'2 out of 3 Diwali stock tips failed': Devina Mehra says forecasts are pure guesswork

Still chasing Diwali stock tips and bold Nifty forecasts for the new Samvat? Devina Mehra, Founder and Chairperson of First Global, has a stark message: stop. In a sharply worded post on X, Mehra dismantles the annual ritual of “expert” market predictions, calling them not just flawed—but fundamentally unknowable. The data, she says, proves it.

Citing a recent analysis by Mint, Mehra highlighted how poorly last year’s Diwali stock picks performed. Only one-third of the recommended stocks closed in the green. 

Even fewer beat the index. In Samvat 2023–24, the hit rate was marginally better—49% of picks were profitable. In both cases, Mehra notes, a random stock selection would have yielded better odds than blindly following expert advice.

And it’s not just stock picks that fall flat. Index projections fare no better, she argues. Referencing the book Noise by Daniel Kahneman, Cass R. Sunstein, and Olivier Sibony, Mehra places such forecasts in the realm of “objective ignorance”—things that are not just unknown, but unknowable.

“We maintain unchastened willingness to make bold predictions about the future from little useful information,” Kahneman writes. Mehra said she instinctively understood this truth decades ago. “That’s why I’ve never given index projections in the 30 years I’ve been in the markets.”

Data from Bloomberg backs her up. Analyzing historical S&P 500 estimates by major Wall Street firms—including Goldman Sachs, Bank of America, Citigroup, and Deutsche Bank—the findings were blunt: these forecasts are, on average, off by 15 percentage points. If they projected a 5% gain, the actual move could be +20% or –10%.

Even more telling, Mehra says, is how often forecasters default to predicting a 0–10% annual return. Yet over nearly a century of S&P 500 history, the market has actually moved within that band only 14% of the time.

Why, then, do these predictions persist? Mehra offers a psychological explanation: safety in consensus. “If everyone else is forecasting 7–10% and you say 20%, you risk looking foolish if you're wrong. It’s safer to be wrong with the herd.”

Mehra closes with a subtle pivot: it may actually be easier to estimate returns over the long term—10 to 20 years—than over a single year. But for those clinging to precise annual forecasts delivered with confidence? She leaves a final reminder: confidence is not competence.