Deep Industries Ltd
NSE:DEEPINDS · BSE:543288 · Gas compression, dehydration, workover rigs & integrated offshore services · Primaegis Equity Desk · 06 Aug 2026
A genuine step-change in revenue, a collapsed order intake, and a guidance pair that cannot both be true. FY26 revenue rose +54.6% to ₹890.71 Cr at a 39.9% EBITDA margin, on gearing of 0.096× and zero promoter pledge. But book-to-bill fell 4.04× → 1.05×, the order book is not verifiable from filings, and management's own FY27 execution and revenue guides are arithmetically incompatible. The stock re-rated +42.3% in a fortnight on a government scheme in which it has disclosed no participation.
Close of record ₹627.00 — settled NSE close, three feeds agreeing to the paisa. Seven agents · adversarial gate REJECT 3 of 5, 29 claims audited · all four sections revised and re-verified.
How to read this report
This desk publishes its audit trail. Chips after a claim tell you where the number came from and how far to trust it:
| filing + date | Primary filing — read off an exchange filing, annual report or investor deck. Strongest tier. |
| raw.json | Traceable to the desk's gathered data — re-checkable at that location. |
| devil gate | Found by the adversarial reviewer, not the analysts — challenged, corrected or withdrawn under attack. |
| GAP | Not disclosed, or not captured. An absence recorded rather than filled. |
| SECONDARY | Unverified media or aggregator sourcing. Carried for context, never load-bearing. |
| DERIVED | Computed by the desk from filed inputs, with the arithmetic shown. |
Colour is signal, never decoration. Green and red mark favourable and unfavourable data only. Variance cells carry ▲/▼ with the delta so direction survives in greyscale.
⚠️ Price of record — the corpus carried FOUR prices for one day
Adopted: ₹627.00, the settled NSE close of 06-Aug-2026 — agreed to the paisa across three independent feeds, with the prior close back-solving exactly to ₹643.60.
The other three: ₹629.50 (raw.json / computed.json, explicitly a mid of three intraday quotes captured ~15:12 IST), ₹628.65 (BSE), and ₹628.15 — which sat inside primary_docs.md's table of daily closes without being one.
A desk that publishes four prices for one day has no price. All five sections are now on ₹627.00; the others survive only where explicitly labelled intraday. Nothing in the conclusions moves — but the restatement was mandatory.
Part A — Fundamentals
A1–A5. What the business did
Deep Industries supplies gas compression, dehydration, workover rigs and integrated offshore services to Indian E&P. FY26 was a genuine step-change: revenue ₹890.71 Cr (+54.6%), EBITDA ₹355 Cr at 39.9%, PAT including NCI ₹197.06 Cr (owners ₹179.94 Cr + NCI ₹17.12 Cr, EPS ₹28.12 on the owners' basis). Q1 FY27 followed with revenue ₹278.92 Cr and consolidated PAT ₹89.14 Cr.
The balance sheet is genuinely conservative and the report should not be read as saying otherwise: gearing 0.096×, promoter pledge zero across all fourteen promoter entities at every quarter-end tested on the primary Reg-31 declaration, no dilution, and a dividend record that clears its own free cash flow. Piotroski 7/9.
devil gate All four basis traps fired. Standalone vs consolidated is live (Q1 standalone PAT ₹55.18 Cr vs consolidated ₹89.14 Cr). PAT-with-NCI vs owners' EPS is live and confirmed. Lease liabilities are not a trap here — right-of-use assets are nil in both years. And screener's Other Income nets the exceptional: −₹139 Cr / −₹228 Cr against an audited gross of +₹69.54 Cr / +₹32.33 Cr.
🚨 Margins are softening — and the desk corrected itself here
An earlier draft dismissed computed.json's margin-pressure signal as an artefact of a unit corruption in the file. The file was regenerated and the flag survives on clean data.
| Basis | FY25 | FY26 | TTM | Direction |
|---|---|---|---|---|
| screener series (clean) | 41.8% | 39.9% | 39.4% | ▼ −190 / −240 bps |
| desk audited derivation | 40.2% | 39.9% | — | ▼ −30 bps |
| quarterly clean OPM, YoY | — | — | −230 bps | ▼ |
Magnitude is base-dependent; direction is not. The mechanism is mix — materials +3.76 pp of revenue, trading goods up 12.4×. The 0.4% margin the corrupt file implied was the artefact; the softening is real.
A11. The guidance does not foot — and Q1 proved which number to test
Q1 FY27 established an identity that settles the whole argument: execution is revenue. ₹279 Cr executed against ₹278.92 Cr booked — a difference of ₹0.08 Cr.
| If this guide binds… | Q2–Q4 revenue/qtr | vs Q1 | FY27 revenue | Implied PAT |
|---|---|---|---|---|
| ₹800 Cr execution | ₹173.7 Cr | −37.7% | −10.2% | ₹255 Cr (27% below guidance) |
| “>25% revenue growth” | ₹278 Cr | flat | +25% | ₹355 Cr ties |
Both cannot be true. The desk calls ₹800 Cr as the number that breaks, on four grounds — chief among them that Q1 annualised already delivers the revenue guide at zero growth, and that the ₹800 Cr figure breaks two guided numbers where “>25%” breaks one.
Falsification test, published: Q2 FY27 revenue below ~₹230 Cr. DERIVED
🚨 Order intake has collapsed while revenue accelerated
| FY25 | FY26 | Q1 FY27 | |
|---|---|---|---|
| Book-to-bill | 4.04× | 1.05× | 1.14× |
The order book is no longer compounding — revenue is being delivered out of it. And it cannot be checked: Reg-30-announced wins over twelve months total ₹525.91 Cr against implied additions of ~₹937.7 Cr, so roughly ₹680 Cr of intake sits below the disclosure threshold. The order book is not verifiable from filings at all.
A8. Forensic — not a fraud file, and the desk says so first
Every reconciliation attempted tied. Inventory foots to the rupee (balance-sheet movement ₹3,461.07 L = cash-flow line ₹3,461.07 L, residual ₹0.00). The working-capital bridge rebuilds to a ₹0.93 Cr residual — 0.19% of a ₹480 Cr receivable book. Payables are being paid faster (110.4 → 85.0 days), the opposite of a stressed signature. Scorecard: 4 🔴 · 6 ⚠️ · 2 ✓, across 15 ranked risks.
The exhibit the desk downgraded on its own initiative
The forensic section's strongest anti-fraud finding was that Dolphin Offshore's separately-filed audited PAT × 25% (₹17.13 Cr) ties to Deep's consolidated NCI (₹17.12 Cr) — to ₹0.01 Cr. On sourcing that figure properly it discovered the same firm audits both the parent and the subsidiary, and downgraded its own exhibit from an independent check to a filing-integrity check. That correction was volunteered, not demanded.
A related observation survives and is sharper: consolidated audit fees fell 19% (₹22.41 L → ₹18.13 L) while revenue rose 55% — in the year of a ₹208 Cr write-off and a Key Audit Matter.
🚨 Dolphin's 58.9% net margin — decomposed, not alleged
₹68.54 Cr of PAT on ₹116.42 Cr of revenue, after depreciation and tax, on vessel chartering. The section ran it rather than flagging it:
- ₹17.42 Cr of other income, +461% YoY — 15.0 pp of the margin
- A ₹13.97 Cr judgement-based deferred-tax asset — 20.4% of PAT, and 62% of the entire group's deferred-tax credit
- The residual 31.9% is the arithmetic complement of a 0.192× asset turnover — ROA 11.3% vs the group's 7.6%, a 3.7 pp premium rather than an anomaly
No unexplained profit — but ~46% of it is non-operating or non-cash, and it is already normalising: Q1 FY27 net margin is 34.6%.
A13. The related-party perimeter — the weakest structural feature
| Exposure | Amount | Why it matters |
|---|---|---|
| Standalone loans & advances that are related-party | ₹372.37 Cr = 99.94% | Effectively the entire standalone lending book. Qualified as standalone — the consolidated non-group figure is ₹134.75 Cr |
| Prabha Energy (consolidated, non-group) | ₹134.75 Cr, +48.7% | An independent director also sits on the borrower's board |
| AGM RPT headroom sought | ₹750 Cr | 84.2% of turnover, blanket and unvalued — including ₹350 Cr into an IFSC entity with no capital, no approval and no bank account |
| AGM Item 6 (found late, in the primary notice) | ₹72 L p.a. | Office of profit for a promoter — more than any FY26 executive-director package — as an ordinary resolution |
The promoter CMD sits on the audit committee that reviews those transactions. The board ran 89 days below 50% independence, and a new independent director was given the audit-committee chair on day one. Two family trusts hold 44.19% while the two managing directors hold 600 shares between them. Governance composite: 2.3 / 5.
A10 / A16. Valuation — two defensible bands that disagree on DIRECTION
The model refused this name: "the call carried no figure that bridges to earnings growth". Traced upstream: the guidance corpus holds a 14-Nov-2025 call with empty metrics, a period already filed, and bridged.eps_growth: null. The refusal is the finding; every forward row below is a DESK EXTENSION.
| Basis | What it is | Base case |
|---|---|---|
| Table A | Filed EPS against the filed-year exit band — the store's own convention, and what the model would have produced had it not refused | +15.4% |
| Table B | A clean forward EPS against a multiple struck on clean earnings — internally consistent on economics | −27.2% |
They disagree on direction, not merely on magnitude, and the disagreement is structural rather than a modelling choice. Table A's most recency-weighted mark is inflated by the very write-off that depresses its own filed EPS — the exceptional is counted twice, in the same direction. Table B's weakness is that it inherits a clean-EPS assumption. Substituting the clean FY26 mark drops the exit median 37%.
Neither is wrong, and the base case is not decomposable into growth and rating. On the evidence available the desk cannot tell you which way this is mispriced — and saying so is more useful than picking. Both tables are published; neither is averaged.
What is not ambiguous is the level. P/E 19.44× against a filed-year band of 6.86–16.09× (n=4) — the observation sits outside its own range. On four filed years that band is thin, and the desk says so. P/B 2.007×, P/S 4.14×, EV/EBITDA 10.66×, on a market cap of ₹4,012.80 Cr.
Context that belongs beside the multiple: the shares ran +36.4% from ₹459.50 on 23-Jul, and +42.3% to the 04-Aug intraday high, after the Cabinet approved the ₹84,084 Cr Samudra Manthan offshore scheme on 31-Jul-2026. The scheme is real. Deep Industries' share of it is currently zero in every filing.
Part B — Technical: Stage 2A, two weeks old
What is genuinely constructive
- The 30-week MA bottomed at 429.17 (05-Jun), turned at 431.60 (03-Jul) and is now 439.51 and accelerating — an early-stage breakout, not a late one.
- Breakout-week volume 10.40 m = 7.30× the ten-week average, clearing Weinstein's 2× test emphatically.
- The desk corrected the gathered file's read.
raw.jsonsaid delivery "collapsed → not institutional accumulation." Half wrong: the delivery ratio fell, but mean daily delivered shares went 80,218 → 513,342 — 6.4×, totalling 19.7% of free float in nine sessions, with an up/down volume ratio of 4.83. - The pullback runs on 0.334 relative volume — constructive, not distributive.
What sits against it
- Price is +42.66% above the 30-week MA. Both frameworks want Stage 2 entries near the line.
- 28-Jul was a distribution day inside the advance — −1.84% on the second-heaviest volume.
- 03-Aug climax volume followed by a 04-Aug stall bar is the buying-climax signature.
- Relative strength is unseasoned. +31.4 pp vs Nifty over 3M, but only +17.3 pp over 12M — and on 23-July this stock was −3.09% year-on-year. It was a twelve-month laggard fourteen sessions ago, and still lags its own cohort by 21.8 pp on 12M.
B4. Does any entry survive a ≤3% stop? No.
The only stop clearing 3% is ₹611.00 (2.55%) — and it is 0.56 ATR wide. That is noise, not structure. Every structure-anchored stop runs 9.73% (below the 569.90 polarity) to 30.62% (below the 30-week MA).
Two compliant entries exist conditionally, and the better one is worth waiting for: a retest-and-reclaim of ₹569.90 — entry ~₹574, stop ₹557, 2.96% risk at 4.71:1, the best R:R on the board. It requires an 8.5% decline first.
Part C — Thesis
The bull case is real: revenue +54.6%, ~40% EBITDA margins, gearing 0.096×, zero pledge, Piotroski 7/9, every reconciliation tying — one of them to ₹0.01 Cr — and a genuine, Cabinet-approved ₹84,084 Cr offshore scheme as a forward catalyst.
The bear case is that the price has re-rated 42.3% in a fortnight, to above the top of its own filed-year range, on a scheme in which the company has disclosed no participation — while order intake collapsed from 4.04× to 1.05× book-to-bill, the order book became unverifiable from filings, and management published two FY27 numbers that cannot both be true.
C4. Alignment — DIVERGING
| Lens | Reads | Direction |
|---|---|---|
| Operating P&L | +54.6% revenue, ~40% EBITDA, gearing 0.096×, Piotroski 7/9 | Positive |
| Order intake | Book-to-bill 4.04× → 1.05×; the book is not filing-verifiable | Negative |
| Guidance | ₹800 Cr execution and “>25% revenue” cannot both hold | Negative — and self-inflicted |
| Technical | Stage 2A two weeks old on 7.30× volume — but +42.66% extended, no entry | Positive on trend, negative on location |
| Valuation | Above its own filed-year range, and the two defensible bands disagree on direction | Unresolvable |
| Governance | 2.3 / 5 | Negative |
C7. PRIMAEGIS opinion — NEUTRAL
Conviction is Low, and that is the finding rather than a hedge. The two defensible valuation bands disagree on direction — +15.4% against −27.2% — because the reported basis double-counts an exceptional and the clean basis inherits an assumption. When the arithmetic cannot separate growth from rating, saying so is more useful than picking.
What holds it up from REDUCE: the growth is real and large, the balance sheet is genuinely clean, and every forensic reconciliation tied. This is not a fraud file and the desk says so plainly.
What holds it down from ACCUMULATE: a price above the top of its own range after a 42.3% fortnight on a scheme with no disclosed company content; book-to-bill at 1.05× on a book that cannot be verified; two FY27 guidance numbers that are arithmetically incompatible; an unresolved ONGC suspension on a customer at 52–70% of the book; and governance at 2.3/5 with ₹750 Cr of unvalued RPT headroom on the agenda in nineteen days.
⬆ Upgrade to ACCUMULATE if any TWO of:
- Q2 FY27 revenue at or above ~₹278 Cr — the growth guide binds, the ₹800 Cr figure was the error, and the contradiction resolves in the shareholder's favour.
- Book-to-bill back above 1.5× on disclosed awards, or an ONGC award after the suspension lifts.
- A Samudra Manthan award with disclosed Deep Industries scope and value.
⬇ Downgrade to REDUCE if any ONE of:
- Q2 FY27 revenue below ~₹230 Cr — the execution figure binds and FY27 revenue falls year-on-year.
- The ₹750 Cr RPT headroom is passed and drawn, or Prabha Energy exposure rises further.
- Any escalation of the ONGC suspension, or its extension to live charters.
- A CARO 3(iii) qualification, or an inventory or receivable write-down in the FY26 annual report.
⚠️ SEBI: Internal analyst opinion only. Primaegis is not a SEBI-registered investment adviser or research analyst. This is research framing, not a buy or sell recommendation, and nothing here is personalised investment advice.
Provenance
Seven agents. The gate sampled 29 claims — 14 supported / 11 weak / 4 unsupported-or-failed — and rejected three of five sections. All four were revised and re-verified.
The gate's best catch, because no analyst would have found it alone. Two sections asserted that “CARE's >130-day operating-cycle downgrade trigger is already breached at 198 days.” CARE reaffirmed A+/Stable in September 2025 on FY25 financials whose cash conversion cycle was 373 days on the identical basis. No agency sets a trigger the issuer is 243 days through and simultaneously reaffirms. CARE's “operating cycle” is therefore not the cash conversion cycle, its definition was never retrieved, and the claim is withdrawn as a GAP. It had been inherited unchecked from the gathered corpus by two independent analysts.
Seven corpus-level defects were corrected at source in a governing corrections file rather than by editing raw.json, which is the evidence of record. Two mattered: the ₹352.9 Cr “FY26 adjusted PAT” on which the guidance file built its “irreconcilable” flag is derived by nobody — which is precisely why three sections adjudicated it three different ways — and is struck in favour of a derivable tax-normalised base on which FY27 guidance is +34%; and the Dolphin barge at “>₹150 Cr/year” is a source error, since ₹281 Cr over three years is ~₹94 Cr/year and the annual framing would exceed the subsidiary's entire guided FY27 revenue.
Where analysts pushed back with evidence, the desk moved. The forensic section declined the fundamental's debtor-day framing on the ground that “on the arithmetic the desk carries neither” — and it was right: stripping both legacy blocks from both years gives 138.9 → 130.7 days, neither section's original figure. Declared gaps, not filled: CARE's operating-cycle definition; the FY26 CARO 3(iii) annexure; GST-vs-book revenue (permanently unrunnable); bulk and block deals after four failed routes; the unexplained ISIN change; and the identity of anyone who transacted in the +42.3% fortnight — the register is 37 days stale and the entire move post-dates it.