Retail
Retail day traders with no coding background are describing a strategy in plain English to Claude and getting a working trading bot back, and some are beating the market with it
One trader told Business Insider his Claude-built bot returned 87% in a month against the S&P 500; another says a fully automated account built the same way is up 106% this year, with both crediting the removal of panic-selling and revenge-trading rather than any secret edge.
The term "vibe coding" undersells what is actually happening here. These are retail traders with no quantitative finance background and, in many cases, no prior coding experience, describing a trading strategy in ordinary language and getting back working, deployable automation. That is a genuinely new capability, not a rebrand of existing retail algo tools, because the barrier that used to separate someone with a trading idea from someone who could actually implement it systematically has effectively been removed for a meaningful slice of the retail population.
What the traders themselves say is working is instructive: not a secret predictive edge, but the elimination of the classic behavioral leaks that erode retail returns. Panic-selling into a drawdown, over-trading out of boredom or anxiety, revenge-trading to win back a loss. A bot that executes a predefined strategy without an amygdala attached will, by construction, avoid all three, and the returns being reported look less like alpha discovery and more like what happens when execution discipline that professionals pay for gets made available for free to anyone who can describe what they want in a sentence.
That said, the framing in these reports, screenshots of a single strong month or a favorable year, is precisely the kind of survivorship-shaped evidence that should be read with real caution. Nobody profiles the vibe-coded bot that lost 40% in a volatile week, and 2026 has not exactly been short of volatile weeks for AI-thematic names specifically. A strategy that removes emotional interference is not the same thing as a strategy that has been risk-managed, backtested across regimes, or stress-tested against the kind of leverage-driven unwind that just hit institutional AI funds this same month.
The regulatory and platform-risk questions this raises are still mostly unasked. A retail trader running a self-built AI bot against their own brokerage account sits entirely outside the compliance infrastructure that governs licensed advisers and broker-dealers, which means there is no supervisory backstop if the described strategy and the coded behavior diverge, or if the model quietly changes how it interprets an instruction over time. The traders profiled here are, functionally, running unsupervised proprietary trading systems built by a general-purpose AI assistant, and the fact that it currently works is not the same as it being safe by design.
Read the original: Business Insider - Retail day traders with no coding background are describing a strategy in plain English to Claude and getting a working trading bot back, and some are beating the market with it. Commentary is the independent editorial view of Share Trading; the original article is credited to its publisher.