The Lynch Screen

Refresh, 19-Sep-2026. The frozen criteria were not amended. What changed: (1) P/E, yield and market cap re-read from Screener at the 18-Sep close for all 57 original names and the price ratio recomputed — no verdict flipped (EIL 1.29, MAYUR 1.10, MPS 1.09 remain WATCH; every Stage-3 exclusion is still below 1.0). (2) Seven names added, all excluded: MACPOWER, TVSSRICHAK, CFF and PANAMAPET at Stage 2; SHREEPUSHK, JGCHEM and WABAG on price. For these seven the callable-debt share (U3) was not extracted this run; under Lynch's rule an unmeasured test is recorded as IE and does not exclude. Categories are my classification from each name’s profile. (3) Company events since 9-Sep that touch a test but do not change a verdict: KMEW (first Q1 call; promoter 49.99%), NRB (pledge cut to 3.10%), INTERARCH (income-tax block assessment), UNIMECH (late senior-management disclosures), EPL (Blackstone exit), ORIANA (~₹540cr of guarantees) — see each profile’s “Changes since baseline” dropdown.

Refresh, 6-Oct-2026. The frozen criteria were not amended; only the price inputs were re-read (Screener consolidated, re-read after the 6-Oct close for all 64 names). Growth rates are unchanged — no company has filed annual results since the last run. One verdict flipped: MPS, WATCH → EXCLUDED, on a price ratio of 0.96 against the 1.0 gate. Its P/E barely moved (24.9 → 24.9); the change is the dividend yield, which Screener now reports as 0.00% against 3.11% before. That is a correction, not a cut: the 3.11% was a trailing figure, and MPS paid nothing in FY26 after the board suspended the dividend to fund the Unbound Medicine acquisition (payout 108% FY24 → 95% FY25 → 0% FY26), which the MPS profile flagged on 2-Sep. Verified at Screener and against the profile; no other name’s yield moved more than 0.3pp. EIL and MAYUR are now the only names on WATCH (1.08 and 1.08). ATLANTAELE’s ratio fell 1.63 → 1.46, but the price gate is not its live test — it is excluded at Stage 2 on the growth ceiling, and a price move cannot rescue a Stage-1 or Stage-2 exclusion. Company events since 19-Sep that touch a test without changing a verdict: DEE shareholders approved conversion of consortium debt into equity on an event of default (debt test), WHEELS raised ₹250cr by QIP, ANUPAM carries the guarantee and pledge behind the Bliss GVS financing, and ORIANA is divesting to Actis in tranches. U3 (callable-debt share) is still unextracted for the seven names added on 19-Sep and remains recorded as IE.

Refresh, 21-Oct-2026 (close). The frozen criteria were not amended; P/E, yield and market cap were re-read from Screener after the 21-Oct close for all 64 names (standalone pages for the six names whose consolidated pages are blank). Two verdicts flipped, both EXCLUDED → WATCH, both on price alone: MPS 0.96 → 1.10 and WABAG → 1.01. Each was second-sourced before being written: implied EPS is unchanged for both (MPS ₹107.1 → ₹106.7; WABAG ₹63.8 → ₹63.6), dividend yields are unchanged, and both stocks fell about 12% since 6-Oct (MPS ₹2,645 → ₹2,326, including −5.2% on 21-Oct; WABAG ₹2,015 → ₹1,782). WABAG clears the gate by 0.01 — a one-percent price move reverses it. No BUY: the closest is EIL at 1.31. Universal Cables (added to the watchlist 13-Oct) was screened the same day as a single-name addendum — see the next note.

Addendum, 21-Oct-2026 — Universal Cables (65th name). Criteria not amended; full work-up in UNIVCABLES_ADDENDUM.md. Classified C4 (cyclical): 5-yr earnings growth 19.5% is in the fast-grower band, but profit fell 51% FY19→FY21 and management itself writes that “the demand of power cables has been cyclical in pattern” — C4 outranks C3, as for DYCL. Stage 1: U1 PASS (institutions 4.8%), U2 PARTIAL (D/E 62% consolidated, 128% standalone), U3 FAIL (60.5% current on the balance-sheet split), U4 PASS (no customer over 10% of revenue), U5 PASS, U6 PASS (dividend raised every year, share count flat). Excluded on U3, like the other 19. Method note: on the borrowings notes the callable share is 46.1% — the first name where the two readings disagree on the verdict; the screen’s method is kept for consistency, and the question matters most for the borderline KMEW (51%) and DYNAMATIC (50.2%). Excluded independently at Stage 2 — 🔴 C4-2, 23.9x on all-time-peak earnings (≈48x normalised), and C4-4, inventory +54% against sales +26% — and at Stage 3 (price ratio 0.83).

Re-priced on the 23-Oct-2026 close; no verdict flipped. GARWARE and SSWL reported Q2FY27 in the window; 5-year growth is restated only on annual results, so neither print changes a Lynch input.

4-Nov-2026 (close), re-priced on the Windows PC. One flip: WABAG WATCH → EXCLUDED at a price ratio of 0.97 — the price rose 8.5% (₹1,710 → ₹1,855) and the P/E moved with it (26.9 → 29.1), so trailing EPS is unchanged; the gate is crossed at a P/E of 28.3. WABAG reports Q2FY27 on 5-Nov — read this verdict as provisional until then. 63MOONS and SYNCOMF are not yet screened.

Start here — how this screen works

Lynch classifies a company before testing it, because what counts as good news depends entirely on what kind of company it is. A low P/E is a bargain in a stalwart and a warning in a cyclical. A missing dividend is fatal in a slow grower and a positive in a fast one.

Classify into six categories the first real judgement

C1	Slow grower — 2–4% growth. You own it for the dividend. “Why else would you own them?”
C2	Stalwart — 10–12%. Bought for a 30–50% gain, then rotated. Recession ballast.
C3	Fast grower — 20–25%. “the land of the 10- to 40-baggers.” Where the tenbaggers live, and where the money is lost.
C4	Cyclical — earnings that fall as well as rise. “It is here that the unwary stockpicker is most easily parted from his money, and in stocks that he considers safe.”
C5	Turnaround — “no growers”, in crisis. Largest absolute gains; least correlated with the market.
C6	Asset play — something valuable the market has overlooked. Low risk, high gain.

What this watchlist turned out to be: 36 fast growers, 19 cyclicals, 7 stalwarts, 1 slow grower, 2 turnarounds — and zero asset plays. The category Lynch rates low-risk and high-gain is entirely absent. Every name here is an earnings-growth bet, which is a concentration of method, invisible on a sector chart.

Six universal tests 20 excluded

U1	Neglect — “the percentage of institutional ownership. The lower the better.”
U2	Debt quantum — “companies that have no debt can't go bankrupt.”
U3	🔴 Debt structure — is the debt callable on demand? “It's the kind of debt, as much as the actual amount, that separates the winners from the losers in a crisis.”
U4	Customer concentration — “25 to 50 percent of its wares to a single customer is a precarious situation.”
U5	Diworseification — acquisitions proven one at a time, or a spree?
U6	Capital allocation — buybacks and dividends, or cash dissipated?

Every one of the 20 exclusions came from U3 alone. See the chart below — it is the single most destructive test in this screen.

Category-specific tests 23 excluded · 6 abstained

Each category is judged on its own terms. The two that did the work here:

C3	The 25% ceiling. “I'm wary of companies that seem to be growing faster than 25 percent. Those 50 percenters usually are found in hot industries, and you know what that means.” — six names excluded on this alone, and MACPOWER and CFF now join them
C4	🔴 The inverted P/E. “Anticipate a shrinking p/e multiple as business recovers, when peak earnings are achieved.” — a low multiple on peak earnings is a sell signal, not a bargain

ABSTAIN is not a failure. Six names lack the published history to be judged at all. Lynch permits deciding on incomplete information — “the scientific mind that needs to know all the data will be thwarted here” — so these are recorded as my failure to establish the facts, not as a defect of the company.

Price 13 excluded

One formula, and it is the book's central one:

(earnings growth rate + dividend yield) ÷ P/E
“Less than a 1 is poor, and 1.5 is okay, but what you're really looking for is a 2 or better.”

With two refusals attached, so the test cannot be misused: “buying a company with mediocre prospects just because the stock is cheap is a losing technique” and “selling an outstanding fast grower because its stock seems slightly overpriced is a losing technique.”

The funnel

🔴 The result, in one chart

Every name with both a growth rate and a price ratio. The shaded box is what Lynch actually wants — earnings growing 15–25% at a price below the growth rate. Note where the names are, and are not.

Slow growerStalwart Fast growerCyclical Turnaround◯ hollow = cyclical at peak earnings

Growth is the 5-year earnings CAGR (10-year where 5 is unavailable). Price ratio is (growth + yield) ÷ P/E. 41 of 65 names have both.

The two halves of the finding.

The well-behaved growers are too expensive. Six names sit in Lynch's 15–25% band — ANTHEM, CARYSIL, GARWARE, SANSERA, SHREEPUSHK and JGCHEM — and all six fail the price gate, at 0.19, 0.69, 0.55, 0.34, 0.88 and 0.76.
The cheap names are cyclicals at the top. Seven of the nine names clearing the gate are cyclicals whose latest earnings are within 10% of their all-time peak (the other two, ATLANTAELE and CFF, are fast growers past the 35% ceiling; PANAMAPET is the new one, at a 5.9x P/E on windfall earnings). Their ratios look wonderful because the E is temporarily high. A conventional low-P/E screen buys every one of them.
The ideal box is not empty — it holds exactly two names, and both are disqualified by what kind of company they are. DEEP is a cyclical at its peak; NLC carries a D/E of 130%. Neither fails on growth or on price.

🔴 Why 19 names died on one test

U3 asks a single question: if a lender got nervous tomorrow, could it demand its money back? Lynch's controlled pair — GCA, carrying $114m of bank debt, “went kaput and was bought out at about 10 cents a share”, while Applied Materials with $17m of debt and $36m of cash went $8 to $36. “The debt burden was the difference.”

Share of total borrowings sitting in current (callable) liabilities, read from each company's FY26 balance sheet. 25 annual reports downloaded and parsed; two hand-verified against the source statements before anything was built on the numbers.

But read this as a finding about India, not about any one company. Indian mid-caps finance themselves on cash credit, packing credit and working-capital demand loans — renewed annually, and in law repayable on demand. Lynch's test is not calibrated for that market, and applying it honestly fails four-fifths of the names that carry debt. The counter-argument deserves stating: in practice these limits are sticky and are rarely pulled from a going concern. But his point is precisely about what happens when a company stops being a comfortable credit — and the frozen criterion excludes on this test alone. It was applied as written rather than softened after the damage was visible.

The three survivors · all WATCH, none a BUY

EIL, MAYUR and MPS are stalwarts. (WABAG, a fast grower, fell out on price on 4-Nov-2026 — see the note at the top.) (MPS dropped out on 6-Oct-2026 when its dividend yield corrected to zero, and came back on 21-Oct-2026 on price alone; WABAG cleared the 1.0 gate by 0.01 the same day — see the refresh note above. The earlier line “both effectively debt-free and growing faster over three years than over ten” was written for EIL and MAYUR and has not been re-tested for the other two.) None is cheap enough. They become buys at a price ratio of 1.5 — EIL at a P/E of 16.2 (-14%); MPS at a P/E of 16.0 (-23%); MAYUR at a P/E of 11.4 (-31%) — recomputed on 4-Nov-2026 (close) prices“If you have a list of companies that you'd like to own if only the stock price were reduced, the end of the year is a likely time to find the deals you've been waiting for.”

Every name · 65

Company	Cat	Verdict	Where it stopped	Growth	P/E	Price ratio	Callable

Disagreement report

The prediction held. Before this screen ran I recorded: “Lynch should resurrect names Fisher killed and reject names Fisher liked. If both screens return the same shortlist, one of the two implementations is wrong.”

The two frameworks agree on exactly one name out of 57 — MPS.
EPL and TRANSRAILL were Fisher survivors. Lynch excludes both — EPL on price, Transrail because ₹571 of every ₹659 it borrows is repayable on demand.
EIL was excluded by Fisher for a shrinking order book. Lynch puts it on WATCH.
Neither is wrong. They are asking different questions of the same companies.

Against the book — 4 of 6 disclosed holdings excluded, but mostly on price.

MAYUR is one of the three survivors. DCMSHRIRAM is the only holding correctly positioned in its cycle — a cyclical trading off-peak.
GARWARE (0.53) and EPL (0.57) are excluded on price alone, on businesses the framework otherwise likes. Lynch's own instruction cuts both ways here: “selling an outstanding fast grower because its stock seems slightly overpriced is a losing technique.” This is not a sell signal — only a statement about where he would add.
DATAMATICS — growth stalled: 1.8% over three years against 13.4% over ten, which is one of his named sell signs.
DEEP is the one worth pausing on. Both frameworks reject it by unrelated routes — Fisher on two consecutive Q4 write-offs, Lynch because an oil-services company at peak earnings on 12.6× is a cyclical at the top, not a cheap stock.

What this screen does not claim

U4, U5 and U6 are incomplete. Three customer-concentration failures were established on evidence — SSWL's Hyundai at 74%, Concord's Indian Railways, Rajesh's Gujarat utilities — and one pass. The rest is IE, so no survivor has been cleared on customer concentration.
The rollout test and the one-shot fragility test were never reached for most fast growers, because the growth band or the price gate excluded them first. Those names are excluded on what was tested.
Insider buying is IE throughout, as pre-registered — Indian disclosure gives aggregate promoter holding, not the “seven vice presidents buying 1,000 shares apiece” granularity Lynch wants.
🔴 Blinding was imperfect, and more so than in the Fisher run — the holdings list was already in context, and the Fisher result was known before this screen ran, a contamination Fisher did not carry. The mitigation is that the criteria were frozen first and the decisive tests are arithmetic. It is a real limitation.
🔴 Three parse failures were caught mid-run and two changed a verdict. A first extractor returned KMEW at −24,444,108. A second captured note references as values and wrongly excluded DCMSHRIRAM. Plotting the data for this page exposed a third: NLC was in the wrong category, sitting inside the ideal zone where no slow grower can be. A parse that looks right is not a parse that has been checked — and a chart catches what a table hides.

Sources: SCREEN_CRITERIA.md (frozen 9-Sep-2026) · CHAIN_STATE.md (20-chapter chain read) · PHILOSOPHY.md · CLASSIFICATION.md · STAGE1.md · PORTFOLIO_LYNCH.md.
Financials from Screener consolidated pages, 9-Sep-2026, every page verified by company-name echo. Debt structure from 25 FY26 annual reports. Standalone artefact — not wired to the main dashboard or the deploy repo.