How the Equitytale Health Score works

The Health Score rates a listed Indian company from 0 to 100 on up to six measures of its filed financials, and compares it against the other companies in its own sector. Every company page states how many of the six we could actually read for that company. It describes financial condition. It is not a recommendation, a price target, or a view on whether the share is worth buying.

What goes into it

Six pillars, each scored 0–100, then combined with the weights below. Banks, NBFCs and insurers are scored differently: leverage is a lender’s product rather than its risk, so the balance-sheet pillar is dropped for them entirely and its weight redistributed rather than counted as a failure.

Not every company supplies every input. A filer that reported no balance sheet has no return on capital and no debt-to-equity ratio, so those pillars are built from whichever measures survived — 74 of the companies we score have a profitability pillar resting on one measure rather than three. The missing ones are left out and the remaining weights rescaled, never counted as zero. Company pages therefore describe what each pillar asks rather than listing inputs that may not have been available for that company.

Why we rank instead of using thresholds

Financial ratios are badly behaved. Across the companies we score, interest cover runs about 19× at the 75th percentile and about 80× at the 90th, and half the market sits at a debt-to-equity ratio of 0.18 or lower. Any fixed scale — “above 20 is good” — either saturates immediately or spends its whole range on a tail almost nobody occupies.

So each measure is scored by where it ranks, not by what it equals. A score of 78 means healthier than 78% of the comparison group. The comparison group is the company’s own sector when that sector has at least 20 companies we can score, and the whole listed universe otherwise — every company page states which of the two it used.

Within that group, each measure is ranked only against the companies that actually reported it, which is usually fewer than the group itself — 245 companies are classified as Financial Services but only 53 have a comparable return on capital. Company pages therefore quote a range rather than a single peer count, because saying “ranked against 245 companies” when one pillar saw 53 would be a precise and false claim.

If fewer than five companies reported a measure, we do not rank it at all and the measure drops out. A percentile drawn from one or two companies is not a weak statistic, it is a meaningless one: rank a company alone and it lands at the bottom of its own distribution, scoring zero for being the only one we could measure.

Two things ranking alone gets wrong

For measures where lower is better, a plain ranking produces two specific errors, and we correct both.

Negative equity is not the lowest debt. A company whose accumulated losses exceed its capital has a negative debt-to-equity ratio, which sorts below every debt-free company in the market. Ranked naively it would score as the safest balance sheet on the site. We score it zero, which is what it is.

Zero is a crowd, not a position. Hundreds of companies carry exactly no debt and exactly no pledged promoter shares. Because ranking places ties at the bottom of their group, the first company with a rounding-error amount of debt would fall far below them. We reserve the top mark for a genuinely clean zero and rank only the companies that actually carry some, so a trivial amount of debt costs a few points rather than a sixth of the pillar.

The opposite applies to valuation multiples: a price-to-earnings ratio near zero is a distress signal, a one-off gain, or a reporting error — never a perfect price. We do not score implausibly low multiples at all rather than crown them.

Bands

The bands describe measured financial condition, not an action. The score contains no entry price, no holding period and no sense of your portfolio, so it cannot support a “buy” — and we are not a SEBI-registered Research Analyst.

When we publish no score

A pillar we cannot measure is left out and the remaining weights are rescaled — a company that never filed a cash flow statement must not score the same as one that filed a bad one. But when less than 55% of the total weight is measurable, we publish no number at all and say so. A confident-looking score built on half the evidence is worse than a blank, because it is the figure that gets quoted.

Where this score is weak

• Turnarounds read as weak. Growth needs four filed annual years, and a company that went from a loss to a profit has no meaningful growth rate — the arithmetic is not defined across zero. Such a company also scores low on “profitable years”. A genuine recovery therefore looks worse here than it is.
• Recent listings often score nothing. Under four filed years the growth pillar drops out entirely, which for many newly-listed companies is enough to fall below the coverage gate.
• Governance is narrow. It reads only the promoter pledge and the promoter stake — both from the latest filed shareholding pattern. It does not yet read the audit opinion, related-party dealings, or board composition, so a company can score well here and still have governance problems this score cannot see.
• Lenders are scored on less. We do not yet ingest asset-quality data — gross and net NPAs, provision coverage — so a bank is judged without the measure that matters most for a bank.
• It is backward-looking. Every input is a filed historical figure. The score says nothing about orders won, management changes, or anything else that has not yet reached a filing.

Price momentum is deliberately excluded

Company pages show RSI and moving averages beside the score, but never inside it. The score is rebuilt once a night; momentum changes every session. Combining them would mean the number we publish in a page’s machine-readable data stopped matching the number a reader sees, as soon as the page was cached. The two are shown together and dated separately.

See also our data methodology for how the underlying filings are sourced and checked, and the disclaimer.