01 — Framing
What we are doing.
We build and document a simulated portfolio of companies we believe are able to lead their industry over the next five to ten years. This page describes the method: the criteria, the score, the points where an analysis stops and the limits of the whole framework.
The content we publish is identical for every reader and takes no account of anyone's financial situation, objectives or risk tolerance. We do not publish entry prices, position sizes or answers about individual portfolios: we explain why in the Transparency and limits section.
The portfolio we document is a simulation: no real capital is invested in it. It exists to make our process verifiable over time, not to be replicated.
02 — The six pillars
What we measure, and with what weight.
Every company receives a score from 0 to 10 on six dimensions. These are the weights, they add up to one and they do not change from one analysis to the next. For every pillar we state what raises the score and what lowers it.
Moat / disruptive potential
How defensible the competitive advantage is over five to ten years, and whether it tends to grow or erode.
Raises the score: network effects that strengthen with scale, real switching costs for the customer, proprietary data that grows with use.
Lowers the score: an advantage a well capitalised competitor can reproduce, dependence on a single channel or customer, replaceable technology.
Team
The quality of those in charge: capital allocation, incentives, ownership stake, honesty with shareholders.
Raises the score: founders or managers with a meaningful stake, capital reinvested with documented returns, mistakes admitted openly.
Lowers the score: recurring dilution of shareholders, pay disconnected from results, promises repeatedly made and missed.
TAM / growth
The size of the realistically addressable market and the credibility of the growth path within it.
Raises the score: a large and structurally expanding market, a current share that is still small, growth driven by the product.
Lowers the score: a market defined elastically to look bigger, growth bought with discounts, dependence on a favourable cycle.
Unit economics
Whether every unit sold creates value: margins, customer acquisition cost, conversion of profit into cash at the current scale.
Raises the score: high and stable gross margins, customers who return more than they cost to acquire, profit that turns into cash.
Lowers the score: growth that worsens margins, profitability propped up by stock based compensation, cash systematically below reported profit.
Financial durability
The ability to get through prolonged stress without having to turn to the market at the worst possible moment.
Raises the score: positive net cash, positive operating cash flow, debt maturities that are distant and spread out.
Lowers the score: material short term debt, tight covenants, dependence on refinancing or capital raises.
Valuation
How much of the future path is already in the price. It is the pillar with the smallest weight: a great business at full price is still a great business.
Raises the score: a price that discounts assumptions more cautious than ours.
Lowers the score: a price that already embeds the most favourable scenario, with no margin for disappointment.
03 — The Generational Score
One number, and its exact meaning.
The Generational Score is the weighted average of the six pillars, rescaled to 0-100: round( Σ (pillar score × weight) × 10 ). No manual adjustment: we publish the number the formula produces.
An example with real numbers, the Duolingo analysis: 6.5 × 0.25 + 8.5 × 0.20 + 7.0 × 0.15 + 8.0 × 0.15 + 10.0 × 0.15 + 7.0 × 0.10 = 7.775 → 77.75 → 78.
| Band | Meaning |
|---|---|
| 80–100 | Full match with our criteria: no weak pillar, a solid thesis across all six dimensions. |
| 65–79 | Good match with at least one explicit reservation: the thesis holds, but one pillar remains the point of attention. |
| 50–64 | Partial match: the company passes the gates but does not rank among the top ten. |
| Below 50 | Does not match our criteria. |
| Partial | Analysis stopped at a gate: the score covers only the pillars assessed up to that point and is not comparable with the others. |
The score measures one thing only: how closely the company matches our selection criteria. It is not a judgement on the share price, does not point to a moment to buy or sell, and is not comparable with third party ratings.
04 — The blocking gates
Where an analysis can stop.
Some flaws are not offset by a good score elsewhere. Three steps of the analysis are blocking gates: the disruptive character of the business, the quality of the team and the strength of the balance sheet. If one fails, the analysis stops there: the score stays partial and the company does not enter the portfolio.
The analysis is published anyway, with the reason for the stop: a reject not put in writing straight away is too easy to rewrite after the fact.
- In portfolio
- The analysis passed every gate and the company is today one of the positions in the simulated portfolio.
- Eligible
- The analysis passed every gate with a full score, but the company is not among the current positions.
- Gate not passed
- The analysis stopped at a gate: we publish the partial score and the reason for the stop.
05 — How an analysis is born
Generated with AI, approved by us.
Every analysis goes through four steps, in the same order, without exceptions. No content reaches publication without a human signature.
Generation
An artificial intelligence system applies the methodology on this page to the company, using financial data from licensed sources and documentary research.
Data verification
Every number quoted is checked mechanically against the source data and receives an accuracy score. Below the threshold, the analysis does not reach review: it goes back or is discarded.
Human review
We read the analysis with the verification report alongside it and approve or reject it with a note. No analysis publishes itself.
Publication
Only after approval. The analysis goes out with authors, date and time of completion and a dated reference price.
The analyses are produced with the assistance of artificial intelligence systems following the methodology described on this page, and are reviewed and approved by Riccardo Priotto and Simone Gallio, who assume editorial responsibility for them.
06 — Updates
When we publish, when we revisit.
An analysis is a photograph with a date. Here we state when we publish new ones and what obliges us to revisit the existing ones.
New analyses
We publish analyses in groups, as they pass data verification and human review, normally several times a month. The process sets the pace, not the news calendar.
Revisiting analyses
We revisit an analysis after every quarterly release from the company and at every extraordinary event that touches the thesis. Every analysis carries the date and time of its latest version.
The score log →
We keep a public log of the scores assigned over the last twelve months, with their changes: promotions, downgrades and analyses stopped at a gate remain available to consult.
07 — The simulated portfolio
Ten positions, every change with its reasons.
The portfolio is where the method stops being theory: the best analyses become positions, with rules stated in advance and not afterwards.
There are three rules. A cap: never more than ten positions; every entry beyond the cap forces an exit, and this obliges us to compare every new idea with the weakest of the ones we hold. Balancing constraints: by sector and by market, because ten theses concentrated on the same risk are a single thesis dressed up as ten. Immutable reasons: every change comes with the reason written at the moment of the decision, which is never rewritten after the fact.
The portfolio is a simulation with no real capital: what we document is the selection process. Weights, dates and reasons stay on record, verifiable over time, without turning these pages into an allocation suggestion.
08 — Transparency and limits
What we disclose, what we don't do.
A method is judged also by what it admits it does not know and does not want to do.
Conflicts of interest
Every analysis discloses whether the authors personally hold the security discussed. We receive no payment from the companies we analyse nor from third parties with an interest in our conclusions.
Sources and data
Financial data comes from licensed sources; quotations are short and attributed; projections are labelled as such. Facts carry their source and date; assessments are recognisable as our own opinions.
What we do not publish, and why
No entry prices, no position sizes, no answers about individual situations. These are deliberate choices: this is research addressed to the public, not personalised advice.
Limits of the method
The score is a judgement of ours: it condenses six dimensions into one number and by construction loses some nuance. The gates reduce false positives, but they also exclude companies that will turn out to be excellent. We prefer this error to its opposite.
For investors and funds
The same engine, your model.
The process behind our analyses is infrastructure: data collection, a scoring grid, automated validation, human review, tracked publication. We can adapt it to your organisation’s model, with your criteria and your weights.