Spend enough time in the subreddits where algorithmic bettors gather and a pattern shows up: people building their own live-game charts in Python and Matplotlib, then asking each other how to keep them running. Where to cache the data, how to stop the thing from breaking mid-game—the list goes on.
Frank Li says he and his ESX co-founders found a variation of that same problem across roughly 3,000 of the users they encountered—and that became the company's wedge.
ESX is a sports-only prediction market built around live in-game trading, with charting and technical tooling layered on top. Li is specific about who it's for. He describes a one-to-ten scale of bettor sophistication, with DraftKings, PrizePicks, and Underdog pulling the recreational end and exchanges like Novig competing on price at the sharp end. ESX is pointed toward the middle: bettors he calls semi-professional, who have been at it a while, are comfortable with data, and want to trade during the game rather than lock a position before it starts.
The template ESX is following isn't a betting company. It's Webull, which built a business on day traders while Robinhood took the mass market. Li's team, consisting of five engineers, took a similar approach, bypassing recreational users in favor of the technical ones.
Reaching those users has been the harder problem. Bettors don't label themselves the way engineers do on LinkedIn, so ESX went to Reddit, where co-founder Andrew Lee was already active. The two of them say they’ve had something like 10,000 conversations with users and prospective users since.
Most weren't useful in the obvious way. Li says the biggest thing he learned is that "getting the answer from our user is not a straight path." Ask someone directly what they think of the product and "they'll be like, 'Hey, the UI is cool,' and then it's useless conversation again." The stronger signal came from watching what people complained about to each other.
The same instinct drives an unusually heavy content operation for a five-person engineering team. Li's read is that the build itself isn't the moat, "it feels like there's less defensibility in technology and building," so distribution gets the attention. Across May and June, ESX pulled roughly 10M views on social, which Li estimates translated to around 100–200 organic users who actually traded in a month.
The commercial picture is early. ESX bootstrapped on about $100K of founder money, launched real-money trading roughly three months ago, and has been 10x'ing volume each month since, with a 50x increase last month. Around 1,000 signups so far. The platform runs on crypto rails and is closed to US users, with free-to-play access and API support serving as a backtesting on-ramp.
The company is now raising $2M to incentivize market makers, fund more content, and cover the legal work toward a US app. The nearer milestone is $1M in monthly volume, which Li wants to hit as NFL season arrives.
The longer version, as he put it: "When people bring up Kalshi and Polymarket, they'll also think of ESX just through our brand exposure."
Listen to the full conversation on YouTube, Spotify, and Apple Podcasts.
