a private market in that chat and put some stakes on the line. Only the people in that group can see and bet on that market, making it a fun social wager. Onit targets a more casual, socially driven audience. Another innovation with Onit is support for distributional outcomes, as hinted by the team, they allow “bell curves on outcomes” and not just binary yes/no bets. This suggests Onit can facilitate prediction markets on numeric outcomes (e.g. “What will the temperature be on Saturday?” or “How many users will this app have next month?”) where users effectively bet on a range or a full probability distribution. Traditional PMs can do this by having many discrete strikes, but Onit appears to provide a more seamless way to express a probability curve (perhaps using automated market makers that cover a continuous range). Such distributional prediction markets are powerful; they generate a full probability distribution for an event, not just an odds of yes/no. It’s like turning a prediction market into a forecasting poll with a payout incentive.
Hedgehog Markets (on Solana): A “no-loss” prediction market. Hedgehog uses DeFi yield to enable no-loss betting, users’ stake is placed into yield-generating pools, and only the yield is paid out as winnings, so losing bettors keep their principal (hence “no loss”). This gamified approach makes participation feel like a free lottery, you risk only the interest on your funds. Markets are typically binary (win vs lose), ranging from sports to crypto events. Market creation is semi-permissionless, communities or the team spin up markets, often with sponsorships. It touts instant settlement and no custody delays, thanks to Solana’s speed. Hedgehog also built tooling for anyone to create their own markets easily.
Inertia (on Avalanche): The team of Inertia are approaching a community first approach, where communities are creating markets on top of their interests, and the UI is much more like reddit, than a prediction market. This is different from the main prediction markets based on price movements. They are also partnering with esports players to do predictions on these types of games, and have a wedge there.
There’s a ton more prediction markets, some include Morpher, across all categories. BRKT, a generalist platform as well, focused on bracket style markets, and more. Here’s a list of some players.
Prediction markets sit at the intersection of finance, information aggregation, and social behavior. It’s no surprise that they inspire a lot of adjacent ideas and novel applications. Here are some of the key emerging concepts related to prediction markets:
Futarchy (Prediction Markets for Governance): “Vote on values, but bet on beliefs.” This phrase by economist Robin Hanson encapsulates futarchy, a form of governance where policy decisions are made by prediction markets rather than by direct vote. In a futarchy, a DAO or government sets a metric for success (e.g. GDP growth, token price, etc. – that’s the “value” to maximize), and then people trade in markets that predict that value under different proposals. The idea is the market’s collective wisdom will determine which proposal would lead to a better outcome, and that proposal is then adopted. While futarchy remained theoretical for years, we’re now seeing real experiments. MetaDAO on Solana is a live example: it’s a DAO that raised $2.2M led by Paradigm to implement futarchy in its governance. Whenever MetaDAO has a proposal, it creates two markets on its token (META), one market speculating on META’s price if the proposal passes, and one if it fails. After some time, if the “pass” market price is higher, that implies traders think the proposal will increase META’s value, so the proposal is automatically adopted. If the “fail” market has a higher price, the proposal is rejected. In July 2024, MetaDAO actually ran such a vote to decide on raising funds for the project; traders signaled confidence (META price was bid up to $435 in the “pass” market vs $386 in “fail”), so the proposal was executed. This is futarchy in action. It shifts decision-making from token-holder ballots to open markets where anyone can back their opinion with money. Proponents argue this yields wiser decisions, as “market participants are driven to make better choices than politicians”, filtering out noise and bias. Futarchy is still experimental (even MetaDAO calls its admins “chief futards” tongue-in-cheek), but if it succeeds, it could transform how DAOs (or even governments) make choices – essentially using prediction markets as a “truth engine” for governance. It’s a perfect illustration of the public-good aspect of PMs: they don’t just predict, they can decide. MetaDAO is focusing on launching DAOs that adopt the model to its full extent.
“Multiverse Finance” and Conditional Markets in DeFi: One limitation of current prediction markets is that the outcome tokens (the shares you buy for Yes or No) aren’t very usable beyond the bet you hold until resolution for a payoff. And you can’t, say, use a Yes-share as collateral easily, because if the outcome goes against you, that share becomes worthless and any loan against it would default instantly. In May 2025, Paradigm’s Dave White introduced “Multiverse Finance,” a framework that creates parallel on-chain universes for each possible outcome of an event. For example, consider an event: “Will candidate X win the election?”
Prediction markets are a rich design space, essentially any scenario with uncertainty can potentially be turned into a market.
We’re seeing them be used for governance (futarchy), for fun social games (Ponder), for trading and hedging (crypto markets, sports, etc.), and as building blocks in DeFi (conditional assets and multiverses).
Finally, as a team that actively considers investments in this sector, it’s worth outlining what it takes for us to back a prediction market.
Here are the key criteria and thoughts we have when evaluating PMs:
Appeal to “Prosumers”: While broad consumer appeal is important for user count, we also know that in prediction markets (and exchanges generally), a small percentage of users often drive the majority of volume. These are the “prosumers”, semi-pro bettors, highly engaged traders, or subject-matter experts who trade frequently. They provide liquidity and depth to markets, making the platform viable. We are interested in platforms that can attract and retain these power users as well. This might be through offering advanced features (e.g. APIs or advanced order types for serious traders, higher betting limits, or integration with external prediction communities like Metaculus), or through focusing on markets that pros care about. The challenge is to serve pros without scaring off casuals. The best platforms might segment the experience, a simple interface for newbies and a more advanced dashboard for heavy traders.
Focus on High-Interest Markets: We tend to favor a focused approach over a platform that tries to offer every possible market from day one. The reality is, just as in sports betting, a handful of markets will drive the majority of engagement. We suspect the same for prediction markets: certain categories will dominate, likely politics, major news events, crypto asset prices, maybe macro-economic indicators, maybe popular culture events. A startup would do well to identify where there is a natural hunger for wagering coupled with information asymmetry that markets can exploit. For instance, Polymarket found traction with political and COVID markets when those were hot topics in the news cycle, and Limitless zeroed in on crypto traders who already like short-term speculation. A common pitfall is spreading too thin and competing on larger markets, instead, we look for founder insight into which vertical or community is the beachhead, and then dominate that. The platform should become the best place in the world to trade X, where X might be one category. Once you own a vertical, you can expand to adjacent ones. This also ties to marketing: it’s easier to acquire users when you’re synonymous with a particular use-case.
Category-Leading UX & Features: Related to the above, the product needs to have best-in-class UX for the markets it serves. If it’s a sports prediction app, it should have the polish and real-time feel of top sports betting apps. Similarly, if it’s a political markets platform, integrating things like poll data, live election results, or expert commentary could greatly enrich the experience. We also note integration with content, a prediction market app could embed a live news feed or a Twitter feed for the topic so users stay in-app to get information relevant to their bets.
In conclusion, our checklist boils down to: great UX, liquidity & focus, community, a plan for openness at scale, and A+ founders who get prediction markets. If those align, we’re very excited to back a PM startup. The upside is not just a big business, but also potentially a product that changes how millions of people bet, invest, and even make collective decisions.
If you’re building here, let’s chat.
Useful links
https://medium.com/collab-currency/futarchy-for-both-ends-of-the-curve-cb730db048ce
https://www.onchaintimes.com/a-chat-with-domer-the-1-trader-on-polymarket/
The Landscape of Prediction Markets & Betting Platforms
Social Graph Ventures
Share Dialog
“Beauty Contest” Social Prediction Games: Not all prediction markets are about objective truth; some are about second-order predictions, predicting what others will predict. This is the idea of the Keynesian beauty contest, and it’s being revived in Web3 social apps. A prime example is Ponder, a mini-app on the Farcaster social network (disclosure: it’s one of our portfolio projects). Ponder is a daily prediction game built on the beauty-contest concept. Each day, it poses a question (often lighthearted or opinion-based, like “Which movie is more iconic?” or “Is a hot dog a sandwich?”). Players can stake a small amount (as low as $0.50) on the option they think the majority of other players will choose. The key is, you’re not voting your personal opinion, you’re trying to anticipate the crowd’s collective opinion. When the poll closes, whichever option got the most votes is the “winning” outcome, and everyone who staked on that shares the prize pool (proportionate to their stake). In effect, it’s a prediction market where the event is “What will the crowd think?”. Success relies on reading the collective sentiment, not luck. Votes are hidden until the poll closes to prevent herd behavior – so you truly have to predict without knowing interim results. The vibe is fun and social, since it lives inside Farcaster, people often discuss the questions, tease each other, and keep streaks. This kind of social betting game might not surface deep truths about the external world, but it’s highly engaging and can onboard users to the idea of prediction markets in a friendly way.
“Information” markets: Jokerace (portfolio company) is pioneering this category, with different iterations, and lastly vote2earn. The mechanism is simple: brands run contests with two periods, an entry period, where entrants can submit their entries, and a voting period, where voters can pay to vote for the best entries. The main use of Jokerace so far has been for community voting, similar to proposals. The largest differentiator is that within Jokerace, the voters are incentivized to vote as early as possible, where votes are cheap at the start, and get more expensive, and if they vote for the winner, meaning, they forecast the winner correctly, then they earn a share of the prize pool. Here’s a dune with some metrics.
Social and Community Engagement: We strongly believe in the power of community around prediction markets. By nature, PMs tap into topics people have opinions on. There’s built-in social energy. Platforms that harness this will have better retention. As seen on Polymarket, once they added a comment section, it became a core engagement driver, users return not just to check odds but to argue and banter. We think future PM apps can go further: imagine features like leaderboards, follower systems, communities. We also think about multimedia: could there be live streams tied to markets? Perhaps a streamer discusses the presidential debate while a prediction market on “Who will win?”
Scalability and Open Market Creation: As discussed, current platforms often restrict user-created markets due to resolution and legal issues. However, we see a lot of value in permissionless market creation if it can be done right. A truly open prediction market protocol (think Uniswap for markets) could unlock long-tail topics and innovation. The question is how to ensure honest resolution and avoid ambiguity in a decentralized way. This might involve decentralized oracles, or new mechanisms like retroactive verification or reputation staked by market creators. We are interested in teams that have novel ideas here. Even a hybrid approach (permissionless creation but with a moderation layer that can veto truly bad markets) might work. The goal is to eventually allow the scale that crypto can enable: thousands of niche markets serving many communities, without everything being bottlenecked by one team’s bandwidth.
Parlays: Even though this is mainstream on sports betting, it’s not yet seen on prediction markets. Parlays fragment liquidity, and calculating odds becomes hard. Strategies like trading correlation markets might help with this, but acknowledging this as a market opportunity is something we value.
Founder-Market Fit: Last but not least, an often underrated factor is the team’s understanding of the space, both its passion and its handling of non-product challenges like regulation. The best founders in this space are usually obsessed with prediction markets themselves.
427
