Devin's Trading Desk: Forex, Futures, Crypto, Stocks & Options
A hands-on build day for people who want to actually build a trading agent, not buy one.
We build the whole thing in the room, together: a data feed, a rule-based strategy, a risk engine, and an LLM layer that reasons about a trade without ever being allowed to place one. Python, Deep Agents and LangGraph, paper-trading against a free OANDA practice account. You leave with a working agent skeleton you built yourself, on your own laptop.
The same skeleton runs on any of five markets — forex, futures, crypto, equities and options — so build against whichever one you actually trade. No ML experience required. You need to be able to read Python and run a command.
This is not a signals group, a course, or a pitch. Nothing said here is investment advice, and nobody is selling a system.
Run of show
12:00 — Doors, coffee, find a seat
12:20 — Welcome: the one rule of the day. The model reasons, the code owns the money
12:30 — Build one: data in, signals out. Clone the repo, wire the OANDA practice feed, get one signal printing on screen
2:00 — Break
2:15 — Build two: the risk engine. Position sizing, the daily loss halt, the kill switch
3:15 — Build three: the LLM gate. An orchestrator, three analyst agents, and a gate that can only veto, never resize
4:00 — Advertised end. We have the room until six
4:15 — Demos: show what you built. A working veto makes a better story than a working entry
5:00 — Open floor and where to go next
What to bring: a laptop, Python 3.11 or newer, and a free OANDA practice account. The repo is public, so clone it before you arrive if you want a head start.
Who this is for: developers, traders who can code, students, and anyone tired of being sold a bot.
Format: a full afternoon of hands-on building. Bring a laptop and leave with something that runs.