261
MONTHS TESTED (FEB 2005-OCT 2026)
27%
FLOW EXPLAINED BY PRICE (MONTHLY, 2015-26)
1%
FLOW EXPLAINED BY REPORTED EARNINGS
-18%
COST OF FOREIGN TIMING, 2004-12 (2013-19: NO DIFFERENT FROM LUCK)
The Evening Number Is an Echo
If foreign money lifted the market, a month of heavy buying should be followed by a rising market. It is not. This month’s foreign flow moves in step with last month’s Nifty return (+0.26) and has no link to next month’s (-0.05).
Daily data say the same thing: a day’s net foreign flow lines up with the previous day’s return (0.21) and not the next day’s (0.03). Push the test out to six months and nothing changes.
CHART
Foreign flow follows last month's market. It tells you nothing about the next.
Correlation of a month's foreign flow with the Nifty's return k months earlier (negative k) or later (positive k). 261 months, approximate 95% significance bound ±0.12 - only k=-1 and k=0 clear it by any real margin.
The sharpest picture comes from the extreme months. We averaged foreign flow around the 11 months in which the Nifty fell 8% or more and the 18 in which it rose 8% or more.
Before a crash, flows look normal. In the crash month they drop. After a rally, they keep rising for two or three more months. Nothing in the months before a big move warned of it.
CHART
Foreign flows were normal before the crashes, then kept chasing the rallies for months
Average foreign equity flow in standard deviations (0 = a normal month). Month 0 is the month of the move. 11 crash months (Nifty -8% or worse) and 18 rally months (Nifty +8% or better), 2007-2026.
Chasing the Price, Not the Profit
The second article of faith is that foreign money comes for earnings upgrades. We tested it against the Nifty’s earnings per share (quarterly since 2013) and its P/E (since late 2015).
Price explains 27% of the monthly swings in foreign flow and 38% of the quarterly ones. Reported earnings growth explains about 1%. Once the price move is known, earnings add nothing, and last quarter’s earnings growth does not predict this quarter’s flow.
CHART
Price explains over a quarter of foreign flow, monthly
Share of foreign equity flow explained (R²) by that factor alone. Monthly: 131 months, Dec 2015 - Oct 2026.
CHART
Same pattern, quarterly
Quarterly: 52 quarters, 2013 Q4 - 2026 Q3. Reported earnings explain almost none of it at either frequency.
Prices move on news and expectations before earnings are reported, so a re-rating can be an upgrade the market saw early. Early or inside information would show up as flows leading price, and we find no sign of that, but net flows hide who traded, and we hold reported earnings, not analyst revisions. The careful claim is that flows follow what the market has already priced in, whether that reflects reported earnings or expected ones. The timing evidence above does not depend on separating news from data.
Valuation points the same way. The Nifty’s P/E peaked at 40 in February 2021. In October 2026 it is 19.1, near its lowest in ten years, and foreigners are still selling. Across the period they have, if anything, bought more when the market was already expensive.
CHART
Monthly net foreign equity flow
₹ '000 crore, Dec 2015 - Oct 2026 (alternating months shown).
CHART
Nifty trailing P/E, same period
Nifty trailing P/E at month end. Foreigners were selling in October 2026 with the P/E near a ten-year low of 19.1 - the opposite of a valuation-driven trade.
The Flow That Ignored the Market
The cleanest test needs money that arrives for reasons unrelated to the market’s own news. JPMorgan announced on 21 September 2023 that Indian government bonds would join its emerging-market index from 28 June 2024, at 1% a month through 31 March 2025. Index funds had to buy whatever the price.
Foreign debt flows jumped about fivefold: from ₹2,612 crore a month before the announcement to ₹12,923 crore after it and ₹13,671 crore through the phase-in, then back to ₹3,879 crore. The price effect is hard to find. The 10-year gilt index moved −0.08% on announcement day, about half a normal day’s move, and the RBI cut rates twice in 2025.
CHART
A rule forced foreigners to buy Indian bonds
Net foreign debt flow, ₹ '000 crore. Announced 21 Sep 2023, inclusion began 28 Jun 2024, phase-in ended 31 Mar 2025.
CHART
The price effect is hard to find
Indian 10-year government bond yield, monthly average (%), same period. No sustained break around the September 2023 announcement or the June 2024 inclusion date.
Does Chasing Cost Them?
Foreign investors buy after rallies and sell after falls. Across 262 months, they put in about ₹6,331 crore a month after years in which the Nifty gained more than 15%, and pulled out about ₹7,999 crore a month after years in which it fell.
CHART
Foreigners buy after the rally and sell after the fall
Average monthly net foreign equity flow (₹ '000 crore), by what the Nifty total return did over the prior 12 months. Jan 2005 - Oct 2026.
Chasing only costs money if the moves chased are about to reverse. We split every month since 2006 into five groups by the size of foreign flow and looked at the Nifty’s return over the next one and two years. After the heaviest-selling fifth of months it returned 12.8% over the next year. After the heaviest-buying fifth it returned 13.7%. The average for all months is 13.9%.
CHART
After the heaviest selling, the next year looked like any other
Nifty total return over the following 12 and 24 months, by fifths of months ranked by foreign flow (scaled to the previous five years), 2006-2026. Average for all months: 13.9% (12m), 26.6% (24m).
The cost was concentrated in one stretch. In 2004–12, foreign timing cost about 18% of the money invested, roughly ₹1.6 lakh crore on ₹4.8 lakh crore of net inflows, and random timing of the same flows would have done worse only 2% of the time. From 2013 to 2019 the timing result was +8%, inside the range of luck.
The individual moments tell the same mixed story. A rupee invested at the Covid low of March 2020, when foreigners were net sellers, has returned about 17% a year since, against 13% for the Nifty over 22 years. A rupee invested at the September 2024 peak, when they were net buyers, has lost about 5% a year.
CHART
The costly moments: selling the Covid low and buying the 2024 peak
Annualised Nifty return from the end of that month to today. 22-year average: 13.2% a year. Selling March 2020 (the Covid low) looks poor in hindsight; buying September 2024 (near the peak) looks poor too. March 2026 is too recent to judge.
The ₹3 Lakh Crore Question
Foreign investors have sold ₹3.09 lakh crore of Indian equities this year, including ₹118,000 crore in March, the month the Nifty hit its 2026 low. Whether that is a mistake like March 2020 or a good call like October 2024 depends on what the market does next. The data cannot say.
It can say what has not happened: foreign money has not led the Indian market, not in 21 years of monthly data.
“Foreign money is a barometer of the Indian market, not an engine of it.”
Primary data sources: NSDL, NSE, AMFI, FRED, ECB.
Methodology: Monthly net foreign equity flow is compared with the Nifty 50’s return at different lags, and with earnings per share (trailing twelve months, quarterly from 2013) and P/E (from late 2015). Timing is compared with random reshuffles of the same flows within each period; forward returns use fifths of months by flow scaled to the previous five years. Results describe timing and association in aggregate flows, not proof of cause or the record of any individual fund.
This analysis is for informational and research purposes only and does not constitute investment advice or a recommendation to buy, sell, or hold any security. Statistical relationships in historical data do not establish causation and may not persist. Flow data is as reported and may be revised.