The method
Our Methodology
Most research tools stop at retrieval — they help you find a filing, search a transcript, or pull a ratio. Clarivest starts where that ends. Once a filing is public, we run a structured forensic review of it, looking for the specific patterns that have historically preceded earnings disappointments, restatements, or governance problems.
What we actually check
We don’t aim to cover everything at once. Depth beats breadth: a small number of high-signal checks, done properly, catch more than a long list of superficial ones. Our current coverage focuses on three areas.
Earnings quality. We look at the gap between reported earnings and the underlying cash generation of the business — persistent divergence between the two is one of the oldest and most reliable warning signs in financial analysis. Related indicators, like unusual movement in accruals relative to sales growth, are scored as part of the same picture.
Working capital trends. Days Sales Outstanding, inventory days, and payable days are tracked over multiple periods, not just the latest one. A steady deterioration in how quickly a company collects from customers, or how much inventory is building up relative to sales, often shows up well before it reaches the headline numbers.
Ind AS vs. adjusted reporting. Where companies report adjusted or management-defined figures (adjusted EBITDA, adjusted PAT, and similar) alongside their Ind AS-compliant statutory numbers, we reconcile the two and flag when the gap between them is growing, recurring, or inconsistently defined from one period to the next.
Each check produces a flag with a plain-language explanation of what was found and why it matters — not just a score.
How scoring works
The rules come first, the model explains. Every check in Clarivest is a codified, deterministic rule — not a judgment call made by an AI model in the moment. The role of the underlying language model is to read the output of those rules and explain it in clear language: what was flagged, what it typically indicates, and what to look at next in the filing. It does not decide the score, and it does not offer an opinion on whether to buy, sell, or hold anything.
This split matters. A model can sound confident about the wrong thing; a rule either fires or it doesn’t, based on numbers taken directly from the filing.
Where the data comes from
We source financial data directly from exchange filings — NSE and BSE XBRL for annual and quarterly financials, prioritizing whichever filing format is available first for a given company, with manual review as a fallback for filers where structured data isn’t available. We do not rely on third-party aggregator summaries for the numbers our checks run on.
What Clarivest is, and isn’t
Clarivest is a research and analysis tool. It surfaces patterns in public financial filings using a structured, rules-based process, so you can investigate further with your own judgment. It does not provide investment advice, is not currently a SEBI-registered Research Analyst or Investment Adviser, and nothing here is a recommendation to buy, sell, or hold any security.
We’ll expand this page as we add more checks — if there’s a specific pattern or ratio you’d like to see covered, get in touch.