About this video
Most people are lying to you about how difficult it is to build a trading bot. In this video, I am breaking down the entire process of how I built a 24/7 AI-powered trading bot that optimises itself while I sleep. We go from a completely blank terminal to a fully functioning crypto and stock dashboard that uses AI to make micro adjustments to its own code. Key Takeaways: - How to use Claude Code to build a Python trading bot without writing the logic yourself. - Why paper trading is essential before you ever touch a real API or real money. - Using scheduled AI routines to perform daily and weekly strategy optimisations. - Setting up a local dashboard to monitor trades, losses, and gains in real time. - Moving from stock trading (with market hours) to 24/7 crypto trading using Binance.
The Finance Industry is Gatekeeping Trading Bots to Keep You Away from the Profits
The finance industry wants you to believe that building a trading bot is reserved for quant geniuses with PhDs and decades of experience. They wrap the process in mystical terminology and complex mathematics to ensure you feel too overwhelmed to even try. I am here to tell you that it is all a facade.
In my latest video, I pull back the curtain on how I built a 24/7 trading bot using AI tools like Claude Code and Warp Terminal. You do not need to be a master programmer or a Wall Street veteran to get this running.
The Strategy: Start with Paper Trading
Before you ever put a single penny of real money at risk, you must master the art of paper trading. My bot runs completely locally, looking at real market data from Yahoo Finance and Binance, but executing trades in a hypothetical environment. This allows you to refine your strategy without the stress of losing capital.
Letting the AI Optimise Your Decisions
The real magic happens when you integrate AI into the feedback loop. Rather than just building a static script, I use Claude Co-worker to schedule daily and weekly reviews.
- Daily Reviews: The AI looks at the previous day of trading and makes micro adjustments to the code.
- Weekly Reviews: The AI assesses the overarching strategy and researches new trends to implement broader changes.
Why This Matters
For twenty years, I have built websites and apps, yet even I felt overwhelmed by the hype surrounding AI trading. The truth is that many people shouting about this online do not actually understand the code behind it. By breaking the bot down into its raw components: a local database, a dashboard, and a scheduled AI routine: we can demystify the entire process.
Stop listening to the "gurus" who say this is impossible for the average person. The tools are here, the data is free, and the code is more accessible than ever. Keep on vibing.
Transcript▾
This is a trading bot that I have built, that is running 24/7 making trades for me pretty much every few seconds. As you can see, it is actually in the green. What is happening is that every day the AI is assessing what happened the day before and making micro adjustments to the strategy. Then, every week, the AI looks at the overall strategy and makes larger adjustments to make me money. Instead of gatekeeping and making this a mystical process, I am going to show you exactly how I built it.
We are doing this in Claude Code. I told it: build a trading bot that looks at the market and trades, but make it completely local and hypothetical. I strongly recommend you do this first to understand and refine your strategy through public sources. I know nothing about trading or what websites to use, so I told the AI to do the research for me. I also wanted a dashboard to monitor progress and theoretical losses.
I ran into some usage limits and had to upgrade, but essentially it built a Python app. I used Warp Terminal, which is an AI-based terminal, to install dependencies like YFinance. The bot watches ten US stocks like Apple and Tesla, scanning every five minutes. It uses two strategies: a moving average crossover and an RSI indicator. If a stock is oversold, it buys the dip; if overbought, it sells.
I then decided to build a crypto version because crypto markets never close. Using Claude Code, I adapted the bot to trade ten major crypto pairs via the Binance API. It scans every five seconds. The magic happens in the scheduled tasks. I created a routine for a daily crypto review where the AI looks at logs and makes micro adjustments to the strategy in the code. I also set up a weekly review to adjust the overarching strategy.
What I learned is that you should not be put off by the excitement people share on things that are relatively simple. To some, this is magic, but when you understand the raw components, it is just a website on a loop making trades. You should not be put off by people acting like this is complicated. It really isn't.