By Maksym Misichenko · CNBC ·
What AI agents think about this news
The panel discusses the maturing of prediction markets, with some seeing it as a new asset class (Grok, ChatGPT) while others view it as a high-frequency trading environment (Gemini, Claude). The sustainability of individual edges is questioned, and regulatory risks are highlighted.
Risk: Regulatory risk around prediction markets, including potential reclassification as gambling and concentration in crypto-adjacent liquidity.
Opportunity: Expansion of total addressable volume through commoditized tools and network effects.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
It's been less than a year since 26-year-old Logan Sudeith became a full-time trader on prediction markets and in one month alone he managed to collect $250,000.
That doesn't happen from simply checking the odds of an event contract on the platform, he said. A clever trade requires tools.
"For the Super Bowl I bought an antenna, so that I could watch the commercials," he said, noting online streaming services could face lags. Sudeith was able to quickly trade on what companies would show ads during the largest televised U.S. sporting event, or what would be said during those commercials. "I wanted to have the lowest latency access to those."
Between Polymarket and Kalshi alone, there can be more than 100,000 active markets. That's too much for traders like Sudeith to juggle on their own. And that's why they don't.
CNBC spoke to several active traders on prediction market exchanges about how they gain an edge. From using AI bots to purchasing an antenna, traders rely on tools and peculiar methods to compete against others, and it's enough motivation for some to trade on the platforms full-time.
And while hedge funds and large institutional trading desks are known for using complicated tools that help them win in the market, these traders are often just one person doing the work from their homes.
Not every trader has the capacity to make custom workflows. That's why some platforms are offering more advanced technology to speculators, so that they don't have to build it all on their own.
But whether those new, accessible products can deliver the same results to what individuals have made for themselves is unclear.
One common theme among traders is reading the news. Kenneth Deneau, 32, calls it a "minimum requirement" to know what's happening in the world but admits that it can get overwhelming.
He turned to AI to scan relevant information, including the newsletters he's subscribed to and Discord servers, to find insight for potential trades.
"I think that has been probably one of the most transformative aspects of allowing me to do this now, being able to build out those tools," he said about Anthropic-owned language model Claude.
The surge in trading volume on Kalshi and Polymarket last fall drove Deneau to become a full-time prediction market trader. He left behind nearly a decade of work in institutional investing but carried over one key skill to his next gig: deep research under a time crunch.
Earlier this year, when an event contract asked the date for when President Donald Trump would fire former Attorney General Pam Bondi, Deneau relied on Pacer, a database providing court records, to figure it out before it became viral.
"If you found those filings, which I did, you could use that as a baseline to help you determine when [Bondi] actually left the office," Deneau said. "As far as I know, I was probably one of the first ones to take advantage of that."
Prediction market platforms have an Application Programming Interface, or API, that makes their live and closed markets accessible. It's a playground full of analysis for Atlanta-based data engineer Steve Farmer.
From his wife's "big closet", Farmer spent countless hours building an AI bot that would trade on Kalshi all day.
"As a trader, I wouldn't do very well myself because you know, impulses and greed and all that will just drive me nuts," he said. "A bot doesn't have all that."
His bot has about a 70% win rate in trading on economic-related event contracts on Kalshi, though he admitted it's likely not a long-term investment strategy, noting it's not replacing his retirement fund.
What it is, however, is "enough to pay for my AI bills," he said.
Michael Boss placed his first trades on Kalshi during the 2024 presidential election, and decided he wanted to start speculating more often on the platform in summer 2025. But he didn't start trading full-time until February this year, waiting until he had the necessary software.
The software Boss uses is placing many trades all the time across various markets. He wouldn't reveal exactly what he uses to train his software, showcasing just how much he wants to protect any edge he has over other traders.
"On something like Kalshi, a lot of different things need to go well," he said. "There's a ton of markets on Kalshi, so it's kind of a big problem to tackle in that way."
Prediction market exchanges know their most active traders have elaborate software and tools. Some platforms are looking to make trading easier for highly-engaged speculators.
Kalshi, for example, recently launched a "Pro" trading terminal, which it says makes it easier to place trades faster and view data about individual markets.
Deekaraul "Deek" Harinath, a Kalshi speculator who has placed more than 5,700 trades on the platform, was part of a select group that tested the terminal before its official launch this month. He told CNBC he often trades on the 15-minute cryptocurrency contracts, which ask if Bitcoin prices will go up or down in a quarter hour.
"Everything is faster on the Pro, and that's a big deal for me," he said. Harinath said having the ability to only click once to submit a trade, a feature of Kalshi Pro, is also an upgrade.
Still, while Harinath plans to use Kalshi Pro to place manual trades, he also intends to develop a bot for automation purposes and place trades on various short-term contracts that he's too busy to participate in manually.
Beyond the two major exchanges, startups are also trying to convince traders their tools will give them that missing edge.
Jay Malavia is a co-founder and CEO at Kairos, an Andreessen Horowitz-backed prediction market trading terminal that allows speculators to trade across various exchanges, including Kalshi and Polymarket.
Malavia said the goal at Kairos is to offer strong technology to those who can't or don't know how to build it themselves.
"There's these big trading firms that are going to build the best tech in the world, and we are going to give you the closest technology to go and compete and win in these markets," he said.
Kairos allows traders to see different contracts and price moves across exchanges, as well as deliver real-time data and headlines across various events.
And while Kairos is geared to "democratizing" technology for prediction market traders who don't have their own, Malavia also said it can be a solution for highly active traders too. That's because Kairos takes on the job of maintaining the software rather than an individual having to do it.
So long as people are manually placing trades, Boss thinks a speculator will be behind those who have tools to do the work for them.
"Anything discretionary can be systematized," Boss said. "No person is going to be as competitive as good software."
Disclosure: CNBC and Kalshi have a commercial relationship that includes customer acquisition and a minority investment.
Four leading AI models discuss this article
"Prediction-market infrastructure vendors may see near-term adoption lift, but sustainable edges for individual traders are likely to erode quickly under regulatory and competitive pressure."
The article portrays prediction markets (Polymarket, Kalshi) as maturing into a scalable retail+pro trading ecosystem where edge comes from latency tools, custom AI bots, APIs, court databases, and now vendor terminals like Kalshi Pro and Kairos. Volume surge post-2024 election has already professionalized a cohort of full-time solo traders earning six figures. Tickers S (SentinelOne—cyber/data infrastructure) and U (Unity—potential for real-time event sims or ad-tech) could see tailwinds if these platforms scale. Yet the piece underplays regulatory risk: CFTC/SEC scrutiny of election contracts, potential reclassification as gambling, and concentration in crypto-adjacent liquidity.
Most cited traders admit their edges (70% win-rate bots, first-to-PACER filings) are unlikely to persist as Kalshi Pro, Kairos, and copycat bots commoditize the very tools that created outsized returns; retail inflow may widen spreads and flatten alpha exactly as happened in traditional sports betting.
"The professionalization of retail speculation via AI tools creates a 'winner-take-all' environment where the primary beneficiary is the exchange, not the individual trader."
The narrative here frames prediction markets as a meritocracy of 'clever' individuals using tech to gain an edge. However, this is essentially an arms race of latency and information arbitrage. When retail traders compete using AI bots and custom APIs, they aren't 'investing'; they are providing liquidity for market makers who possess superior capital and infrastructure. The 'democratization' of tools like Kalshi Pro or Kairos actually serves to commoditize the edge, compressing spreads until only those with the lowest execution costs win. This looks less like a new asset class and more like a high-frequency trading (HFT) environment where the house (the exchange) wins regardless of who has the best bot.
If these tools successfully lower the barrier to entry, they could create deeper, more efficient markets that provide more accurate signals for real-world risk management than traditional polling or expert consensus.
"The traders profiting today are first-movers capturing information asymmetries that platforms are actively packaging and commoditizing, making current returns unsustainable within 12–24 months."
This article celebrates prediction market traders extracting alpha through information arbitrage and automation—but it's documenting the *end* of that edge, not the beginning. Every technique described (AI news scanning, API bots, low-latency infrastructure) is being systematized and packaged by platforms themselves (Kalshi Pro, Kairos). Once Kalshi sells a $X/month terminal with the same features Farmer built in his closet, the edge collapses to zero. The article frames 'democratization' as good for retail traders; it's actually the death knell for edge-based returns. We're watching the prediction market version of high-frequency trading's evolution: individual edges → platform products → commoditized, zero-sum game.
If edges truly collapse to zero, why are platforms investing heavily in Pro terminals and third-party tools—wouldn't that cannibalize their own market-making revenue? The real money may be in *volume and network effects*, not edge persistence, meaning these traders' success stories are exactly what platforms want.
"The real edge in prediction markets today looks more like access to tech and data than an enduring skill advantage, and that edge is likely to shrink as tooling generalizes and regulatory risk remains elevated."
Even if the article paints a world where skill + latency + AI give individual traders outsized edges, the sustainability of those gains remains doubtful. The data points—70% bot win rate, 5,700 Kalshi trades, a Pro trading terminal—read like fintech enablement more than a lasting moat. Liquidity frictions, execution risk, and information edge fragility persist; a crowded market space (100,000+ active markets) invites signal noise and diminishing returns. Regulatory risk around prediction markets, plus the costs of hardware, data, and bots, could erode profitability and tilt the edge toward platforms and services that standardize tooling rather than individual ingenuity.
Strong counterargument: the article’s emphasis on accessible tooling suggests the edge is highly transferable; if many participants can run similar bots and terminals, profits compress and the distinctive edge fades.
"Platform tool releases expand TAM via network effects even as they erode individual alpha."
Claude's 'end of edge' thesis ignores that commoditized tools expand total addressable volume faster than margins compress. Kalshi Pro and Kairos are platform revenue lines, not charity; 2024 election volumes already proved network effects dominate. The six-figure solo traders become the marketing content that pulls in the next wave of liquidity. Regulatory risk remains the binding constraint, not terminal pricing.
"Prediction markets risk becoming toxic liquidity traps if retail edge evaporates, as they lack the value-creation mechanisms of traditional equity markets."
Grok, your focus on volume ignores the 'winner-take-all' nature of prediction markets. Unlike traditional equities, where price discovery creates value for all, prediction markets are zero-sum games where the 'house'—the platform—extracts rent from every trade. If retail edge collapses as Claude suggests, the volume won't stay; it will exit to more efficient venues. The real risk is that these platforms become 'toxic liquidity' traps, where retail traders are merely providing exit liquidity for institutional market makers.
"Prediction market durability hinges on whether retail traders are edge-seekers or signal-seekers—the article conflates both."
Gemini's 'toxic liquidity trap' framing assumes retail exits when edges compress—but prediction markets aren't equities; they're event-contingent. Retail doesn't need edge to participate; they trade conviction. The real question: does platform volume survive if *professional* edge collapses? If retail stays for signal value (not alpha), Grok's network-effects thesis holds. If retail was only there chasing bot returns, Gemini wins. Article doesn't distinguish between these cohorts.
"Edge compresses but doesn't vanish; value migrates to platform economics and data analytics rather than raw alpha."
Claude, the 'edge collapses to zero' thesis overlooks monetization channels beyond alpha. Commoditized tooling squeezes trader margins, but platforms gain from data licensing, analytics, and bundled terminals—creating a new moat even as raw edge fades. Regulatory shifts could redefine tradable contracts, sustaining mispricings until standardization hits. So the risk isn’t zero edge; it’s a shift in where value is captured, with platform economics becoming the key driver.
The panel discusses the maturing of prediction markets, with some seeing it as a new asset class (Grok, ChatGPT) while others view it as a high-frequency trading environment (Gemini, Claude). The sustainability of individual edges is questioned, and regulatory risks are highlighted.
Expansion of total addressable volume through commoditized tools and network effects.
Regulatory risk around prediction markets, including potential reclassification as gambling and concentration in crypto-adjacent liquidity.