Sona BLW Precision Forgings NSE: SONACOMS
A genuinely good business at a price that requires it to become a materially better one. The franchise is real — 24% operating margin against a 17% peer median, a ₹240bn order book at 5.4x revenue, net cash, IND AA+. But at 68x earnings and 7.94x book on a 10.5% ROE, the desk’s three retained valuation anchors converge on ₹540–₹640. The reason is not scepticism about growth — it is that the growth is already in the trailing numbers: the base case’s +21% FY27 revenue produces EPS of ₹11.35 against a TTM ₹11.03, +2.9%.
🚨 Recent Developments — six weeks that repriced the stock
- 22-Jul-2026 — DENSO twin-JV (Reg-30). 2W/3W EV business slump-sold to a wholly owned subsidiary for ₹8,932 mn; DENSO subscribes 49% at an enterprise value of ₹17,500 mn. Sona keeps 51% and majority board rights, so JV1 stays consolidated. A second greenfield JV2 for 4W+ high-voltage powertrain is DENSO 51% and therefore equity-accounted — its revenue never enters Sona’s top line. Reg-30 22-Jul-26
- 23-Jul-2026 — record Q1 FY27 + a robotics vertical. Revenue ₹1,310 Cr (+54%), PAT ₹181 Cr (+45%), BEV at 44% of automotive product revenue, an all-time high. A Robotics & Physical AI vertical unveiled with a ₹8bn cumulative order book, plus ₹9,800 mn of itemised automotive wins. press release 23-Jul-26
- 16-Jun-2026 — ₹626 mn robotics capex board-approved, five weeks before the vertical was disclosed publicly. Capital and orders preceded the narrative — a genuine positive on management credibility. Reg-30 16-Jun-26
- 25-Jun-2026 — Ind-Ra affirmed IND AA+/Stable, but cut rated bank limits to ₹7,250 mn from ₹9,250 mn (−21.6%), with a negative trigger at net adjusted leverage >1.5x and a base case assuming no additional debt. Ind-Ra 25-Jun-26
Street chatter check. The R11 Telegram corpus (12 groups, 23-Jul-26) circulated all of the above. Verified against primary filings: BEV 44% CONFIRMED · twin-JV CONFIRMED · DENSO 51% of JV2 CONFIRMED · 49% at ₹17.5bn EV CONFIRMED · “₹800Cr robotics” PARTIAL (a cumulative order book, not quarterly wins) · “10x by FY35” PARTIAL (an aspiration, not guidance). The circulating ₹730 Cr and ₹940 Cr order figures are refuted as robotics numbers — both are automotive.
A four-business composite with a genuine moat — and one project that has stalled
Precision-forged driveline components, electric traction and suspension motors, starter motors, railway braking and coupler systems, and radar sensors with perception software. Twelve plants across India, USA, Mexico, China, Serbia, Belgium and Germany. No listed Indian peer matches all of it — Happy Forgings earns a 30% margin on pure forging with essentially zero EV driveline content; Divgi TorqTransfer is the closest product match at one-thirteenth the revenue.peers.md
Where the revenue comes from
| Product | FY26 | Q1 FY27 | Read |
|---|---|---|---|
| Differential gears | 22% | 19% | Core franchise · global share 8.7% |
| Differential assembly | 17% | 18% | Gears + assemblies = 37%, retained 100% post-JV |
| Traction & suspension motors | 11% | 18% | +7pp in one quarter · lowest margin in the portfolio |
| Starter motors (micro-hybrid + conventional) | 23% | 20% | Comstar legacy · global share 4.2% |
| Railway (brakes + suspension/coupler) | 14% | 10% | Falling in absolute rupees — see A5 |
| Sensors and software | 2% | 1% | Novelic (Serbia) — the robotics IP seed |
Geographic mix, corrected. India 50%, North America 26%, Europe 15%, Asia 9%. The widely-circulated “Europe 26% / NA 15%” is transposed and wrong — confirmed against the rendered slide and Ind-Ra’s independent FY26 prose. Tariff exposure is roughly twice what the secondary source implies, partly offset by four plants inside USMCA.presentation Q1FY27 slide 22
The BEV denominator trap — and the fact the headline obscures
Q1 FY27 BEV revenue of ₹436 Cr is 44% of automotive product revenue — a narrower denominator. Against total consolidated revenue of ₹1,310 Cr, BEV is 33%, and only the 33% is comparable to the FY26 figure of 35%.
And FY26 BEV revenue actually declined — ₹12,235 mn to ₹11,542 mn, −5.7%, while total revenue grew 25%. The Q1 “+107%” is measured off the weakest quarter of a down year for BEV. Real, but not a clean acceleration off a rising base.presentation slide 34
What DENSO’s mark does — and does not — tell you
⚠ The “DENSO paid 2x the internal mark” claim is dead — killed at the devil gate
Three sections of the draft treated the ₹17,500 mn enterprise value against the ₹8,932 mn internal slump-sale price as ~2x third-party validation. The arithmetic kills it: 0.49 × ₹17,500 mn = ₹8,575 mn — exactly the cash leg of the consideration, to the rupee. And 0.51 × ₹17,500 mn = ₹8,925 mn ≈ the ₹8,932 mn total.
That is structural necessity, not coincidence: DENSO subscribes ₹8,575 mn of primary capital; pre-money equity is ₹8,925 mn; DENSO lands at exactly 49%; JV1 pays the parent and ends debt-free at a ₹17,500 mn equity value. For the structure to close at 49%, the cash leg had to equal ₹8,575 mn. The internal mark’s level was set by the funding design, and the valuer’s opinion then blessed a number equal to the structuring requirement.
What survives is genuinely useful, and it reads the other way. DENSO — ¥7,540 bn revenue, R&D at ~9.2% of sales — paid 4.54x the unit’s FY26 turnover for a non-controlling 49%. The group’s own multiple on the same basis is 10.68x sales. DENSO paid 42.5% of what the market pays for Sona. And the deal is value-neutral by construction — ₹857.5 Cr received equals 49% of ₹1,750 Cr given up — so the resulting ₹688 Cr equity credit that takes P/B from 7.94x to ~7.12x is fair-value recognition on an asset already owned, not a cheapening.
Strategy: “Sona 2.0” and the 10x-by-FY35 aspiration — assessedRequired CAGR 25.9% vs 26.0% delivered — but off a base 7.3x smaller
Supporting it: 10x in a decade requires a 25.9% CAGR; delivered FY18–FY26 is 26.0%. The required rate is what this company has already been doing. Order book at 5.4x revenue; net cash and IND AA+ for funding capacity; three named engines (new verticals organically and inorganically, look-east — eastern markets already 56% → 59% — and intelligent/connected systems).
The qualifier is not optional: that 26.0% was delivered off a ₹612 Cr FY18 base, on a company now 7.3x larger, and management itself attributes >85% of it to three acquisitions. Repeating 25.9% from ₹4,449 Cr means adding ~₹40,000 Cr of revenue — roughly nine times the entire current business. The rate is demonstrated; the rate applied to this base is not.
Against it: the historic engine returned ~9.8% at the PAT level (₹270 Cr on ₹2,750 Cr invested). The only forward forecast in existence is already below the required run-rate by year two (Ind-Ra: +20–25% FY27, +15–20% FY28). And one of the two headline growth vehicles is structurally excluded from the metric — JV2 is equity-accounted, so a 10x revenue aspiration cannot be delivered through it.
Verdict: the required CAGR sits inside demonstrated capability, at a much smaller scale, via a mechanism whose return record is at best neutral. Unfunded for anything transformational. Treat as ambition, score against the milestones. The word in the filing is “aspiration”; the circulating ₹35,000 Cr FY35 figure is a third-party inference and must not be attributed to management.
⚠ Project status: railway has stalled on the revenue line
The Escorts Kubota railway division — consolidated in Sep-25, and the destination of the bulk of ~₹1,800 Cr of QIP proceeds — shows revenue share falling from 14% (FY26) to 10% (Q1 FY27). Converted to rupees that is a fall from a ~₹163–208 Cr quarterly run-rate to ~₹130 Cr — between −20% and −37%, depending on which FY26 share basis is used (the slide implies 14%, Ind-Ra says 11%).
Why it matters far more than a 10% share suggests: Ind-Ra’s “ROCE is likely to gradually improve from FY27” is the sole sourced defence of a 68x multiple on an 11% ROE, and railway is its main engine. Mitigants, stated fairly: the ₹12bn railway order book is ~2.3x the annualised Q1 run-rate and sits on a 12-month PO basis, so this may be order-timing lumpiness; FY27–28 capex is directed “mainly towards railway division”, which is not how a company treats a shrinking asset; and no segment disclosure exists — this is an estimate from pie-chart shares, not a measurement.
Growth accelerating, every quality metric deteriorating — and 62% of the margin fall is unexplained
⚠ The margin break predates the explanation everyone is using
Ind-Ra attributes the FY26 step-down to “the addition of the railway division, which offers slightly lower margins” — and three of the desk’s own sections relied on it. On the company’s own quarterly series that attribution cannot hold.
| Quarter | Op profit ÷ sales | Margin | QoQ | Railway consolidated? |
|---|---|---|---|---|
| Mar-25 | 231 ÷ 865 | 26.71% | −25bp | No |
| Jun-25 | 206 ÷ 854 | 24.12% | −259bp ← largest step in the series | No |
| Sep-25 (railway consolidates, +33.3% QoQ) | 284 ÷ 1,138 | 24.96% | +84bp | Yes |
| Dec-25 | 296 ÷ 1,200 | 24.67% | −29bp | Yes |
| Mar-26 | 296 ÷ 1,258 | 23.53% | −114bp | Yes |
| Jun-26 | 293 ÷ 1,301 | 22.52% | −101bp — ten-quarter low | Yes |
The largest single step-down is Jun-25, one full quarter BEFORE railway consolidated — in a quarter when revenue fell 4.2% YoY. And the first two railway-carrying quarters printed higher margins than that break quarter. Total decline Mar-25 → Jun-26 is −419bp, of which the pre-railway step alone is 259bp = 62%.
Because the cause is unidentified, the reversal cannot be dated — which is precisely what Ind-Ra’s 23–24% band and management’s “progressively more visible from quarter 2” both rest on. Three untested candidates: operating deleverage on a shrinking quarter; the light-rare-earth magnet substitution, which management’s own “the same way for the last 5 quarters” dates precisely to Q1 FY26 — the break quarter; and tariff absorption. Settling document: the Q1 FY26 statutory expense-line breakup, locatable and unparsed.
| ₹ Cr | FY22 | FY23 | FY24 | FY25 | FY26 | TTM |
|---|---|---|---|---|---|---|
| Revenue | 2,131 | 2,655 | 3,185 | 3,546 | 4,449 | 4,897 |
| EBITDA | 560 | 675 | 902 | 967 | 1,081 | 1,169 |
| EBITDA % | 26.3 | 25.4 | 28.3 | 27.3 | 24.3 | 23.9 |
| PAT | 362 | 395 | 518 | 600 | 629 | 686 |
| ROE % | 18.1 | 17.2 | 19.5 | 10.9 | 10.5 | — |
| FCF | 101 | 198 | 374 | 360 | 180 | — |
| WC cycle (days) | — | — | 95 | 87 | 122 | — |
Organic growth — the reconciliation that replaces every single-number version
Three organic figures circulate. All three are arithmetically right and none alone is the number to model.
| Measure | Value | Arithmetic |
|---|---|---|
| FY26 organic | +8% to +12% | Revenue +₹903 Cr; railway inorganic ₹489–623 Cr |
| Q1 FY27 organic | +37% | (₹1,301 − ₹130 railway) ÷ ₹854 − 1 |
| Two-year ex-railway stack — the honest run-rate | +14.6% CAGR | ₹891 Cr (Jun-24) → ₹1,171 Cr (Jun-26) |
Why the first two differ: Jun-25 was the trough — ₹854 Cr, below Jun-24’s ₹891 Cr, with FY26 BEV revenue simultaneously falling 5.7%. FY26’s weak organic number and Q1’s strong one are the same fact seen from either side of the same trough. Neither is a run-rate.
⚠ And the base effect expires this quarter
Sep-25 already carried railway, so from Q2 FY27 reported year-on-year growth decelerates mechanically, whether or not anything goes wrong. Against that, Ind-Ra’s +20–25% FY27 band requires Q2–Q4 to average ₹1,346–1,420 Cr against Q1’s ₹1,301 Cr — sequential acceleration in every remaining quarter. The order book compounds it: the automotive book has already declined at a 0.80x book-to-bill, and 91% of the quarter’s itemised automotive wins have SOP in FY28–FY29.
Cash conversion — and the fraud filter
Verdict: CLEAN of manipulation
Over five years only ₹220 Cr — 5.3% of cumulative EBITDA — leaked to working capital and non-cash items, and 85% of that is FY26 alone. A persistent accrual problem leaks every year; this leaks in one. Capex at 1.75x depreciation implies 81% owner-earnings conversion. Derived capex reconciles to Ind-Ra's independently within 2.6%. No capitalisation game (CWIP flat, depreciation/sales falling), no revenue pull-forward, no other-income prop, and the Jun-26 QoQ PAT decline decomposes fully with zero residual.
But the FY26 deterioration is real
CFO ÷ operating profit 77% → 80% → 61%. FCF/PAT 72% → 60% → 29%. WC cycle 87 → 122 days; debtors 73 → 94; inventory 82 → 107. FY26 dividend of ₹199 Cr exceeded FCF of ₹180 Cr, funded from QIP cash. Ind-Ra attributes the inventory build to railway’s part-year revenue against a full-year balance sheet — but if railway is shrinking rather than annualising, that explanation has no unwind.
Quality scores — Piotroski 5/9 and Beneish −2.08, reconstructed and interpretedThe composition matters more than the number
Piotroski 5/9, independently reproduced from the accounts. Passes: positive ROA, positive CFO, CFO > net income (the accruals test), no new shares issued in FY26, rising asset turnover. Fails: ΔROA, Δmargin, Δleverage, and Δcurrent ratio (not determinable — current assets and liabilities are not disclosed, so scored 0; the score is 5 with a hard bound of 6).
The honest read: all three points on levels are earned; every direction-of-travel signal except asset turnover is negative. A 5 built as “levels good, trends bad” is a worse read on a 69x stock than a 5 built the other way round — because a 69x multiple is a claim about the trend. Every failure is nonetheless the signature of a heavy investment-and-dilution phase, not of accounting stress.
Beneish −2.08 against a −1.78 threshold — no flag. But report the margin honestly: Beneish is weak for acquisitive companies, and Sona’s days-sales-in-receivables index of 1.288 and sales-growth index of 1.255 both push the score up toward the threshold. A genuine pass with less margin than the headline implies. If FY27 crosses −1.78, reopen the file.
Altman Z ≈ 21.5 — no distress risk whatsoever, corroborated by IND AA+, net cash and 46x interest cover. But 89.4% of that score is the market-cap term, which is circular for valuation: a higher share price mechanically produces a “safer” Z. A valid solvency statement and an invalid valuation argument.
Thirteen quarters of P&L — full tableMargin 30% (Dec-23) → 23% (Jun-26); EBITDA flat for three quarters
| Quarter | Sales | YoY | QoQ | EBITDA | EBITDA % | PAT | Other inc. | Dep. | Tax % | EPS |
|---|---|---|---|---|---|---|---|---|---|---|
| Jun-23 | 731 | — | — | 203 | 28% | 112 | 3 | 51 | 25 | 1.91 |
| Sep-23 | 787 | — | +7.7% | 220 | 28% | 124 | 4 | 53 | 24 | 2.12 |
| Dec-23 | 782 | — | −0.6% | 233 | 30% | 134 | −0.4 | 56 | 21 | 2.26 |
| Mar-24 | 884 | — | +13.0% | 247 | 28% | 148 | 9 | 60 | 21 | 2.54 |
| Jun-24 | 891 | +21.9% | +0.8% | 249 | 28% | 142 | 9 | 61 | 25 | 2.42 |
| Sep-24 | 922 | +17.2% | +3.5% | 252 | 27% | 144 | 13 | 63 | 25 | 2.32 |
| Dec-24 | 868 | +11.0% | −5.9% | 234 | 27% | 151 | 41 | 67 | 26 | 2.43 |
| Mar-25 | 865 | −2.1% | −0.3% | 231 | 26.75% | 164 | 53 | 65 | 24 | 2.64 |
| Jun-25 | 854 | −4.2% | −1.3% | 206 | 24.08% | 122 | 32 | 67 | 26 | 2.01 |
| Sep-25 | 1,138 | +23.4% | +33.3% | 284 | 25% | 170 | 21 | 72 | 25 | 2.78 |
| Dec-25 | 1,200 | +38.2% | +5.4% | 296 | 25% | 150 | −14 | 75 | 25 | 2.43 |
| Mar-26 | 1,258 | +45.4% | +4.8% | 296 | 24% | 187 | 33 | 74 | 24 | 3.09 |
| Jun-26 | 1,301 | +52.3% | +3.4% | 293 | 23% | 179 | 35 | 77 | 26 | 2.90 |
Sep-25 is the railway consolidation — the +33.3% QoQ jump, the largest sequential move in the series and substantially inorganic. Dec-25’s −₹14 Cr of other income dragged PAT flat YoY on +38% sales; the line swings ±₹50 Cr a quarter on a ₹240 Cr PBT base because forex is netted into it. Depreciation has risen in nine of ten quarters (₹61 → ₹77 Cr) and the effective tax rate has climbed from 16% (FY22) to 26% — a silent drag that ate ~3.4pp of annual PAT CAGR.
Modelling note: Q1 FY27 depreciation of ₹77 Cr annualises to ₹308 Cr against FY26’s ₹288 Cr, with ₹300–400 Cr/yr of capex still to come. Any FY27E D&A below ~₹310 Cr is too low — an error that overstated the desk’s own first-pass EPS by ₹0.91.
The growth is already in the trailing numbers
The sharpest number in the valuation work
The base case’s +21% FY27 revenue growth produces EPS of ₹11.35 against a TTM EPS of ₹11.03 — an increase of 2.9%. Because TTM already contains three quarters of railway and the margin is still falling, the growth is already in the trailing numbers. Forward P/E is 67.3x against a trailing 69.2x. A full year of base-case delivery de-rates the multiple by under two points. The stock is not “cheap on forward” in any meaningful sense.
What 7.94x book actually requires
Solving the justified-P/B identity for the ROE that supports 7.94x, with the payout and cost of equity stated explicitly rather than assumed:
| Payout | r = 12.0% | r = 13.43% (CAPM) | r = 15.0% |
|---|---|---|---|
| 33% — FY26 actual | 16.9% | 18.9% | 21.1% |
| 50% | 21.3% | 23.9% | 26.6% |
| 70% | 30.9% | 34.6% | 38.6% |
At Sona’s actual 33% payout, 7.94x book requires an ROE of 16.9%–21.1%. The listed-history maximum is 19.5% (FY24) and the current level is 10.5%. The multiple asks the company to return to — and then hold — the very top of its own range, from roughly half that level today. (The desk’s own first-pass claim of “~24%” is withdrawn: it corresponded to a 50% payout that was never stated. Sona pays 33%.)
And the identity is numerically unstable here, so it must bound rather than price: drop ROE to the listed-history mean of 15.24% and justified P/B collapses from 7.9x to 1.56x. A 3.6pp ROE change moves it five-fold. Anyone quoting a point justified-P/B on this name is quoting noise.
Why the “below its own median” argument fails
The panel prints “PARTIAL HISTORY: 1134/1260 samples”. Sona listed in June 2021, so a 5–10 year percentile range does not exist for this name. Reconstructing the history independently:
| Date | Price | Trailing EPS | P/E | BVPS | P/B | P/S |
|---|---|---|---|---|---|---|
| Jul-2021 | ~₹433 | ₹3.76 | 115x | ₹24.01 | 18.0x | 15.9x |
| Jul-2023 | ~₹574 | ₹6.75 | 85x | ₹39.15 | 14.7x | 12.6x |
| Jul-2025 | ~₹460 | ₹9.67 | 48x | ₹88.34 | 5.2x | 8.1x |
| Jul-2026 | ₹763.75 | ₹11.03 | 69x | ₹96.20 | 7.94x | 9.70x |
Book value per share went ₹24 → ₹96 across the window. A percentile computed while the denominator quadrupled is measuring the denominator, not the multiple. The panel’s own ₹1,257 P/B “fair value” is the arithmetic consequence — 13.07 × ₹96.20 — applying a multiple earned on a ₹24–39 book to a ₹96 book. That single method is +64.6% above CMP and it sits inside the panel’s composite. Trading at 68x against a 69.5x “median” is not evidence of cheapness — the median records how expensive the stock used to be.
Where Sona trades against its peers
| Company | P/E | P/B | P/S | EV/EBITDA | ROCE | ROE | OPM |
|---|---|---|---|---|---|---|---|
| Sona BLW | 65.4 | 7.94 | 10.68 | 44.4 | 14.2% | 11.3% | 24% |
| Bharat Forge | 91.7 | 10.8 | 6.16 | 38.0 | 12.6% | 12.0% | 17% |
| Uno Minda | 55.8 | 9.9 | 3.45 | 31.3 | 19.6% | 19.3% | 11% |
| Schaeffler India | 50.4 | 10.3 | 6.52 | 35.8 | 27.3% | 20.2% | 18% |
| Endurance Tech | 42.3 | 5.8 | 2.70 | 20.8 | 17.8% | 14.9% | 13% |
| ZF CV Control | 53.7 | 7.3 | 6.58 | 41.3 | 19.4% | 14.6% | 16% |
| Sansera Engg | 60.0 | 6.5 | 5.77 | 32.9 | 14.1% | 11.5% | 18% |
| Happy Forgings | 52.4 | 7.4 | 10.14 | 34.0 | 18.0% | 15.0% | 30% |
| Ramkrishna Forg | 99.2 | 3.5 | 2.69 | 22.2 | 5.6% | 2.56% | 15% |
| Divgi TorqTransfer | 62.0 | 4.6 | 8.24 | 41.6 | 10.2% | 7.62% | 20% |
| Peer median (9) | 55.8 | 7.3 | 6.16 | 34.0 | 17.8% | 14.6% | 17% |
The largest premium is on the one multiple that gets no QIP excuse
P/E is +17% and P/B +9% above the peer median — precisely because their denominators are depressed by the ₹2,400 Cr QIP still being absorbed. But P/S is +73.4% above the median, and sales have no such excuse. Remove it and Sona is priced at the top of its cohort — on a ROCE 20% below the median and a ROE 23% below it. The premium is earned on three things: a 24% margin against a 17% median, the only disclosed multi-year order book in the set, and net cash against a levered cohort. It is not earned on returns.
What the price implies — reverse-DCF
| Exit P/E in 2036 | Required FY36 EPS | Implied 10-yr EPS CAGR |
|---|---|---|
| 68.24x — today’s, i.e. no de-rating ever | ₹39.36 | +14.6% |
| 55.8x — peer median | ₹48.14 | +16.9% |
| 42.3x — cohort floor | ₹63.50 | +20.2% |
| 25x — a mature auto-ancillary | ₹107.44 | +26.7% |
₹763.75 implies a 10-year EPS CAGR of 14.6% to 20.2% (or a 28–37% free-cash-flow CAGR on the FCFF frame). The only sourced forward numbers produce +12.3% EPS in FY27. The gap is closed only by the 10x-by-FY35 aspiration — which requires a 25.9% CAGR, which management labels an aspiration, whose historic mechanism returned 9.8%, and whose funding the sole rater has modelled as unavailable. Note also that FCF yield is 0.38%–0.79%: the equity does not currently generate enough free cash to be valued on cash flow at all.
De-rating alone — no earnings miss required
| De-rate to | Multiple | On FY27E EPS ₹11.35 (growth delivered) | vs CMP | On TTM EPS ₹11.03 (de-rating alone) | vs CMP |
|---|---|---|---|---|---|
| Cohort median | 55.8x | ₹633 | −17.1% | ₹615 | −19.4% |
| Happy Forgings — best process comp | 52.4x | ₹595 | −22.1% | ₹578 | −24.3% |
| Own band-low P/E | 51.29x | ₹582 | −23.8% | ₹566 | −25.9% |
| Endurance — cohort floor | 42.3x | ₹480 | −37.1% | ₹467 | −38.9% |
With earnings delivered exactly to the Ind-Ra-consistent base case and no operational disappointment, multiple compression to the peer range produces −17% to −37%. With no growth at all, −19% to −39%.
Scenarios — the desk’s own grid
| BEAR | BASE | BULL | |
|---|---|---|---|
| Probability | 30% | 45% | 25% |
| Revenue FY27 | +14% → ₹5,072 Cr | +21% → ₹5,383 Cr | +27% → ₹5,650 Cr |
| Q2–Q4 avg required | ₹1,257 Cr (−3.4% vs Q1) | ₹1,361 Cr — +2.3% QoQ every quarter | ₹1,450 Cr (above Ind-Ra’s ceiling) |
| EBITDA margin | 22.0% | 23.0% (Ind-Ra floor) | 24.0% (Ind-Ra ceiling) |
| Implied incremental margin (observed: −2.5%) | 5.6% — more generous than observed | 16.8% — needs 19pp of improvement | 22.9% — needs 25pp |
| FY27E EPS | ₹9.95 | ₹11.35 | ₹12.65 |
| Exit multiple | 38.6x (observed Oct-2025) | 52.4x (Happy Forgings) | 72x |
| Reference level | ₹384 | ₹595 | ₹911 |
| vs CMP ₹763.75 | −49.7% | −22.1% | +19.3% |
Probability-weighted 12-month reference level: ₹611 (−20.0%). 24-month: ₹722 (−5.5%). Sensitivity tested across six single-input changes, the weighted level sits in ₹611–₹652 — no single input moves it to the current price.
The shape matters as much as the level — this is a barbell, not a slope
The TradingView grid’s modal 40% case is a mild decline, and its entire positive expected value is carried by a 39% bull tail. Strip the bull leg and the remaining 61% of probability mass averages ₹601, or −21.3%. The investor is not being offered a gently rising base case with tail risk — they are being offered a coin-flip between a −8% grind and a +54% re-rating. Note also that the same panel prints a normalised fair value of ₹622.29, 18.5% below the current price, while assigning 39% to +54.2%. Those two outputs are not reconcilable.
Withdrawn at the gate: the desk’s own sum-of-the-parts, in full — it applied a multiple to whole-group earnings and then added an external enterprise value for a unit already inside those earnings, double-counting the eDrive business; and its “peer-median 34x” was a mislabelled EV/EBITDA, not a P/E. The desk’s DCF is reported as a diagnostic, not an anchor: it cannot reach the current price at any sourced growth rate.
Five risks that can impair the thesis on their own
Consolidated from 20 entries to 15 at the devil gate — the undisclosed shareholders’ agreement appeared seven times across the pack, the promoter litigation six, the compliance-officer churn seven. A reader would have counted 20 risks where there are about nine.
| # | Risk | Sev. | Early-warning marker |
|---|---|---|---|
| 1 | Valuation and the re-rating that produced it. 68x on a 10.5% ROE; the month’s ~₹8,950 Cr of market-cap addition is several times the aggregate disclosed value of everything new. PEG 3.8–5.6 | HIGH | ROCE fails to exceed 18% and ROE 17% by FY28 |
| 2 | Margin structurally lower, and 62% of the decline unexplained. Ten-quarter low of 22.5%, below Ind-Ra’s 23–24% band on the comparable basis; the largest step predates railway; cause unidentified ⇒ reversal undatable | HIGH | Q2 FY27 margin ≤22.5%; VA/employee cost 5.8→4.9→4.4 is the only company ratio still deteriorating |
| 3 | Railway appears to be shrinking — destination of ~₹1,800 Cr of QIP capital and the engine of the ROCE-recovery case, on a 12-month-PO book of ₹12bn | HIGH | Q2 FY27 railway share at or below 10% again |
| 4 | The undisclosed SHA/JVA. Reserved matters could neuter the nominal 51%; a DENSO put would be a fair-valued liability plus a ~₹857 Cr contingent cash call; loss of control forces a ≈₹1,403 Cr one-off P&L gain — 223% of FY26 PAT, presentational not economic | HIGH | No postal ballot by Q3 FY27; the FY27 AR note on control over Sona eDrive |
| 5 | Customer concentration the desk cannot size — top 5 at ~50% of revenue, an Ind-Ra rating weakness, all customer names refused | HIGH | A fall in the “20 fully ramped” EV programme count |
| 6 | The automotive order book has already declined ₹235bn → ₹232bn at 0.80x book-to-bill; 91% of Q1 itemised wins have SOP in FY28–FY29; honest organic run-rate 14.6% | MED | A second consecutive quarterly decline |
| 7 | North America 26% of revenue, tariff-exposed, with zero specific management commentary and no analyst question. Four USMCA plants mitigate; the exposure is unquantified in both directions | MED | NA share below ~22%; margin commentary naming duties |
| 8 | M&A returns thin while Sona 2.0 doubles down. ~9.8% at PAT level. Headroom to the 1.5x trigger is ~₹4,200–4,570 Cr — but rated facilities total only ₹725 Cr, so that is covenant arithmetic, not availability | MED | Any acquisition above ~₹1,500 Cr; any new rated facility |
| 9–15 | Working-capital normalisation unproven · s.50B slump-sale tax of ~₹137 Cr unsized by the company · rising effective tax rate · JV1 minority interest and JV2 equity-method drag · treasury income declining · promoter-block litigation (downgraded HIGH→MED at the gate — the listed entity is not a party and no order reaching its Sona shares is in evidence) · key-man risk · information deficit | MED | See full risk matrix in the desk file |
The “FII exodus” is an index event, not an opinion
✓ The pre-flagged −9.7pp FII decline is largely mechanical — and now quantified
- 98.8% of the entire decline sits in two quarters — Mar-25 (−3.34pp) and Sep-25 (−6.28pp). The other ten quarters net to −0.12pp.
- Both were falling quarters. Across the entire subsequent +62.5% advance, FIIs were net +0.14pp — there is no distribution-into-strength pattern.
- The cause is dated and external: Sona BLW was removed from the MSCI India Standard Index effective 26-Aug-2025, moving to the Smallcap index, with an estimated ~US$163 mn (≈₹1.4k Cr) of passive outflow against the desk’s own sizing of the move at ~₹1,600 Cr. That closes 85–90% of the quarter.
- The marquee holders did not leave: GIC 6.84% → 6.54%, MAS 1.66% → 1.56%, Norges newly in the top three, and the FPI investor count actually rose 312 → 321.
- Blackstone is not in the FII line at all — Singapore VII Topco III was a co-promoter and fully exited its 20.50% on 13-Mar-2023 at ₹410, eighteen months before the window opens.
- Retail is the seller into strength: public float 8.37% → 6.77%, shareholder count −15% since Sep-25. The one quarter retail added was the crash quarter.
Verdict: smart money is NEUTRAL, drifting mildly in, with a wholesale change of nationality. The structural consequence to carry: free float 72% (non-promoter) but non-institutional public float only 6.77% — downside liquidity now depends on institutions trading with each other, and crowding risk has migrated from foreign to domestic hands.
Governance — what the DENSO deal did and did not discloseLegally clean, disclosure-poor — and one genuinely large unknown
No breach, and the legal tests are verifiable from the filing’s own figures. The slump sale is 6.0% of standalone net worth and 9.3% of turnover, so it is not an “undertaking” under s.180(1)(a); and Reg 23(5)(b) exempts holding-company-to-wholly-owned-subsidiary transactions from the audit-committee and shareholder-approval requirements. Credit where due.
But a landmark ₹893 Cr carve-out was executed with no shareholder vote, no NCLT scrutiny, an unnamed valuer, no published valuation and no fairness opinion — the structure lawfully routes around both a vote and the majority-of-minority protections a scheme would have carried. On a transaction of this significance the silence on the fairness process is itself the finding.
The audit-committee timing flag, with its mitigant. A Non-Executive, Non-Independent director was inducted on 22-Jul-26, the same day as the DENSO resolutions. Mitigants: the committee is 75% independent against a two-thirds floor, independent-chaired, and the Reg 23(2) proviso means only independent directors may approve related-party transactions — so she cannot vote on the JV1 royalty RPTs. Compliant, and poor optics. Both halves belong in the record.
The largest unknown, and it is unverifiable by construction: the SHA/JVA reserved matters, put/call options and deadlock mechanics are not public. Reserved matters could reduce Sona’s nominal 51% to protective rights and force deconsolidation; a DENSO put would be a financial liability under Ind AS 32/109. The desk cannot independently verify the control assertion on which the entire consolidation case rests.
Two things the gate corrected in the company’s favour. (i) The “three compliance-officer changes in three months” framing is withdrawn — the record is one unplanned senior-legal exit plus a two-month interim bridge that resigned explicitly “consequent to the appointment of a permanent appointee”, resolved with an M&A specialist installed five weeks before the DENSO signing. Deal preparation, not distress. (ii) The claim that the promoter-estate injunction “mechanically precludes both a sale and a pledge” is withdrawn as legal overreach — the order runs against estate and Aureus-level assets, one layer above Sona’s register, and no primary order reaching Aureus’s Sona shares is in evidence. Forward diary item: mandatory auditor rotation falls due at the 31st AGM in 2027 — an incoming auditor will inherit the Ind AS 110 control judgement on Sona eDrive.
Trend quality A−. Entry quality at ₹763.75, F.
✓ Resolved at the devil gate — ₹768 is a genuine all-time high
The draft flagged the pre-Aug-2025 high as “the most important unresolved level in the analysis”, with a P/E back-out suggesting the 2024 cycle high sat somewhere in ₹690–780 — leaving open the possibility that ₹768 was the underside of a 2024 distribution top. The gate resolved it: the prior all-time high is ~₹741.65 and the 2024 cycle high was ₹718.55 intraday on 10-Sep-2024, the QIP day. Both sit below ₹768, and both fall inside the band the P/E back-out predicted — two independent methods agreeing. So this is a breakout into clean air, not into old supply. And the ₹700–742 prior-cycle-high zone becomes a fifth method inside the support confluence — resistance turned support. (Sourcing is secondary and single-source; a monthly chart settles it definitively, but the conclusion holds under either candidate high.)
The central tension — resolved, not averaged
Bullish — and it is real
- Stage 2 Advance confirmed, price above a rising 30-week MA by 18–22%
- +62.5% over 12 months against a NIFTY that fell 2.97% — +65pp of alpha
- RS +9.09% above its own RS moving average, “strongly outperforming”
- Breakout on 4.0x rupee turnover (₹594 Cr vs a ₹148 Cr 50-day average), closing in the top of its range
- A textbook high-tight flag beneath it: five sessions of closes within ₹0.71
- Four of nine peers within 7% of 52-week highs — narrow leadership, but leadership
Cautionary — and not resolvable by one session
- 6.8 ATR above the 50-day mean — climactic-extension territory
- Volatility 42% above its own one-year norm (ATR 2.8% vs 1.97%)
- “Cheat Entry: No — not in pullback zone”
- Base health: largest-spread and highest-volume bars inside the base were both down bars
- The setup’s own backtest: 29% win rate, avg −2.6%, n=7 — a 95% confidence interval spanning 0% to 62%
- The model’s own 40%-probability modal case (₹705) sits below its own stop (₹709.80)
The momentum readings resolve; the extension readings do not. “Momentum weakening” provably describes the 23–29 July flag — the four sessions before the breakout netted +0.10% — and was overtaken by the tape. But 6.8 ATR of extension is a current-state fact about where price sits relative to its own mean, and it resolves only through time or through price. The honest verdict is a genuine Stage 2 advance and a bad place to open a position — two separable judgements that must be reported separately.
Levels, and the stop-tolerance arithmetic
| Level | ₹ | vs CMP | Significance |
|---|---|---|---|
| Band high | 1,209 | +58.3% | Model output |
| Position Planner T3 / T2 / T1 | 979.55 / 898.65 / 844.70 | +28% / +18% / +11% | Arithmetic (4.0R / 2.5R / 1.5R), not structural — zero structural information |
| Fair-value line | 862.28 | +12.9% | Model output |
| 52-week high — genuine ATH | 768.00 | +0.56% | The pivot. Clean air above. |
| CMP | 763.75 | — | Top of the 52-week range (98.8%) |
| Prior all-time high — resistance turned support | 741.65 | −2.9% | The fifth method, and the entry that fits the risk rule |
| Flag shelf / fast EMA / 2024 cycle high / composite value / ATR stop | 720.50 · ~721 · 718.55 · 717.63 · 709.80 | −5.7% to −7.1% | Four-method confluence — where the setup lives or dies |
| 50-DMA region | ~618–653 | −14% to −19% | Stage-2 integrity zone |
| Band low / 200-DMA region | 582 · ~567–606 | −24% to −26% | Structural fail-safe |
⚠ At ₹763.75 the ≤3% stop rule and full-capital deployment are mathematically incompatible
| Stop | Level | Distance | in ATR | in ADR(20) | Survivable? |
|---|---|---|---|---|---|
| Ideal 1% | 756.11 | ₹7.64 | 0.36 | 0.34 | No — intraday noise |
| Max 3% | 740.84 | ₹22.91 | 1.07 | 1.01 | No — one average day’s entire range |
| Indicator ATR×2.5 | 709.80 | ₹53.95 | 2.51 | 2.38 | Yes — but 2.4x the rule |
A 3% stop sits exactly one average daily range below the close. It would very likely be hit by ordinary noise within days without the setup having failed in any way — a full loss with zero information gained. Note the trap: a tighter stop improves nominal R:R to 3.53:1. That is an illusion — tightening inside the noise band raises the ratio and collapses the probability faster. Better R:R, worse expectancy.
✓ The one setup that fits the risk rule — created by resolving the all-time high
Entry A′ — a shallow retest of the old all-time high. This did not exist in the original draft, because ₹741.65 was not known to be a level.
| At today’s close | Entry A′ — on a −2.9% retest | |
|---|---|---|
| Entry | ₹763.75 | ~₹748 (reclaim/hold above the old ATH) |
| Stop | ₹709.80 | ₹734 — just below ₹741.65, structural not arbitrary |
| Risk | 7.06% — 2.4x the rule | 1.87% — inside the rule, near the 1% ideal |
| R:R to ₹844.70 | 1.50 : 1 | 6.91 : 1 |
| R:R to the fair-value line ₹862.28 | 1.83 : 1 | 8.16 : 1 |
If ₹741.65 fails, the breakout has failed — which is exactly what a stop should measure. Why it may not come: a shallow 3% retest is the least likely pullback depth after a +6% breakout bar on 3.37x volume; the stock may keep going, or overshoot to the ₹710–723 confluence (where a reclaim of ~₹727 with a ₹706 stop is 2.89% risk and 5.6:1). What does NOT create a valid entry is a further advance — chasing is the one action that cannot improve the arithmetic.
Invalidation is a close below ₹720.50 — the flag shelf. Below it, the 30-July bar is retrospectively a one-day blow-off and the five-session base is void. Loss of the ~₹618–653 50-DMA region on expanding volume is a Stage 2 → Stage 3 character change, thesis-level rather than trade-level. Next earnings 28-Oct-2026 — 89 days, an event and not a thesis.
For an existing holder this is a different problem entirely. The desk’s own prior signal was a BUY at ₹484.65 on 14-Oct-2025, now +54% to +58% marked to today. A holder is risking open profit, not capital: the ₹741.65 and ₹720.50 references are legitimate trailing stops that still retain +53% and +48.6% of gain. An entrant deciding whether to risk 7.06% of capital 0.56% below resistance is not the same trade and should never get the same answer.
What the adversarial reviewer found
Every section of this report was attacked before publication. The gate returned REJECT-SECTIONS with 41 required fixes: 8 claims supported, 9 weak, 6 unsupported, and 6 of 10 checked numbers failed. Four of the five analyst sections were rebuilt from scratch; the other two were corrected in place. The gate is a feature, so here is what it caught.
Killed
- The “DENSO paid 2x the internal mark” inference — a structuring artefact, not validation
- A sum-of-the-parts that double-counted the eDrive unit and used a mislabelled multiple
- A justified-P/B of “~24% ROE required” that silently assumed a 50% payout on a 33%-payout company
- An Ind-Ra FCF “bridge” that was arithmetically wrong and double-counted land already inside capex
- A ₹15 Cr royalty booked as incremental bull-case profit — it eliminates on consolidation
- A starred “24-month value equals today’s price” coincidence that moved ±6% on trivial inputs
Corrected in the company’s favour
- ₹768 confirmed as a genuine all-time high — clean air, and it created the only rule-compatible entry
- The −9.7pp FII decline quantified as an MSCI index event (~US$163mn), closing 85–90% of it
- Promoter litigation downgraded HIGH → MEDIUM — the listed entity is not a party
- The “three CS changes in three months” flag withdrawn as an overstated bridging arrangement
- The audit-committee flag given its Reg 23(2) mitigant — the inducted director cannot vote on the RPTs
- Risk list consolidated 20 → 15; the SHA appeared 7 times, the litigation 6, CS churn 7
☠ Devil’s Objections — the five strongest bear arguments
- The margin break predates railway, so the “structural mix” story is only half-sourced — and the unexplained half is 300bp. EBITDA margin went 27% (Mar-25) → 24% (Jun-25) before the Escorts Kubota division consolidated in Sep-25. Ind-Ra’s railway attribution — relied on by three of the five analyst sections — cannot explain the single largest step-down in the series. Something took 300bp out in a quarter when revenue fell 4.2% YoY, and because the desk has not identified it, it cannot date its reversal either.raw.json quarterly · Ind-Ra 25-Jun-26
- The honest organic run-rate is ~15%, not 37% or 52% — and the base effect expires this quarter. Ex-railway the two-year stack is ₹891 Cr → ₹1,171 Cr = +14.6% CAGR. Sep-25 already carried railway, so reported growth decelerates mechanically from Q2 FY27 — while Ind-Ra’s +20–25% band requires Q2–Q4 to average ₹1,346–1,420 Cr against Q1’s ₹1,301 Cr. The order book compounds it: the automotive book has already declined ₹235bn → ₹232bn at 0.80x book-to-bill, and 91% of the quarter’s itemised wins have SOP in FY28–FY29.presentation slides 17/19/20
- Railway — the destination of ~₹1,800 Cr of QIP capital — appears to be shrinking in absolute terms, and nobody in the pack computed it. Share fell 14% → 10%; the railway order book is ₹12bn on a 12-MONTH PO basis, not the 10-year auto basis. Implied run-rate ~₹208 Cr/quarter against ~₹130 Cr in Q1. If railway revenue is falling, the “ROCE improves from FY27” case — the entire defence of an 11% ROE at 68x — loses its main engine.slide 22 · slide 20
- The DENSO deal creates no net value at DENSO’s own price, and the pack banked it three ways as if it did. ₹857.5 Cr received = exactly 49% × ₹17,500 mn given up — an identity exact to the rupee. The ₹1,750 Cr “mark” is not new value; the ₹688 Cr equity credit taking P/B to 7.1x is fair-value recognition on an asset already owned, not a cheapening; and the base-case s.50B tax on the ~₹5,461 mn surplus (~₹137 Cr) is unsized anywhere, reducing both the cash inflow and the pro-forma net cash the M&A headroom rests on.Reg-30 22-Jul-26
- The 10x aspiration has no visible funder, and the lender lines were just cut. Sona 2.0’s mechanism is explicitly “organically and inorganically”; management concedes “more than 85% of that growth came from just 3 strategic decisions” on which the return is ~9.8% at PAT level; Ind-Ra’s base case assumes zero new debt with limits cut to ₹7,250 mn from ₹9,250 mn. The ₹4,300–4,700 Cr of covenant headroom is arithmetic, not availability — at ₹725 Cr of rated facilities, a deal that size requires the rating exercise the negative trigger describes.Reg-30 23-Jul-26 · Ind-Ra 25-Jun-26
A good business at a price that requires it to become a better one
The franchise is real and not in dispute: 24% operating margin against a 17% peer median, an order book of ₹240 bn at 5.4x revenue that no listed Indian peer discloses an equivalent to, net cash with 46x interest cover and an IND AA+ rating, and a BEV share of automotive product revenue at an all-time-high 44% on a structural power-source-neutral hedge. On 22–23 July the company added a DENSO twin-JV and a robotics vertical, and the stock has gone from ~₹620 to a genuine all-time high.
But the price is 68x earnings and 7.94x book on a 10.5% ROE, and the desk’s three retained valuation anchors converge on ₹540–₹640. The reason is not scepticism about growth — it is that the growth is already in the trailing numbers. Meanwhile de-rating alone, with earnings delivered and no operational disappointment, gives −19% to −39%.
C4 · Alignment: DIVERGING
The fundamentals and the technicals disagree about the price, not the business
Both sides agree the franchise is good and the July catalysts were real. They part company on whether ₹763.75 is a sensible price to own it at. Valuation says the market is paying 68x for a business whose margin is at a ten-quarter low, whose returns have halved, whose automotive order book has started shrinking, and whose only sourced forward forecast delivers +2.9% EPS against trailing. Technicals say the tape is in a confirmed Stage 2 advance with 65pp of alpha, breaking to genuine all-time highs on 4x volume.
What that means for timing — three consequences:
- A divergence of this shape usually resolves through time rather than price. The 24-month weighted reference level of ₹722 (−5.5%) against a 12-month ₹611 (−20%) is exactly that shape — the gap closes more through delivery than through collapse, but not fast enough to make ₹763.75 a good entry.
- The near-dated event is binary and close: 28-Oct-2026. Q2 FY27 tests three of the four load-bearing bear findings at once — margin recovery, whether the automotive book declines a second consecutive quarter, and whether railway is genuinely contracting. Eighty-nine days is a short wait for that much information.
- For anyone acting on the technicals, the arithmetic is unforgiving at this price and workable 2.9% lower — Entry A′ at ~₹748 with a ₹734 stop is 1.87% risk and 6.9:1. That is the only circumstance in which the desk’s rating and the tape can both be respected.
C5 · Catalysts to watch
| Date | Event | What it tests |
|---|---|---|
| 28-Oct-2026 | Q2 FY27 results | Three bear findings at once: margin ≥24.0% vs ≤22.5% · automotive book-to-bill back above 1.0x vs a second decline · railway share ≥12% vs a second contraction. Plus revenue ≥₹1,346 Cr, which Ind-Ra’s own band requires |
| ~Q3 FY27 | Shareholder-approval notice / postal ballot | The likely first public window into the SHA terms and the royalty rate |
| On/before 31-Mar-2027 | JV1 completion; ₹8,575 mn received | Net of the ~₹137 Cr s.50B tax the company has not quantified. An extension announcement is the first slippage signal |
| Q2–Q3 FY27 | First robotics revenue line | A falsifiable near-dated test of the only growing part of the order book |
| ~May–Jun 2027 | FY27 results + Annual Report | ROCE >15.4%, WC cycle <100 days, FCF >₹360 Cr — and the segment disclosure that would settle railway and the margin gap |
| 2027 (31st AGM) | Mandatory auditor rotation | An incoming auditor inherits the Ind AS 110 control judgement on Sona eDrive |
C6 · Watchlist verdict: MONITOR. Not TRACK — the desk does not want a position at this price and the valuation gap is too wide to work into on ordinary weakness. Not PASS — the franchise is good, the order book is real, the technical leadership is intact, and there is a dated near-term event that could change the answer. MONITOR, with the 28-Oct-2026 print as the scheduled re-underwrite.
⬆ Upgrade to NEUTRAL if, by the Q2 FY27 print on 28-Oct-2026, any two of: EBITDA margin ≥24.0% on the operating-profit basis · revenue ≥₹1,346 Cr, meeting Ind-Ra’s own floor arithmetic · automotive + railway book-to-bill back above 1.0x · railway revenue ≥₹165 Cr, refuting the contraction. Upgrade to ACCUMULATE only if all four land and the price has corrected into ₹595–₹640 — the business proves the recovery and the market has already paid for the disappointment.
⬇ Downgrade to SELL if, by 28-Oct-2026 or the FY27 results in May-2027, any two of: EBITDA margin ≤22.5% for a second consecutive quarter · a second consecutive quarterly decline in the automotive order book — the first has already fired · railway revenue below ~₹130 Cr again · Ind-Ra revising the FY27 band down or moving to Negative · a debt-funded acquisition requiring facilities above the ₹725 Cr rated limit · or SHA disclosure revealing a DENSO put or reserved matters that force deconsolidation.
Rationale. The valuation is stretched on the desk’s own work — three retained anchors converge on ₹540–₹640, and de-rating alone with earnings fully delivered produces −19% to −39%. That alone justifies NEUTRAL. What takes it to REDUCE is that the deterioration is not only in the multiple: margin at a ten-quarter low with 62% of the decline unexplained and therefore undatable, ROE halved and not recovered two years post-QIP, the automotive order book already declining at 0.80x book-to-bill, and railway — the QIP destination and sole sourced engine of the ROCE-recovery case — apparently contracting. The probability-weighted 12-month reference level is ₹611 (−20.0%) and does not reach the current price under any single-input change tested. Conviction is MODERATE, not HIGH, because the technical trend is genuinely strong and can persist past fair value; the two sharpest bear findings both rest on inferred segment data that a single disclosure could overturn; and 28-Oct-2026 is only 89 days away and can validate the bull outright.
Holder (desk’s prior BUY at ₹484.65, now +54–58%): trim into strength, not exit into weakness. ₹741.65 and ₹720.50 are legitimate trailing references retaining +53% and +48.6%.
New buyer: do not initiate at ₹763.75, where the only structural stop is 7.1% away and R:R to the nearest real level is 0.08:1. The single defensible entry is a −2.9% retest of ₹741.65 — and even that is a trade against the desk’s valuation view, taken on technical structure alone and sized accordingly.