Value

Value Screens That Survive Contact With Indian Data

Saral Team · 1 August 2026

The value premium is the oldest documented pattern in cross-sectional equity returns, and it is also the one most likely to hand a retail investor a portfolio of genuinely bad companies. Both things are true, and the gap between them is almost entirely construction.

The premium and the argument about why

Fama and French’s 1992 paper established book-to-market as a systematic driver of returns, and HML — high minus low book-to-market — became one of the three original factors. The premium has been found in essentially every market with enough history to test.

What it means has been argued about ever since. Fama and French’s own reading is risk-based: cheap firms are distressed, distress is a risk investors want compensating for, and the premium is that compensation. Lakonishok, Shleifer and Vishny’s 1994 paper argues the opposite — that the premium is a behavioural error. Investors extrapolate recent performance too far into the future, so companies with a bad few years get over-discounted and companies with a good few years get over-priced, and value is the reversal of that mistake.

The distinction matters practically. If value is compensation for risk, it should persist forever and you should expect real losses in the states of the world where the risk shows up. If it is a behavioural error, it can decay as more capital learns about it — and the 2015–2020 global value drought is at least consistent with that.

Either way the drawdowns are real and long. Value underperformed growth for most of a decade in developed markets. Any honest presentation of a value strategy leads with that.

Two multiples, because each is broken differently

The instinct is to pick one ratio. Both common choices have a failure mode severe enough that using either alone produces systematic errors.

Price-to-book breaks on asset-light businesses. Book value measures accumulated accounting capital. For a bank or a cement manufacturer that is a meaningful number. For a software services firm, a consumer brand, or a pharma company whose value is in a molecule pipeline, most of what the business is worth was expensed rather than capitalised and never appears on the balance sheet. Those companies look permanently expensive on P/B, and a P/B screen systematically excludes entire sectors for an accounting reason rather than an economic one.

Price-to-earnings breaks near zero earnings. As profit approaches zero, P/E approaches infinity; as profit crosses into loss, P/E goes negative. A naive “low P/E” sort puts loss-making companies at the very top, because negative is less than positive. This single bug has probably destroyed more retail capital than any other screening error in India, because the resulting portfolio looks like deep value and is actually a basket of companies in trouble.

Blending both and requiring positive profit before ranking removes most of the damage from both. Where P/B is uninformative, P/E carries the signal; where earnings are noisy, book value anchors it.

price_to_book      = market_cap / total_equity_annual_consolidated
price_to_earnings  = market_cap / net_profit_annual_consolidated

Cheap for a reason, or cheap by mistake

The distinction that matters is between a company that is temporarily out of favour and one that is correctly priced for permanent decline. No screen resolves this cleanly, but several filters shift the odds:

Positive profit, as a hard gate before ranking. This is not a quality filter — it is a data-integrity filter that stops the P/E sort from inverting.

Positive book value. Companies with negative net worth produce a negative P/B, which sorts to the top of a “low” ranking for the same reason. Accumulated losses have exceeded paid-in capital; that is not a discount.

Index membership. Restricting to Nifty 500 constituents as of the rebalance date does two jobs at once: it removes the untradeable tail, and it applies a crude ongoing-viability test, since sustained decline eventually causes index exclusion.

Surveillance exclusion. Stocks under NSE’s Graded Surveillance Measure trade in periodic call auctions with up to 100% margin and 5% price bands. Many will screen as very cheap. None of them are buyable at the screened price.

Wider diversification. Forty names rather than fifteen. Value’s return distribution has a fat left tail — the value trap that never recovers — and the defence against a fat left tail is more names, not better selection.

Consolidated, and lagged

Two data choices that determine whether the backtest means anything.

Consolidated over standalone. Indian companies file both. Standalone accounts exclude subsidiaries entirely, so for any group structure the standalone balance sheet describes a holding shell rather than the business. Screening a mixed universe on standalone numbers compares a full operating company against a fragment of another.

Publication lag. FY2024 numbers are not public on 31 March 2024. They are filed weeks or months later. A screen that reads them from the fiscal year end is trading on information that did not exist, and the resulting backtest is not optimistic — it is fictional. Any fundamentals field has to be shifted to its actual publication date before the as-of join.

Rebalance quarterly, not monthly

Annual fundamentals change once a year and are confirmed four times. Rebalancing monthly against a signal that updates annually means eleven months of paying costs to trade on the same information. Quarterly matches the filing cadence.

Try it

The Value 40 template screens Nifty 500 constituents, requires positive annual consolidated profit, builds both multiples from filed figures, blends them into a z-scored cheapness composite, and holds forty names equally weighted with a quarterly rebalance.

Two comparisons are more informative than the headline number. Run it against the Quality 30 template and note how few names overlap — quality and value select nearly disjoint sets, which is why they combine well. Then find the strategy’s worst twelve-month stretch and ask honestly whether you would have held through it. That question determines whether a value allocation is right for you far more than the average return does.