IRB Infrastructure Developers Ltd.

This is not an accusation and not a finding that any manipulation occurred. Beneish's own paper is explicit that the score identifies companies whose accounting CHARACTERISTICS resemble those of a sample of past manipulators -- nothing more. StockProof never labels a company "manipulator", "red flag" or "suspicious", and never shows any list of companies ordered by this score. The eight components below -- factual statements about figures the company itself filed -- always come before the composite number, which is the inference, not the fact.

The M-Score is calibrated on 74 US-listed companies from 1982-1992 later found to have manipulated earnings (Beneish, M. D. (1999), "The Detection of Earnings Manipulation", Financial Analysts Journal 55(5)) -- not on any Indian company or era. Measured on StockProof's own Nifty 500 universe as of 2026-09-02, 13.6% of companies score above Beneish's own published cutoff. On any realistic base rate of actual earnings manipulation, that flag rate is overwhelmingly false positives, not overwhelmingly real ones.

No "Selling, General & Administrative Expense" element exists anywhere in NSE's filed data. The SGAI component below is approximated as Employee Benefit Expense + Other Expenses, divided by revenue -- a reasonable stand-in for a manufacturer, but for a services company this proxy is closer to the entire cost base than to SG&A specifically. This is disclosed because it is a genuine deviation from Beneish's published specification; its coefficient (-0.172) is the smallest of the eight, which bounds how far the approximation can move the composite.

This measures something different from a financial-distress score like Altman's Z-Score, and the two essentially never agree. Measured on StockProof's own 389-company overlap as of 2026-09-02: only 5 companies were flagged by both, an agreement rate (intersection ÷ union) of 6.2%, and a correlation between the two scores of +0.088 -- closely matching the independently measured +0.092 over 3,242 Malaysian-listed companies. A low Z-Score and a high M-Score are not corroborating each other; they are unrelated measurements.