An experiment in what an agentic-AI research framework can do — published in full so you can check its work, not a tip service to follow. Generic research and education, not advice, not a personal recommendation, not FCA-authorised. Hypothetical book, not real money. Capital at risk.
Portfolio Lab · Research Lab
Portfolio Lab — transparent, low-cost portfolios with real out-of-sample performance, net of all costs. View the portfolios →
Portfolio Lab · Research Lab

Research Lab

This is an experiment into what an agentic-AI framework can do — whether disciplined AI agents, held to one rule (a claim only counts if there's evidence it actually predicts returns), can research investment ideas with discipline and evidence. We publish everything in full so you can check the system's work — that openness is the proof, and the whole point. It is deliberately detailed, but it is not a tip service and following it is not the idea: the experiment is the framework and how it performs, not any single stock.

You'll find our analysis of investment signals (which work, which don't, and why) and the concepts behind them — the evidence engine for how the framework thinks. It's a process under construction, published as it develops. Generic research and education only: no advice, no recommendations, no real money.

Please read. This is research and education about a process — not financial advice, not a personal recommendation, and not a tip service. Portfolio Lab is not authorised or regulated by the FCA. Nothing here is tailored to you or suggests that you should buy or sell anything. The book is hypothetical; capital is at risk and past or hypothetical performance is not a reliable indicator of future results. Always do your own research. Read our full disclosures & risk notice →

Signal Research work in progress

The evidence engine behind how we pick — which signals actually predict returns, tested on our own data. Keep-decisions and reject-decisions both matter. See all signal studies →

C+ OVERLAY ONLY
Quality / Gross Profitability — is it still true?
Quality
F AVOID
Congressional Trading — the 'Pelosi tracker'
Information / political
C+ SITUATIONAL
Share Buybacks — only the real ones, only when cheap
Capital allocation
B+ OWN THE ENDS
Free-cash-flow yield — does buying cheap-on-cash beat the market?
Value / cash flow

Concepts

The ideas our framework rests on — how we think, in plain English. See all concepts →

Evidence over narrative
A good story isn't a reason to buy. A claim only counts if there's evidence it predicts returns — and we throw out the ones that fail, even when they sound clever.
Confluence — stacking independent signals
Conviction should come from how many genuinely independent reasons agree, not how loudly one shouts. The trap is counting the same fact twice.
Mean reversion to a quality-backed trend
Prices revert to a long-run path — but only if the business is genuinely good. The difference between a beach ball held underwater and a falling knife.
Margin of safety — price is the risk
The risk isn't a wobbly chart, it's the price you paid. The gap between price and value is the cushion that lets you be wrong and survive — a hard pre-condition.
Signals are overlays, not bets
Most signals are real but short-lived — good at timing entry and size, bad at being the reason to hold for a year. Durable strands carry the position; signals sit on top.
How to test a signal honestly
The wrong benchmark, survivorship, trusting the white paper, forcing a straight line. The traps we design around so a result can actually prove us wrong.