DGPredict to DG3 Terminal | Here’s what led us here.

Ronit & Charan, DG3

It started in Dehradun, India.

Two poker players, a whiteboard, and a question we kept coming back to: how do we build something people already desperately want? We both knew gaming. We both knew markets. And when we looked at the prediction market landscape in early 2024, the answer felt obvious - liquidity and demand was there, but the experience was broken. Crypto sports trading was fragmented across venues with no clean way to move between them. So we decided to take a shot at it.

That became DGPredict. A simple aggregator that pulled liquidity from Azuro, Overtime, and SXbet and let users place trades across venues in one place. It wasn't glamorous, but it was useful and for a while, it felt enough.

What the data and traders told us

Over the next year, we did something that changed everything: we started talking to our heaviest users. Not surveys. Not analytics dashboards. Daily conversations with 20+ whale traders - people running $1M+ in monthly volume on prediction markets. We watched how they worked. We asked what frustrated them. We showed up in their DMs and Discord and we listened.

The pattern that emerged was almost universal. These traders were spending 20 to 30 minutes before every single trade scanning for the right market, cross-referencing news, pulling player histories, checking prices across multiple books, then finally executing. And here's what surprised us: roughly half of them had already started duct-taping LLMs into their workflow. They were doing it manually, on their own with no infrastructure built for it.

Charan had also lived this firsthand. Last hockey season, he identified a +6% edge on an NHL market. Eighteen minutes to confirm the thesis across five sources. By the time he moved, the line had shifted eighty basis points. He was right about the market and still lost. That's not a trading problem. That's a workflow problem.

The product we built as an aggregator inherited the problems of every venue it sat on top of. When they had instability, so did we. When their UX was broken, ours was too. We were seeing customer churn that we couldn't fix from the inside. So we both went back to the drawing board.

The real insight: the wedge isn't a better book

The prediction market industry grew up fast. In 2025, $44 billion was routed through these markets. Kalshi cleared CFTC. Robinhood and CME walked in. What was once a crypto-native niche became something closer to financial infrastructure.

But the tooling never caught up. The platforms traders were using looked exactly the same as they did two years ago - pretty charts, no context, no speed, no intelligence. The market had matured. But the products didn't.

That's the gap >| DG3 is built to fill.

A multi-agentic workflow terminal that reads the edge in under 30 seconds, with every claim sourced, and executes across four venues in under 100 ms. The trader's only job becomes being right about markets. Everything else is ours to handle.

What >| DG3 actually does

We built DG3 around four architectural decisions that everything else flows from: an event-first data hierarchy where research connects to real-world events, not just contracts; an agentic RAG system with source playbooks tuned for each category of market; a Rust execution layer that delivers sub-100ms signed fills; and DG3's own proprietary data CLV, sharp consensus treated as first-class signals from day one.

The three features we're most excited for traders to try:

Agentic Search in Discover.
Type something like "find me high-edge UCL markets this week" and DG3 surfaces the seven that actually matter, ranked by edge × velocity in under twenty seconds. No keyword matching. No tab-switching. Semantic search with sports context built in.

Edge Finder.
A live ranked signal feed across 1,000+ markets, with one-click execution in under 100ms. Every position ranked by CLV - the only honest scoreboard in prediction markets, and the one metric that separates sharps from everyone else.

Write to trade.
You type your strategy in plain English and the terminal converts it into code for you to trade.

A non-trader founder would have built dashboards. We built workflows. That difference shows up everywhere in how >|DG3 is scored, how it surfaces signals, and how it executes. Because we've both sat on the other side of that trade, watching a good call expire because the tooling was too slow.

Who we're building this for

Outsiders call prediction markets gambling. Insiders know they're priced like commodities. CLV compounds, win-rate doesn't. A casual trader chases outcomes; the sharp chases the side that disagrees with the close, every position, and lets variance do the rest. Prediction markets are becoming a truth-seeking machine - when the liquidity is there and the market matures, the price of an event becomes one of the most reliable signals in existence. We're still early.

The first 100 traders we're looking for aren't tourists. They're people already active on Polymarket who hate the meta-work more than anything - running $1K to $50K per position, five or more positions a week, with real, researched opinions on their markets. Not vibes. Not hunches. Actual edge.

What they get isn't just early access. It's a direct line to us, features shipped weekly from what they actually ask for, Sharp Score visibility from trade one, and the chance to help us build the default operating system for sports prediction trading, together.

The only benchmark we care about

A year from now, our measure of success isn't a user count. It's whether any serious sports prediction trader runs a position without >| DG3 in the loop. And if so, why?

We're two days away from finding out.

If that's you or if you know that trader, the door opens June 3.

~ Ronit & Charan