Systematizing fundamental investing.
Qi’s Fundamental Intelligence Lab explores how AI and quantitative methods can make fundamental investing more systematic, scalable, measurable, and continuously improving — without replacing the judgment of the fundamental PM.
Fundamental investing, with a new intelligence layer.
This is not a quant model looking for more fundamental data. The portfolio remains fundamentally driven. AI expands the PM’s ability to read, remember, compare, and recognize patterns; quantitative methods test whether those patterns are repeatable and useful.
The PM remains the core
Thesis, variant perception, business quality, earnings path, catalysts, risk/reward, sizing, and portfolio construction remain fundamentally driven.
AI becomes research memory
AI helps maintain company state, compare evidence across time, retrieve historical analogs, and make implicit fundamental pattern recognition explicit.
Quant becomes validation
Statistics act as a measurement and falsification layer: which patterns repeat, when they work, and whether they add information beyond conventional factors.
A fundamental-first investment operating system.
The workflow follows how a strong fundamental investor actually thinks: understand the business, recognize a setup, remember analogs, make a portfolio decision, and learn from the outcome.
Fundamental Research
Build and update the living company thesis from management, financials, competitors, customers, suppliers, and industry evidence.
Pattern Recognition
Identify inflections, contradictions, cycle setups, management changes, competitive shifts, and second-order read-throughs.
Systematic Memory
Store timestamped company states and pattern records so today’s setup can be compared with prior quarters, peers, and historical analogs.
Decision Support
Layer analogs, quantitative validation, risk, liquidity, correlation, and portfolio context on top of fundamental conviction.
Learning
Compare thesis vs outcome, pattern vs realization, and AI vs PM judgment so the PM’s pattern library compounds.
Turn implicit PM judgment into cumulative intelligence.
The lab begins with fundamental questions, not return optimization.
Can AI recognize fundamental inflections before the sell side changes estimates?
Can management credibility be tracked consistently enough to improve fundamental judgment?
Can second-order read-throughs become part of a reusable pattern library?
Which patterns does an exceptional fundamental investor recognize repeatedly — and how can we preserve their context?
Public ideas. Private machinery.
The public lab can share philosophy and selected research while keeping the actual implementation moat private.
Public research
- Fundamental investment philosophy
- High-level system architecture
- Research questions
- Selected conceptual frameworks
- Research notes and learnings
Private lab
- Fundamental ontology and scoring rubrics
- Pattern library and historical analog retrieval
- Agent procedures and evaluation
- Point-in-time datasets and validation
- Portfolio decisions, sizing logic, and decision history
Day 001.
Start with the pattern recognition of great fundamental investing.
A strong fundamental PM is constantly matching today’s evidence against an internal library of prior situations: the wording change that matters, the low-quality beat, the inventory setup seen before, the competitor comment that changes a thesis, or the inflection that appears before consensus numbers move.
The working hypothesis is that modern AI can help preserve, retrieve, and structure this implicit knowledge — while quantitative methods test which patterns are genuinely repeatable. The objective is not to automate stock picking, but to make fundamental judgment more systematic and cumulative.
How can we turn the pattern recognition of an exceptional fundamental investor into a systematic, measurable, and continuously improving investment process — without losing fundamental judgment?