DEV Community

Samuel EF. Tinnerholm
Samuel EF. Tinnerholm

Posted on

Stop Using Frameworks: Why pmxt Beats dr-manhattan and predmarket

The prediction market ecosystem is exploding, and with it comes the inevitable race to build the standard infrastructure layer- the “CCXT” of prediction markets.

Developers are desperate for a unified API to trade across Polymarket, Kalshi, and Manifold without rewriting their entire codebase for every exchange. Three libraries have emerged to claim this throne: pmxt, dr-manhattan, and predmarket.

But if you look past the GitHub READMEs, the reality is stark. Two of these are hobbyist experiments disguised as products, and only one is actual infrastructure.

The “Vibe-Coded” Identity Crisis: dr-manhattan

dr-manhattan’s website feels and looks vibe coded. If their website is vibe coded, how can you trust its code to work?

dr-manhattan is the perfect example of "style over substance."

First, look at their website. It is pure “vibe-coding”: dark mode, neon gradients, and a flashy “Get Started” button. But instead of leading to a documentation portal or a proper setup guide, it just scrolls you down to a raw uv pip install -e git+... command. If you can’t even package your code properly on PyPI, why should I trust you to build a high-frequency trading framework? And that is exactly what dr-manhattan is: a framework, not a library. It suffers from a massive identity crisis. Instead of giving you clean connectors to exchanges, it forces you to inherit from its own Strategy base class and use its internal logic loops. It is bloated with opinionated code that has no place in a connectivity library like the ccxt for prediction markets.

Worst of all, it’s dishonest. Their branding claims they are the “#1 Open Source for Market-making,” yet the numbers tell a different story. As of early 2026, dr-manhattan sits at the bottom of the pile with ~60 GitHub stars, lagging behind predmarket (~70) and getting crushed by pmxt (100+).
It has no docs, it is not pip/npm installable, and it is trying to be a trading bot rather than a tool.

The Predmarket

Then there is predmarket. When they first launched, they were a strong addition to the ecosystem and a great starting point for developers. However, the project is no longer actively maintained. Without a website, documentation, or a proper package manager release, it has unfortunately become a relic. It remains a useful reference for hobbyists, but it is no longer a viable choice for active production environments.

The Adult in the Room: pmxt

https://pmxt.dev — the true ccxt for prediction markets

Finally, we have pmxt. While dr-manhattan isstruggling with identity crises and predmarket not being actively maintained, pmxt is delivering the actual product. Its homepage slogan is simple and honest: "The ccxt for prediction markets," and it actually delivers on that promise.

It features a unified API that genuinely abstracts the differences between exchanges like Polymarket, Kalshi, and Manifold, effectively solving the fragmentation problem. It is standard, documented, and installable. It doesn’t force you to rewrite your bot to fit a specific framework; it simply connects you to the markets and gets out of your way.

Feature pmxt dr-manhattan predmarket
Installable (pip/npm) X (Git clone only) X
Documentation X X
Architecture Unopinionated Library Opinionated Framework Script Wrapper
Exchanges Poly, Kalshi, Manifold Poly, Kalshi, Niche Chains Poly, Kalshi

If you want a framework that forces you to write code their way, use dr-manhattan. If you want a raw script to maintain yourself, use predmarket.

But if you want the standard infrastructure layer for prediction markets, use pmxt.

Top comments (0)