Lookback: How PSU-Defence Stocks Traded from Euphoria to Correction in 18 Months

The order-book story carried the pack to a ₹12 lakh crore market-cap expansion. The margin story took back nearly a third of it before the year was out.

The PSU-defence trade between March 2024 and September 2025 will be studied for a long time, not because it was unusual in shape, but because it was so faithful to the template of every Indian thematic cycle that came before it. There was a credible policy tailwind. There was a measurable order pipeline. There was a brokerage cohort willing to put a multiple on the pipeline rather than the earnings. And there was, eventually, the moment where execution slipped behind narration. By the close of the September 18, 2025 session, the pack had spent five clean months grinding lower, and the conversation in dealing rooms had quietly shifted from "how high" to "how much further down".

The story did not start in March 2024. It started two budgets earlier, when the indigenisation push under the Defence Acquisition Council began converting line items into actual order placements at Hindustan Aeronautics, Bharat Electronics, Mazagon Dock, Cochin Shipyard, Garden Reach, Bharat Dynamics and BEML. By the spring of 2024, with a general election in the air and the strategic-autonomy narrative running through every TV panel, those order books had hit a size that retail screeners could no longer ignore. That was the trigger. The sustain came from a combination of three things: a defensible export-order story, a domestic capex cycle that finally looked real, and a flow imbalance where the buying community kept growing while the float, in classic PSU fashion, did not.

Caption: The eighteen-month arc on a single weekly frame, showing how the parabolic late-2024 leg gave way to a stair-step decline through the second half of 2025.

The cycle anatomy split cleanly into three acts. The first ran from March 2024 into the post-result melt-up of June and July, when the election outcome, despite being narrower than the street had positioned for, was read as continuity for capital expenditure on defence. Volumes in the pack were heavy but orderly, with delivery-based buying from domestic mutual funds and the non-FII institutional book absorbing whatever paper the QIP route brought to market. HAL and BEL did the heavy lifting in this leg, with the shipyards and the missile names rotating in behind them. This was the phase where the multiple expansion was, at the time, defensible. Order-inflow guidance was being raised, and execution was tracking.

The second act, from August 2024 into late January 2025, was the parabolic phase, and it was the part that should have given the long-only community pause. Price moves decoupled from any incremental change in fundamentals. The smaller names in the pack, the ones with thin floats and concentrated promoter holdings, ran the hardest, which is the classic late-cycle signature of any thematic. New retail accounts on the discount brokers became a measurable share of daily volume in BEML, GRSE and Bharat Dynamics. Brokerage notes through this stretch did what brokerage notes do near tops: they raised target prices to validate the spot, framed multiples on FY27 and FY28 earnings, and reached for the export option value as the residual to bridge the gap between fundamental and screen. The blow-off high for the pack arrived in the back half of January 2025, with the Nifty India Defence index posting its highest weekly close of the cycle in that window.

The third act, the unwind, took longer to start than the bears wanted but, once it started, behaved exactly like a retail-heavy theme deflating into supply. The first crack came with the Q4 FY25 results in May 2025, where a handful of names in the pack delivered execution misses and margin compression, and a few of the larger ones flagged that the order-to-revenue conversion was going to be slower than the street had assumed. The second crack was a flow event. The FII book, which had only been a marginal participant in the rally, treated the May results as the cue to take the chips that had been put on the table at much lower levels. The DII book, dominated by mutual funds whose defence-themed schemes had absorbed heavy inflows through the rally, found itself running out of fresh subscription money to defend the prints. By July, the pack was making lower highs and lower lows on the daily, and by mid-September the market-cap give-back had reached the rough one-third figure that the press was repeating in the lead-up to the September 18 close.

Caption: The corrective phase in detail. Each rebound attempt failed at a lower level, with the late-July rally failing first and the September drift confirming the trend.

Leadership inside the pack rotated in a way that, in hindsight, was a useful signal. The early phase belonged to the engineering and electronics names, with HAL and BEL leading on relative strength because both had the cleanest combination of order book, execution track record, and free-cash visibility. Through the middle of the cycle, leadership shifted to the shipyards. Mazagon Dock, in particular, ran with a momentum that detached from its trailing earnings, helped by the long-tail visibility of the submarine and warship orders. By the parabolic phase, leadership had migrated to the smallest, most operationally levered names, which is almost always the warning. When BEML and Bharat Dynamics started delivering the daily percentage gains that HAL and BEL had delivered six months earlier, the pack had stopped being a fundamentals trade and become a flow trade. In the unwind, the order reversed cleanly. The smallest names cracked first and hardest, the shipyards followed, and HAL and BEL gave up the least in percentage terms while still surrendering meaningful absolute value.

Cross-sector comparison through this window made the defence story look less unique than its supporters argued. PSU-defence shared the eighteen-month period with a number of other thematic pockets that ran on similar logic, namely PSU power, capital goods and railways. PSU power, anchored by NTPC and Power Grid, moved on the energy transition and the capex narrative, peaked in the same January-February 2025 window, and corrected through similar machinery. The railway pack, with IRCTC, RVNL, IRFC and the wagon makers, delivered an even sharper version of the cycle, with a higher beta on the way up and a faster unwind on the way down. The capital-goods names, more diversified and less retail-saturated, gave back less. What lagged and what led was useful. IT services, weighed down by a soft global discretionary environment for most of the window, lagged for almost the entire eighteen months. Private banks delivered a flat-to-modestly-positive arc, doing nothing exciting but also avoiding the round trip. FMCG and pharma did defensive duty in the second half, with selective rotation flowing into both as the thematic packs unwound.

The macro context through the window did not make the unwind any easier. The RBI through 2024 had held its policy rate steady before beginning a measured cutting cycle into early 2025, which kept liquidity comfortable enough for the rally to extend, but the global rate environment, the dollar index and crude prices all introduced second-order pressure on the FII book through the second half of the window. The general government continued to flag defence indigenisation in budget speeches and policy releases, but by mid-2025 the question on the desks of institutional buyers had stopped being whether the orders existed and started being whether the order book could be converted to cash flow at the margins that the spot multiples assumed. That is a question every theme eventually faces, and it is rarely answered in the bulls' favour at peak multiples.

Caption: The intraday character at the end of the window. Each attempted gap-up was sold into through the day, with the closing-hour prints marking the weakest part of the session.

Brokerage behaviour through the cycle was a textbook of incentive and timing. Through the first leg, the upgrades were defensible, with target-price increases tracking actual order-book additions and execution improvements. Through the parabolic phase, the upgrades became valuation-extension exercises, where rolling-forward to FY27 and then FY28 earnings was used to keep the multiple from looking absurd at the spot. By the time the May 2025 results landed, the same houses began trimming targets, downgrading select names from buy to hold, and writing reports that politely acknowledged that execution was now the variable to watch. By August and September, the downgrade flow was steady, and the language in the notes had shifted to "watch for stabilisation in margin trajectory before adding". The cycle of recommendations lagged the price action, as it almost always does, which is its own lesson for anyone using brokerage consensus as a primary signal.

The historical analog that maps best to this cycle is the public-sector capital-goods and infrastructure run that played out between 2006 and 2008. That earlier cycle had a similar combination of a credible long-cycle order narrative, a flow regime that pushed multiples well past trailing earnings, and a retail participation rate that escalated through the late stages. It also unwound on the same combination of execution slippage and the unwinding of the flow regime that had supported the multiple. The defence cycle of 2024-25 was less violent on the way down than the 2008 unwind, partly because the macro backdrop was kinder and partly because the domestic institutional book, larger and more sticky in 2025 than it was in 2008, slowed the descent. But the shape was familiar, and anyone who had traded the earlier cycle would have recognised the rotation through the leadership pack and the late-cycle widening of participation as the signals they were.

What was structurally different about this cycle, and worth noting for the next one, was the role of the retail discount-broker book. In 2008, the marginal incremental buyer in the late phase of a thematic was either a high-net-worth individual through a full-service broker or a small-cap-focused mutual fund. In 2025, the marginal incremental buyer was a retail account on a discount broker, often investing through systematic transfer plans into thematic mutual funds and, separately, holding direct equity positions in the same names. This created a flow correlation that, on the way up, amplified the rally, and on the way down, amplified the redemption-driven selling pressure on the same names. The asset-management industry's defence-themed schemes, which had launched in number through 2024 and absorbed sizable inflows, became net sellers as redemption pressure built into the second half of 2025. That was a layer the earlier analog did not have, and it is the layer that thematic investors will need to model into the next cycle.

Two qualitative observations on the September 18, 2025 session itself are worth recording, because the close that day did not look like a panic. It looked like resignation. The pack opened mixed, attempted a recovery in the first hour, faded through the late morning, and closed near the lows in a session where volumes were no longer matching the daily averages from the rally peak. The fear gauge was not extreme. The intraday option chains in the larger names showed call-writing activity at strikes that, six months earlier, would have been spot. There was no margin-call cascade and no circuit-filter scramble. What there was, instead, was the slow draining of the speculative interest that had driven the parabolic phase, and the absence of any catalyst on the immediate horizon to reverse the trend.

Position-wise, by mid-September 2025, the desks that had run the trade well had cut the smaller, more operationally levered names first, retained partial positions in HAL and BEL where the underlying cash-generation visibility remained intact, and rotated freed capital into pockets that had lagged the eighteen-month window, including selective IT services and a few of the larger private banks. That rotation was already showing up in the relative-strength reading by the end of the window, which is usually the first clue that the next leadership has begun assembling itself even before the headline indices acknowledge it.

There is one final framing worth offering for the lookback. The PSU-defence cycle of 2024-25 was not a story of fundamentally bad businesses being overpriced. The order books were real, the execution improvements were real, and the strategic case for indigenisation was, and remained, real. What got mispriced was the rate at which a real story would convert into earnings that justified the multiples the market had assigned. That is a more sober lesson than the usual "thematic blew up" narrative, and it is the lesson that makes the cycle worth studying. Real stories with real order books are exactly the stories that the market over-prices most aggressively, because the resistance to selling them is the highest. The same template will appear under a different sector label in the next cycle. The signature, the rotation pattern and the retail-flow amplifier will be familiar.

VERDICT

Stance: NEUTRAL. Horizon: 3mo. Rationale: The eighteen-month round trip ended with the pack still digesting valuation reset and execution scrutiny, with no immediate catalyst for a sustained re-rating but the larger names back at multiples where the cash-flow visibility began to reassert itself, leaving the trade as a stock-picker's terrain rather than a thematic call.