Bubbles, Crashes And Signals: Can Statistical Indicators Predict Financial Instability In India?

On May 17, 2004, trading on the Bombay Stock Exchange was halted as the index plummeted 15.52% in a single day, the darkest session in BSE history. Analysts called it a surprise. The dot-com crash, the 2008 collapse and the 2020 COVID freefall were each labelled the same way.

This piece, by Ahaana Bisarya of NIST International School, examines whether that label holds up, or whether India's markets leave statistical fingerprints before they collapse. But India's markets have lived through the Harshad Mehta scam, speculative peaks and sudden freefalls for four decades, and the word “surprise” reappears every time. That description captures how a crash feels, not whether it was foreseeable. If markets leave statistical fingerprints before they collapse, the real question isn't whether crashes can be predicted, it's whether anyone is watching the dashboard.  

What Is a Bubble and Why Is India Different?

A bubble occurs when a security's price persistently exceeds what its earnings or fundamentals can justify. A stock at 30x earnings is fairly valued if it's growing 40% a year; the same multiple signals a bubble if growth is stuck at 5%. In India, massive foreign capital inflows drive real growth, but the same inflows fuel speculative pricing, and a genuine growth story can act as cover for irrational exuberance. Since a security's “true” fundamental value is itself uncertain, spotting that gap is inherently difficult.

The Statistical Toolkit

Three tools do the heavy lifting. P/E deviation from the historical mean asks not whether a valuation is high, but how far it's drifted, and for how long. A right-tailed unit-root test on Indian indices during the 1999–2000 dot-com period found clear exuberance in the Nifty 50, Nifty Next 50, Nifty 500, and BSE 100, but none in the Sensex or Nifty Bank, showing bubbles can hide in some indices while sparing others.

The Variance Bounds Test checks whether price volatility exceeds dividend volatility. If it does, prices are moving irrationally. Applied to the CNX Nifty around the 2008 crisis, it found indirect evidence of a bubble driven by herd behaviour.

The Phillips-Shi-Yu (PSY) Method detects self-reinforcing price loops in real time. Applied to the Nifty 500 from 2003 to 2021, it flagged bubble episodes around 2007 and again around 2017.

What the Signals Said: 2007 to the 2024–26 Midcap Mania

Before 2008, PSY flagged bubble conditions in the Nifty 500 around 2007, but investors chasing momentum discounted the warnings. The clearest recent test came from India's mid- and small-caps. Through 2024, the BSE SmallCap and MidCap indices surged 30.7% and 28.3%, dwarfing the Sensex's 10–13% gain. 

Valuation concerns surfaced by September 2024, and by March 2025 a correction was underway: the Nifty Midcap 150 fell 20.5%, the Smallcap 250 fell 25.4%. That same month, SEBI and AMFI publicly called out “froth” and “irrational exuberance,” mandated fresh stress tests, and restricted lump-sum inflows into some funds, regulators acting within months, not years. By July 2026, the Nifty 50's P/E sits around 20.9, below its long-run median, while mid- and small-caps trade at roughly 33.5 and 32.8, stretched, but well off their 2024 extremes.

A separate January–February 2025 selloff, including a 1,000-point single-day Sensex drop, came from US tariffs and foreign outflows (roughly ₹61,000 crore between January and March), not a valuation bubble. The RBI defended the rupee; SEBI curbed short-selling. It's a reminder that not every crash is the kind these tools are built to catch.

Why Early Warning Systems Fail Anyway

Once a signal is widely known, traders exploit it, changing the dynamic it was meant to capture. India's Financial Sector Stress Index typically flags trouble only a few months ahead, a tight window for policymakers. Retail algo-trading and F&O activity have made momentum loops move faster than older models expect, and sentiment gauges can sit stuck at “extreme greed” exactly when a warning is needed most.

Institutional caution compounds this. The IMF's 2025 assessment of India found valuations broadly in line with earnings, but noted the RBI had already raised risk weights on unsecured personal loans back in November 2023, a reactive move. By 2025, the RBI had eased some of those same weights again to spur credit, regulators oscillating between tightening and easing rather than acting ahead of the curve, wary of the blame that comes with popping a bubble too soon.

Who Collects the Bill?

The tools work: PSY caught 2007 and 2017; variance-bounds tests found irrational pricing around 2008; SEBI and AMFI acted on midcap froth within months in 2025. The limitation isn't mathematical, its institutional will and speed. The deeper question is who's responsible for acting on these signals: regulators, investors, or markets themselves. Periodic instability may not be a systemic failure at all, but an inherent feature of a fast-growing emerging economy.