After a decade in equity research, we recognized the old playbook was broken. So we built a different one - technical signals first, fundamentals second, AI throughout. A short-term trading system designed for absolute returns in any market environment.
After ten-plus years in equity research, one conclusion became unavoidable: the traditional approach to picking stocks is breaking down. Not because the analysis is wrong, but because the world the analysis was built for no longer exists. The edge has moved, and most of the industry hasn't moved with it.
The information advantage that used to come from deep fundamental work has compressed. Everyone has the same data, the same 10-Ks, the same transcript access. The edge from "knowing the company better" is increasingly marginal - and the market prices that marginal edge faster every year.
Same frameworks, same DCF models, same consensus estimates. By the time a research note is published, the market has already absorbed the conclusion. The research isn't wrong - it's just late. And late research, no matter how thorough, has no commercial value.
Reading filings, building models, comparing peer sets, tracking estimates - this is the bulk of what equity analysts do, and it's the bulk of what AI now does faster and more consistently. The human analyst's role is shrinking. The question isn't whether AI replaces research - it's what comes after.
The traditional approach layers technical analysis on top of fundamental research - you find a company you like, then you try to time it. We do the opposite. We screen for stocks that are technically ready to move - showing accumulation, momentum shifts, or pattern breaks - and then we apply fundamental research to decide how much to commit. The chart tells us when. The fundamentals tell us how much.
The system is rules-based and repeatable. No gut calls, no narrative drift. Each step has a defined input and a defined output - if a candidate doesn't pass a step, it doesn't advance. AI runs the screening and pattern recognition; human judgment runs the fundamental validation and risk management.
AI-driven technical screens scan both US and Hong Kong equity universes for stocks showing accumulation patterns, momentum shifts, volume anomalies, and multi-timeframe signal convergence. The system surfaces candidates that look like they're about to move - before they move.
Fundamental research confirms there's a real business behind the chart. We're not buying charts - we're buying companies whose charts are telling us something. If the fundamentals don't support the technical signal, the candidate is cut. No exceptions.
Position sizing is driven by conviction level and risk parameters - not by a fixed model. The stronger the technical signal and the fundamental validation, the larger the position. Risk is defined before entry, not adjusted after.
Day trades exit intraday - US positions are closed before the session ends, no overnight risk. Swing trades in HK are held for days to weeks, monitored against their entry thesis with predefined exit rules. The system tracks every position in real time - if the technical signal fades or the pattern breaks, the position exits. Capital is recycled into the next opportunity.
The opportunity cost of capital is the most underpriced risk in investing. You can be right about a company - the business grows, the earnings compound, the thesis plays out exactly as you expected - and still lose money. Because capital deployed at the wrong time is capital that isn't working.
We design for timing because timing is where returns actually live. A stock that returns 30% over twelve months but sits flat for nine of them is not the same as a stock that returns 30% in three weeks. The first ties up capital. The second frees it. In a world where AI compresses the time between signal and price action, the edge belongs to whoever acts on the signal first - and recycles capital fastest.
We're selective about who we work with, but we read every message. If you're an investor, a technologist, or just curious about the approach, send a note.