Indicators run first
Trend, regime, RSI, MACD, Bollinger Bands — the structured setup is calculated before any model sees the data. The AI reads context, not raw price.
Who made this?
A developer who got curious, then built a lab
Hugo's Trading Lab exists because I wanted to understand how to actually make trading decisions — what goes into them, what the science is, and where the guesswork starts. Then use AI to make those decisions for me, armed with all of it.
Where this started
Curiosity, not a trading background
I'm Hugo Bento, a software developer based between Spain and Portugal with over 15 years of experience building digital products. In late 2025, I got curious about crypto trading — not the speculative kind of curious, but the engineering kind: what actually goes into a trading decision?
I had no finance background. I didn't know what separated a disciplined system from a hunch. I wanted to understand indicators — what RSI, MACD, Bollinger Bands, and trend signals actually measure, what they're reliable for, and where they stop being useful. The things that can be calculated exactly, I wanted calculated exactly.
But I also knew that turning all of that into a buy or sell call — weighing the signals against each other, accounting for current exposure, deciding whether the setup is strong enough to act — was the part I couldn't do alone. That's where AI comes in. Not to go wild with it. To make the call with all the information in front of it that I don't have the expertise to interpret.
So I built a system that does exactly that: calculate everything that can be calculated, then hand it to an AI with clear rules and let it decide — and show its work every time.
What got built
A system that shows the work
The lab runs a 10-coin universe on Binance paper trading. Every hour, it reads indicators, passes a structured market snapshot to an AI model, constrains the output against current portfolio exposure and cash limits, and either places an order or holds. Every decision — including the holds — gets logged.
The portfolio is public. The orders are public. The reasoning behind each one is attached. If the AI gets something wrong, it shows up in the ledger.
"The point is not to project omniscience. The point is to make the system inspectable."
This is learning in public. The lab does not claim to be profitable yet. It claims to be honest about what it is doing and why.
The approach
Rules first, AI second
AI interprets, not invents
The model works inside a fixed structure. It explains the action in plain language. It cannot ignore exposure limits, available cash, or the option to do nothing.
Paper trades, real ledger
Orders go through Binance's demo environment. Positions, P&L, and allocation are all real numbers — just not real capital. The experiment is honest about what it is.
The person running it
About Hugo
I'm a software developer with 15+ years of experience, currently working as a Senior Software Engineer at ChartMogul and as a freelance product developer. My stack is primarily Ruby on Rails, with a strong focus on design and user experience.
I've worked at companies like Cint, carwow, and Runtime Revolution. I've been using AI tools as part of my engineering workflow since they became practical, which made building an AI-in-the-loop trading system a natural experiment to run.
The lab is built the same way I approach all my work: clear architecture, honest interfaces, no hidden state. Claude is the AI co-pilot for both the development and the trading decisions — which feels appropriate.
Important note
This is not financial advice
Hugo's Trading Lab publishes rules-based analysis with AI-assisted decision summaries for informational purposes only. Nothing here should be treated as investment advice. Crypto markets are volatile and carry significant risk.
The lab is designed to be observable and honest about its process — not to tell you what to do with your own capital. Always apply your own judgement, and never put money at risk that you cannot afford to lose.
Follow the experiment
The portfolio is public. Holdings, orders, and the reasoning behind each decision are all there.