Disclosure: This post may contain affiliate links. If you click one of these links and make a purchase, I may earn a commission at no additional cost to you. As always, I only recommend products and services I truly believe in.
AI Investment Diary – Vol.6
AI Investing #6: A “Disastrous” Failure Built My New Strategy “C”
Until now, my approach to **AI investing** has been to use AI as a brilliant co-pilot. I, the human, made the final call in an “AI-assisted” framework. But to further enhance my portfolio’s diversification, I decided it was time to step into a new territory: the world of **system trading**, where human emotion and discretion are completely removed, and the AI takes the lead in capturing market patterns.
This post is the full story behind my new strategy, codenamed “C”—a journey that was far from smooth and reveals the failures and discoveries that are reshaping my approach to AI-powered finance.
[INSERT_TABLE_OF_CONTENTS_HERE]The Grind Begins: Data, Code, and a Partnership with AI
Building an AI system trade from scratch is less glamorous than it sounds. It began with the painstaking task of data collection. I opened **Google Colab**, a programming environment, and used the **Yahoo Finance API** to download 15 years of historical stock data for every component of the S&P 500. The AI would tell me what data it needed, and I would write the code to fetch it. It felt like a true partnership.
This process alone took hours. Watching endless streams of data scroll by, I felt a surge of excitement. I was stepping into a domain once inaccessible to individual investors.
The First Model: A “Disastrous” Backtest Result
That excitement didn’t last long. After prepping the data, the first model my AI and I built—a weekly rebalancing strategy—produced backtest results that were, in a word, “disastrous.”
The performance chart showed my strategy lagging far behind the market average. In an attempt to leverage the AI’s processing speed, I had increased the trading frequency. This turned out to be a classic mistake: the model was just reacting to short-term market noise, racking up transaction costs, and bleeding money. For a moment, my heart sank. “Is system trading really only for the pros?” I wondered.
Rebuilding from Failure: The True Value of AI Collaboration
Just as I was about to give up, the AI prompted me: “Let’s analyze the factors that caused this model to fail.” This began a new dialogue. I fed the failed model’s transaction history back to the AI, and we dug deep. “Why did losses accelerate in this phase?” “Which indicators were acting as noise?” This is the core value of collaborating with an AI. It helps you overcome the limitations of human biases and knowledge gaps. This iterative process of analysis and refinement is central to a successful **AI investing** plan.
If you’re interested in the foundational theories behind this kind of portfolio analysis, I highly recommend exploring resources on quantitative finance. You can find many excellent books on the subject on Amazon that break down these complex ideas.
The Final Evolution of My AI Portfolio
After this rigorous validation process, I decided to integrate “Strategy C” as a new pillar of my portfolio. I reallocated approximately 12% of my total assets from my satellite holdings to this new AI system trading strategy. This marks a new stage in my portfolio’s evolution, making my overall asset allocation strategy more active and deeply integrated with AI.
Before
After
Trusting the AI, Not Blindly Following It
Strategy C operates on a defined logic: it invests in 30 stocks from the S&P 500 that it determines to have a statistical edge at any given time. Unlike my “Strategy G,” it’s not feasible to manually review the fundamentals of 30 different companies on a frequent basis.
However, this isn’t a “set it and forget it” approach. My role has shifted. I am now the architect of the logic the AI follows. I am the one who must rigorously backtest the results and continuously improve the logic in response to changing market dynamics. By repeatedly validating the logic I create, I can execute this **AI investing** strategy **with my own responsibility, my own intention, and with full confidence**. This is the new relationship with AI that I’m striving for.
In the next “Knowledge” installment, I’ll dive deeper into the fascinating world of “quantitative investing” and “factors” that form the backbone of Strategy C.
Comments