01 — PrincipleThe governing principle
A research report is a claim about the future made on the basis of the past. The only honest way to publish one is to state the assumptions explicitly enough that a reader can disagree with a specific input rather than the whole conclusion. That is the standard we hold ourselves to, and it is deliberately uncomfortable: it means publishing the number that breaks our own thesis, the sensitivity table that shows where the model is fragile, and the scenario in which we are wrong.
We are not neutral. Neutrality is a posture, not a method. A report that lists eight considerations on each side and declines to weight them is not objective — it is evasive. We take positions. We just take them in writing, with inputs attached, so that being wrong is a falsifiable event rather than a matter of interpretation.
Before publication, every report must survive one question: if this conclusion is wrong, what is the single most likely reason? That reason is written into the bear case in the same words used internally. If it cannot be written plainly, the report is not ready.
02 — PipelineThe research pipeline
Each report moves through six stages. No stage is skipped, and no report is published from a partial pass.
Scope definition
The question is written as a single falsifiable sentence. "Is NVIDIA's revenue durable?" is not a question. "Does the FY2027 revenue run-rate survive a 20% reduction in hyperscaler capital spending?" is.
Source harvest
Ten years of filings, every earnings release and transcript, supplier disclosures, competitor filings, regulatory dockets, and customs/trade data. Everything is stored with a citation key so that any figure in the final report can be traced in one step.
Model construction
The financial model is built bottom-up from unit economics — units, average selling price, cost of goods, and mix — rather than from a growth rate applied to last year's revenue. Top-down growth rates are treated as an output to be explained, not an input.
Earnings-quality rebuild
Reported net income is rebuilt into cash. Working-capital drift, capitalised costs, non-operating gains, and related-party flows are isolated and sized as a percentage of the operating result.
Adversarial review
A second analyst is tasked with destroying the thesis. Their brief is not to balance it but to find the failure mode. Findings are incorporated or rebutted in writing.
Publication & revision
The report ships with a dated changelog. After each subsequent quarterly filing it is re-cut, and the revision is logged even when the rating is unchanged.
03 — StandardsQuantitative standards
A small number of conventions are applied without exception, because they are the ones most often used to obscure rather than reveal.
- Cash conversion is reported alongside every earnings figure. Operating cash flow as a percentage of net income, and free cash flow as a percentage of net income, are stated for each period discussed.
- Non-operating income is separated, always. Gains on equity stakes, interest income, and one-time items are removed from operating performance and sized independently.
- Working capital is treated as a signal, not a footnote. Days sales outstanding and days inventory outstanding are computed and compared across periods. A widening gap between revenue growth and receivables growth is treated as a finding.
- Market share is triangulated. Where a share figure is cited, it is cross-checked across at least two independent methodologies — typically unit shipments and revenue — and any divergence is disclosed.
- Estimates are labeled. Anything not directly reported is marked "Farstar estimate" in the accompanying chart note.
04 — ValuationValuation framework
We use three complementary approaches and treat disagreement between them as information rather than noise.
Scenario-weighted earnings multiples
Bear, base, and bull cases are defined by explicit revenue, gross margin, and operating expense assumptions. Each case is assigned a probability and an exit multiple justified by the growth and return profile of that case. The output is a probability-weighted value, published alongside the individual scenarios so the weighting can be disputed separately from the assumptions.
Reverse discounted cash flow
Rather than forecasting a value from assumptions, we invert the question: at today's price, what growth rate and terminal margin must the business deliver to justify the valuation? This reframes the debate from "is it cheap" to "what has to be true," which is a question with a defensible answer.
Cross-checks
Comparable-company multiples, historical multiple ranges for the same issuer, and sum-of-the-parts are used as sanity checks. Where a cross-check diverges materially from the primary method, the divergence is disclosed rather than averaged away.
05 — RatingsRating definitions
| Rating | Expected total return | Time horizon |
|---|---|---|
| High Conviction Long | Greater than +30% | 12–24 months |
| Constructive | +10% to +30% | 12 months |
| Neutral | −10% to +10% | 12 months |
| Cautious | −10% to −30% | 12 months |
| High Conviction Short | Less than −30% | 12–24 months |
| Under Review | Not rated | — |
Expected total return includes assumed dividends and is measured against the issuer's primary listing currency. Ratings are assigned on a 12-month view unless stated otherwise.
06 — DisclosuresConflicts & disclosures
Farstar Capital and its analysts hold no positions in any security covered on this site as of the most recent publication date. Any position taken in future will be disclosed here and in the footer of the relevant report within five business days of establishment.
Compensation
Farstar Capital receives no compensation from issuers, investment banks, or any party with a commercial interest in the conclusions of its research. Research is funded exclusively by subscription and licensing revenue from readers who have no influence over coverage decisions or conclusions.
Coverage selection
Coverage is initiated by Farstar analysts. We do not accept requests to initiate, suspend, or alter coverage, and we do not provide pre-publication drafts to issuers or their representatives.
Published coverage is filed under seven sector categories — Big Tech, AI, Biotech, Digital Assets, Fintech, Semi (semiconductors) and Platform Economy. A category with no published report is shown as open rather than omitted, so the shape of the coverage we intend to build stays visible, and a new report simply files into its category.
Limitations of this research
This research is provided for informational purposes only. It is not investment advice and does not constitute an offer, solicitation, or recommendation to buy or sell any security. It does not consider the specific investment objectives, financial situation, or needs of any particular person. Securities discussed may not be suitable for all investors. Forward-looking statements are estimates and are inherently uncertain; actual results may differ materially. Farstar Capital makes no warranty as to the accuracy or completeness of third-party data. Past performance is not indicative of future results. Readers should conduct their own independent diligence and consult a licensed adviser before making any investment decision.
Third-party data
Company financial data is sourced from issuer filings with the U.S. Securities and Exchange Commission and from issuer earnings releases. Industry and market-share data is sourced from named third-party research providers and is identified as such in each chart note. Estimates attributable to Farstar are labeled "Farstar estimate."
07 — ContactContact
Corrections, challenges to our analysis, and data requests are welcome and are read by the analyst who wrote the report. We respond to substantive disagreements with the underlying argument, not with a restatement of the conclusion.
research@farstar.capital
corrections@farstar.capital