Research

The work
underneath.

Longer than a note and slower to write: studies that state what was measured, over what period, and what was found — published in full, with the data behind every figure.

See both papers The latest study
  • Survivorship-free
  • Published in full
  • Reproducible
  • Limitations stated
2 papers

What is published here.

One study from this desk and the century-scale work it extends. Choose one to read it in full.

  1. Paper 01 Marathon Trends research

    India · NSE · 3,214 companies · 1994–2026

    How Concentrated Are Equity Returns in India?

    Atul Suri & Devendra Agarwal Marathon Trends Advisory Private Limited

    The median NSE stock returned 1.29× over its life. The mean returned 15.4×. Thirty-nine companies out of 3,214 produced half of every rupee of net gain the market ever made — and once the 425 delisted firms most studies quietly drop are put back, the median return halves.

    • 3 Sep 2026
    • 17 pages
    • Market structure
    Read it on this page Open the PDF
  2. Paper 02 Reference literature

    United States · CRSP · 29,754 stocks · 1926–2025

    One Hundred Years in the U.S. Stock Markets

    Hendrik Bessembinder W. P. Carey School of Business, Arizona State University

    Twenty-nine thousand seven hundred and fifty-four U.S. stocks over a hundred years. The mean buy-and-hold return is above 30,000%; the median is −6.9%. Shareholders gained $91 trillion — and 46 firms produced half of it.

    • 21 Mar 2026
    • 15 pages
    • Reference literature
    Read it on this page Open the PDF
Paper 01 Marathon Trends research Market structure SkewnessSurvivorship biasConcentrationNSEMultibaggers

How Concentrated Are Equity Returns in India?

Survivorship-Free Evidence from Indian Equities, 1994–2026

India · NSE · 3,214 companies · 1994–2026

  • Atul Suri & Devendra Agarwal
  • Marathon Trends Advisory Private Limited
  • 3 Sep 2026
  • 17 pages
Read the paper

JEL G11, G12, G14, G15, N25

  • 3,214 Companies studied
  • 39 Produced half the gains
  • 1.29× Median lifetime return
  • 32 Years of data

The median NSE stock returned 1.29× over its life. The mean returned 15.4×. Thirty-nine companies out of 3,214 produced half of every rupee of net gain the market ever made — and once the 425 delisted firms most studies quietly drop are put back, the median return halves.

We study the cross-section of long-horizon buy-and-hold returns for 3,214 companies listed on India's National Stock Exchange between November 1994 and August 2026, using a survivorship-free universe that reincorporates 425 delisted companies. We document that the distribution of lifetime returns is extraordinarily positively skewed: the median stock returns 1.29× while the mean return multiple is 15.4×, and 87% of stocks underperform the cross-sectional mean. Positive skewness in the wealth distribution generates severe concentration of aggregate gains — 39 stocks (1.2% of the universe) account for one half of all net rupee gains, and 88 stocks that returned 100× or more account for approximately two-thirds. Reincorporating delisted firms lowers the fraction of stocks with positive returns from 62.7% to 55.9% and raises the fraction suffering near-total loss from 2.6% to 10.4%, confirming that survivorship bias materially flatters conventional return statistics. Our results extend the U.S. evidence of Bessembinder (2018, 2026) to a major emerging market and are robust to winsorization, to bootstrap resampling, and across listing cohorts.

Keywords positive skewness; buy-and-hold returns; survivorship bias; return concentration; Indian equities; emerging markets

What it found

Four results, in plain terms.

  1. 01
    The typical stock is not the market

    The median NSE company returned 29% in total over a median 8.4 years — roughly 3% a year, below a fixed deposit. The mean multiple of 15.4× is twelve times the median, and 87% of stocks fell short of it.

  2. 02
    Thirty-nine names carried half the market

    Ranked by net rupee contribution, 9 stocks produced a quarter of all net gains, 39 produced half, and the cumulative total peaks at the 907th name — the best 28% of companies exactly offset the losses of the other 72%.

  3. 03
    A hundred-bagger is slow, not sudden

    88 companies returned 100× or more and account for about two-thirds of net gains. The median one took 30 years and compounded at 22% a year. That is patience, not a sprint.

  4. 04
    Survivorship bias is first-order

    Put the 425 delisted firms back and the positive-return share falls 6.7 points, near-total losses rise from 2.6% to 10.4%, and the median return is halved from 60% to 29%. Their own median outcome was −95%.

Cumulative share of net gains, by rank

3,214 NSE companies, ranked by net rupee contribution
0% 25% 50% 75% 100% 1 10 100 1000 3,214 Company rank by contribution to net gains (log scale) 9 stocks 0.28% of the market → 25% of gains 39 stocks 1.21% of the market → 50% of gains 139 stocks 4.32% of the market → 75% of gains 907 stocks 28.22% of the market → 100% of gains

Own the 1.21% of the market that contributed most, and you hold the names behind 50% of every rupee of net gain. Miss 39 stocks out of 3,214, and that 50% is what you forgo.

Half of every rupee of net gain came from 39 companies. Cumulative gains peak at the 907th name and decline thereafter — the remaining 72% of the market subtracted, in aggregate, exactly what the best 28% added. The four marked points are the paper’s published thresholds; the line between them is an interpolation, not a measured series.

The multibagger census

Survivorship-free universe, N = 3,214
2× 1,26239.3%
5× 80124.9%
10× 53016.5%
25× 2939.1%
50× 1725.4%
100× 882.7%
500× 150.5%
1,000× 30.1%
Counts of companies whose lifetime buy-and-hold multiple met or exceeded each threshold. The 88 hundred-baggers account for roughly two-thirds of aggregate net gains.

Outcomes improve with time held

Companies binned by length of listed life
Length of listed lifeCompanies Positive returnBecame 10-baggersMedian CAGR
Under 5 years 1,215 42.2% 1.1% −6.5%
5–10 years 572 52.4% 9.3% +1.4%
10–20 years 730 61.1% 19.0% +4.4%
20 years and over 697 77.3% 46.6% +8.2%
The compounding mechanism in one table: over longer horizons the right tail has more room to extend. It partly reflects long lives being conditioned on survival, which the paper states rather than obscures.

What survivorship bias hides

Survivorship-free (3,214) against survivor-only (2,789)
Survivorship-free · 3,214 companies Survivor-only · 2,789 companies
Positive lifetime return 55.9%62.7% +6.7 pp
Lost money 43.7%37.0% −6.7 pp
Beat a fixed deposit (6.5% p.a.) 43.9%49.2% +5.3 pp
Near-total loss (>90% decline) 10.4%2.6% −7.8 pp
Became 10-baggers 16.5%18.9% +2.4 pp
Median total return 29.2% survivor-only reads 60.0% +30.8 pp
"Bias" is the survivor-only figure minus the survivorship-free one — the distortion introduced by omitting delisted firms. The 425 delisted companies had a median return of −95% and an 87.8% loss rate.

The fifteen largest contributors to net gains

Ranked by net rupee contribution
#CompanyMultiple YearsCAGRShare of net gains
01 Adani Enterprises 4,855× 32 31% 10.5%
02 Eicher Motors 1,399× 32 26% 3.0%
03 Infosys 1,286× 32 25% 2.8%
04 Divi's Laboratories 942× 23 34% 2.0%
05 Sun Pharmaceutical 803× 32 24% 1.7%
06 Samvardhana Motherson 700× 30 24% 1.5%
07 Havells India 666× 25 29% 1.4%
08 Aurobindo Pharma 651× 31 23% 1.4%
09 Radico Khaitan 613× 23 32% 1.3%
10 Pidilite Industries 613× 32 22% 1.3%
11 Titan Company 547× 32 22% 1.2%
12 TVS Motor Company 518× 25 29% 1.1%
13 Bajaj Finance 517× 32 22% 1.1%
14 Godrej Industries 517× 24 30% 1.1%
15 Schaeffler India 502× 25 28% 1.1%
Company names appear to describe what happened over three decades. They are not recommendations and are not necessarily held in any Marathon Trends strategy.
Paper 02 · the literature

What this work
stands on.

Our own paper is an extension, and an extension has to say what it extends. The study below is not ours — it is cited, credited and linked to its author.

Paper 02 Reference literature SkewnessWealth creationCRSP

United States · CRSP · 29,754 stocks · 1926–2025

One Hundred Years in the U.S. Stock Markets

Hendrik Bessembinder · W. P. Carey School of Business, Arizona State University · 21 Mar 2026

Twenty-nine thousand seven hundred and fifty-four U.S. stocks over a hundred years. The mean buy-and-hold return is above 30,000%; the median is −6.9%. Shareholders gained $91 trillion — and 46 firms produced half of it.

29,754Stocks studied
$91TNet wealth created
46Firms made half of it
−6.9%Median lifetime return

This paper is the work of Professor Hendrik Bessembinder of Arizona State University. It is reproduced here as the reference our own study builds on, and is not Marathon Trends research. The author's own copy.

Read the paper
  1. 01
    $91 trillion, from 3.72% of firms

    The first 27,999 firms — 96.28% of the century's listings — collectively matched Treasury bills and produced zero net wealth. The remaining 1,082 firms produced all $90.96 trillion of it.

  2. 02
    The median stock lost money

    Only 48.22% of stocks delivered a positive lifetime buy-and-hold return, only 41.17% beat Treasury bills over their own listed months, and only 27.60% beat the market.

  3. 03
    Concentration is tightening

    Half of net wealth creation took 89 firms when measured to 2016. Measured to 2025 it takes 46. Nineteen of the thirty largest wealth creators since 2017 do not appear on the list through 2016.

  4. 04
    Time in the market, not spectacular rates

    The thirty highest cumulative returns came from a median annualised return of just 13.0% — compounded over an average of 93.9 years. None of the thirty highest annualised performers appears on that list.

The open question

Why it sits on this page

Bessembinder (2018) was the first study to document systematically how concentrated long-term market outcomes are when measured in money rather than percent. This is its century-scale update, and it is the direct antecedent of our own work.

Whether that result is a structural property of equity markets or a feature of a mature, deep, heavily researched U.S. market is an open empirical question — and it is the question our India paper was written to answer. Emerging markets differ in ways that could plausibly amplify or dampen the effect: shorter listing histories, higher failure and delisting rates, thinner liquidity, concentrated promoter ownership.

The answer turned out to be that the pattern recurs in India, if anything more starkly.

Please read. These papers are empirical research published for information and educational purposes. They are not investment advice, not research analyst recommendations, and not an offer or solicitation to buy or sell any security. Company names appear solely to describe historical patterns of wealth creation; they are neither recommendations nor necessarily held in any Marathon Trends strategy. Past performance is not indicative of future results. Papers last updated 2026-09-24. Full disclosures.
From the page to the process

The tail is where the return lives.
Holding it is the harder half.

The concentration result is why the process is built around riding strength rather than predicting it — and why the exit rule matters as much as the entry.

Why trend following Speak with the team