Famous last words.
It's March 2025. I fire up Visual Studio for the first time in over a decade. GitHub Copilot is open in another tab. Claude is standing by.
I've got a plan: build a machine learning trading bot. Learn ML the way I learn best—by building something!
"Two weeks," I think. "Maybe three if I hit some snags."
I have my first XGBoost model trained in 48 hours. It's running. It's making predictions. It's working.
The report shows £90K a day profit.
Done! Project complete!
...Wait, let me just check that code.
Oh. Oh no. Not £90K. Not even close.
And that's how it all started! The "ya, ya, ya, NOOOOOOO" of TradeBOT.
Nine Months Later
It's November 2025.
I've tested over 50 technical indicators.
I've trained more XGBoost models than I can count.
I've built an entire trading platform from scratch (I named it SPIRIT).
I've debugged model predictions at 3 AM.
I've migrated to cloud servers because my home infrastructure literally blew up under the amount of data.
And I've learned one critical lesson:
YouTube lied to me.
Why I'm Starting This Blog
For nine months, I've been quietly working on TradeBOT. Testing, failing, learning, iterating.
I've documented everything—daily logs, technical decisions, architectural diagrams, performance reports. Hundreds of pages of internal notes tracking every win, every failure, every lesson learned.
But it's all been private.
Here's the thing: I've watched dozens of YouTube videos about trading bots. They make it look easy:
"This simple strategy returns 300% per year!"
"I built a bot in a weekend and it's printing money!"
"Anyone can do this!"
It's not easy.
And I think that's worth talking about.
Not because I want to discourage anyone—quite the opposite. I want to share the real journey. The messy, challenging, occasionally frustrating, but ultimately fascinating process of building something this complex.
And So It Begins
Well, sort of. You're jumping in about nine months into the journey, so feel free to read the About page for the full backstory on how I got here.
Going forward, I'll be posting weekly updates on what I'm working on—the wins, the losses, and everything in between. Where possible, I'll give you some background on why I'm working on a particular idea.
Why You Should Follow Along
If you're interested in:
Algorithmic trading (especially the messy reality)
Machine learning for finance (not just tutorials)
Building complex software projects (infrastructure, testing, deployment)
Following a real R&D journey in real-time (wins and losses)
...then this blog is for you.
I'm not selling a course. I'm not pitching a "guaranteed profitable system." I'm not here to tell you this is easy.
I'm here to document what actually happens when a cybersecurity professional with a trading obsession decides to teach himself ML by building a bot.
It's been nine months of daily work, and I'm still not sure if it'll work.
But I'm going to find out, and you're welcome to join me.
What's Next
Over the coming weeks, I'll be publishing:
📊 The SPIRIT platform architecture — how I built a modular trading system
❌ Why my first 20 models failed — and what I learned
🌊 The whipsaw problem — and my ML solution
☁️ Cloud migration story — when my home server died
📈 Daily insights — from current testing
If you want to follow along, subscribe (it's free). You'll get updates when new posts go live.
If you've got questions, thoughts, or you're working on something similar — drop a comment. I'd love to hear from you.
And if you think this content could help someone else — share it. The more people interested in real algo trading development (not just YouTube hype), the better.
Let's Find Out Together
Can machine learning beat the market?
Can a stubborn cybersecurity guy with a trading obsession figure this out?
I don't know yet. But after nine months of "ya, ya, ya, NOOOOOOO," I'm determined to find out.
Welcome to TradeBOT.
— Tim
