The story behind the bots.
I’ve been actively investing in the markets since 2010. My investment horizon has exclusively been a long-term outlook (15+ years), primarily focused on a buy-and-hold (B&H) strategy.
Is AMD a solid company? Is it trading at a bargain? Okay, I’ll invest in it with the assumption that I will steadily enhance my position over the course of several decades. When looking at the grand scheme of things, the B&H strategy will effectively yield unrealized gains as a lengthy investment horizon tends to flatten out the “noise” or the volatility associated with earnings releases and macro and microeconomic trends.
As Redditors infamously state, “stocks only go up”, then what’s the point of employing a different strategy? After all, B&H has been tested against time, only Black Swan events (e.g., 1987 “Black Monday” Crash, September 11 Attacks, 2008 Global Financial Crisis) could disrupt the inevitable. Don’t get me wrong, I am pro-B&H, always have been, always will be. What I have trouble with is following my intuition in exiting positions. If my position in AMD doubles, my gut tells me it’s time to sell, but that same gut is telling me I don’t want to miss out on more potential gains. The fear of exiting or trimming positions and missing out on compounded gains far outweighs any corrections and downturns.
I figured why not employ strategies that force me to recognize gains? Albeit minimal, recognizing small gains and either banking the profit or reinvesting will open avenues to explore strategies and learn more about the investing world. This was the impetus for developing AI bots that autonomously trade with allocated capital based on a set of parameters.
I first started developing AI bots in 2026. Since then, I have come across numerous issues, had Claude troubleshoot the same bugs repeatedly, and refined bot strategies daily. This is my journey in understanding the world of AI trading bots.
This blog documents my journey building autonomous trading bots with an AI coding agent as the backbone. The bots trade leveraged and unleveraged index ETFs through a retail brokerage API, using strategies ranging from simple dip accumulation to RSI mean reversion and multi-signal confluence.
Building trading bots with AI is a new frontier. Most resources are either too theoretical or skip the messy parts. I wanted to document the real process: the bugs, the strategy iterations, the infrastructure headaches, and the reasoning behind why each strategy was kept, changed, or retired.
This is not a performance log. Figures in the posts are illustrative and are there to show how a strategy behaves, not to report what any account returned.