Adani Green's -77.18% abnormal return in the ten trading days after Hindenburg published on 24 January 2023 is the number a trader should care about. The constant-mean benchmark gives it a t = -5.362. On publication day, the abnormal return was just -0.42%.

The gap between those returns is the trading story. The paper reads it backwards.

What the study measures

Singh, Azad, Tiwari and Rajak conduct a single-firm event study across nine listed Adani Group stocks: ACC, Adani Enterprises, Adani Green, Adani Ports, Adani Power, Adani Total Gas, Adani Wilmar, Ambuja Cement and NDTV. They treat an activist short seller as an information intermediary. Allegations of stock manipulation, accounting irregularities and weak governance should cause prices to reset if they carry value-relevant information that the market has yet to price. The Supreme Court judgment of 3 January 2024 supplies the second shock. In principle, it could reverse some of the damage from the report. The paper calls it "the favourable judicial outcome,".

The methods are familiar. The authors calculate log returns from investing.com daily closes and use "relevant proxy indices" without identifying an index. They estimate expected returns with three models: constant mean (CRM), a market model (MAM) and CAPM. Each uses a 40-day estimation window. The 21-day event window comprises the "anticipation" window, covering the ten preceding days, the event day itself, and the "adjustment" window, covering the ten following days. For each firm and window, the authors calculate cumulative abnormal returns (CARs) and buy-and-hold abnormal returns (BHARs). Their t-test scales daily stdev from the estimation window by the square root of the window length.

The authors report negative reactions across all nine stocks and significant reactions in five: Green, Enterprises, Wilmar, Power and Total Gas. They put Total Gas's decline at about 83% by 28 February 2023. For the judgment, they report a significant positive reaction only in Adani Power. Judicial developments, they conclude, "may only partially mitigate the adverse market reaction."

When was the trade available?

On the report date, returns were small or went the wrong way. Enterprises returned -1.35% (CRM), Ports -0.77% and ACC +0.53%. Wilmar gained 5.19% (p = 0.127). The losses followed publication. Green's adjustment-window CAR reached -77.18% under CRM and -55.46% under CAPM. Wilmar recorded -27.07% (p = 0.014), with a significant adjustment CAR under all three benchmarks.

The 21-day totals include ten days before publication, when nobody could have traded on the report. Green's anticipation CAR was +6.46%. Including it leaves a headline total of -71.14%, understating the move after publication.

The paper reads these results as semi-strong efficiency. It says an event study requires prices that "respond to the news promptly and precisely." Yet a ten-day decline is post-event drift, the only pattern in these tables a trader could act on. The limitations section worries that short windows "may not fully capture delayed investor responses." Here, most of the result comes from a delay already inside the window.

This remains an in-sample study of one event affecting nine names exposed to the same report. We found no discussion of transaction costs or short-sale constraints anywhere in the paper.

The five 'significant' firms do not line up with the p-values

Adani Enterprises makes the significant list. Its total CARs are -44.92%, -47.67% and -45.04%; the corresponding p-values are p = 0.138, 0.119 and 0.137. For its adjustment-window CARs, the p-values are 0.059, 0.058 and 0.057. Only the adjustment-window BHAR clears 5%, with 0.047, 0.047 and 0.046. The authors themselves judge CAR more suitable for such short windows.

ACC is absent from the list despite adjustment CAR p-values of p = 0.049 (MAM) and 0.044 (CAPM). Its CAPM total CAR has p = 0.046, and its adjustment BHAR has p = 0.045 and 0.041. That gives ACC five cells clearing 5%, against three for Enterprises. Enterprises does pass under all three benchmarks on a single row, its adjustment BHAR. The paper never specifies that as its selection rule. Such a rule would also need reconciling with its preference for CAR over BHAR.

Enterprises exposes the test's power problem as well. Daily AR stdev in its estimation window is 0.065, scaling to 0.297 over 21 days. A -45% total CAR in Adani Enterprises consequently fails the t-test. Every event generates 9 firms × 3 models × 4 windows × 2 measures, or 216 unadjusted tests. Testing proceeds firm by firm, without a group-level test. The paper cites Kolari and Pynnönen (2010) on cross-sectional correlation, yet does not apply that correction to the nine stocks struck by the same report. We flagged the same separation between text and tables in a Turkish rating-announcement study.

Searching the tables for Adani Power's court rebound

Table 3 presents the Supreme Court event for ACC, Green, Enterprises, Wilmar and Ports. No total CAR is significant; the lowest p is 0.343. Wilmar under CAPM supplies the strongest event-day cell: +4.67%, with t = 1.966 and p = 0.057.

A continuation table contains the remaining four firms, but has no event label and a garbled column header. The only significant column is negative. It reports adjustment CARs of -34.69%, -38.41% and -32.84% (p = 0.001, 0.000, 0.001). The highest event-day t is 1.552 (p = 0.129). Another column exactly repeats Adani Green's Table 3 CARs: event-day returns of 4.42%, 5.58% and 5.34%, followed by adjustment returns of -19.10%, -7.99% and -12.02%. The stdevs, however, are 0.042, 0.036 and 0.044, versus 0.095, 0.093 and 0.097. We could not locate a significant positive Adani Power figure anywhere in the printed tables.

The prose changes windows as it moves between firms. The "modest positive" returns attributed to Enterprises, ACC and Ports correspond to their event-day CARs. Most of their totals are negative. Enterprises records -13.05%, -0.73% and -2.17%; Ports records -11.33%, +6.08% and -4.32%. Green and Wilmar receive negative descriptions even though their event-day gains, at 4.42-5.58% and 3.64-4.67%, are the largest in Table 3.

The conclusion that the judgment was "insufficient to fully offset" the earlier damage has a deeper measurement problem. Each event uses its own 40-day benchmark window. A 21-day CAR around January 2024 cannot establish recovery from a 2023 loss. No figure in the paper measures that recovery.

The 83% fall in Adani Total Gas is a price move

Table 1 reports an Adani Total Gas maximum close of 3918.90 and a minimum of 714.25 across 74 observations, a gap of about 82%. The Total Gas figure therefore describes a raw price decline. Its cited endpoint, 28 February 2023, also falls beyond the ten-day adjustment window. The continuation table's largest CAR is -42.14%. Green has the largest abnormal loss printed anywhere for Event 1, so placing Total Gas first ranks price moves. Table 3 offers a trader no reason to buy on the judgment: the lowest total-CAR p is 0.343.

We could not test these results ourselves. Doing so requires daily histories for Indian-listed Adani shares and an Indian benchmark index; we hold neither. Replacing them with US names would remove the company-specific allegations at the heart of the paper.

Adani Green's and Adani Wilmar's post-report drift survives every benchmark the authors use. The judicial leg warrants another look only when a labelled Event 2 table for Adani Power, with its own stdev, is available.