The Uncomfortable Arithmetic of Patience

A new paper was perfectly timed in its distribution this week. Hendrik Bessembinder, updated through December 2025: one hundred years of US stock market data, 29,754 companies, USD91 trillion in wealth creation above Treasury bills.

The number that stopped me wasn't the USD91 trillion. It was the median buy-and-hold return across all those stocks: minus 6.87%.

More than half of all stocks that ever listed destroyed value. Nearly 60% underperformed a cash benchmark. Only 28% beat the market index. The greatest wealth-creating machine in human history delivered a negative median return to the individual companies within it. The $91 trillion was real. It was generated by only 3.72% of firms.

The standard, and market-led, response to this data is passive ownership. Five hundred names, broad exposure, no stock-picking required. But the problem is that the S&P 500 is not a broad basket. It is eight stocks doing most of the structural work, held alongside hundreds of value destroyers as ballast. When those eight represent 30% of global market cap, buying the index does not spread your risk - it spreads your exposure across a distribution most of which is negative, while your returns flow almost entirely from a handful of extreme winners. You are not avoiding concentration. You are simply inheriting yesterday's concentration, passively and retrospectively, without having made a single active decision about it.

That matters because staying passive through a market that is increasingly concentrated is still a decision. It is just one made by default rather than conviction

The more interesting question is what it would actually take to distinguish the 3.72% from the 59%. Bessembinder's data gives a partial answer, and it is counterintuitive. The stocks with the highest cumulative returns over the century - Altria at 4.4 million percent, Vulcan Materials at half a million - did not get there through extraordinary annual returns. Altria's annualised return was 16.5%. The median annualised return among the top 30 long-run performers was 13%. What separated them from the rest was not genius. It was duration. Altria's annualised return was 1.18 times Vulcan's. Its cumulative return was 8.81 times larger. Small differences, held long enough, become enormous distances.

This is the part that behavioural finance makes strange. The mathematics of compounding is simple. Its emotional execution is almost impossibly hard.

Investors talk about conviction investing as though it is primarily an intellectual exercise - finding the right companies, building the right thesis. It is that. But Bessembinder's data suggests the intellectual part may be the easier part. The harder part is the staying. The willingness to hold through the periods when the thesis looks wrong, when the stock is flat for three years, when the narrative has turned and the weight of received opinion is against you. The 100-year winners did not deliver straight lines. They delivered decades of compounding interrupted by crises, drawdowns, long periods of apparent mediocrity. The investors who captured those returns had to sit with uncertainty for years at a time. Most did not.

This is where the two questions connect. Breadth feels like safety. Five hundred names feels more prudent than fifty. But within equities, breadth is not risk reduction — it is median dilution. The genuine risk reduction is across asset classes. Within equities, the question is whether your concentration is deliberate and forward-looking, or accidental and backward-looking. And then, separately, whether you can hold it when the reason to sell feels more vivid than the reason you bought.

The concentration data in Bessembinder's update sharpens this further. In his original 2018 study, 89 firms accounted for half of all net wealth creation through 2016. Updated through 2025, the number is 46. Thirty firms accounted for 61% of the USD48 trillion created in just the last nine years. Nvidia alone: 9.3% of that. Compare that to Exxon Mobil, the top cumulative wealth creator through 2016, which accounted for 2.9% of everything created over the preceding 90 years.

Whether that concentration continues is a key investment question of the next decade. Bessembinder asks it directly in his conclusion: will AI accelerate winner-takes-all dynamics, or open space for more specialised firms to thrive? Most forecasts assume continuation. But every previous concentration in the 100-year dataset eventually redistributed. The question is never whether. It is whether you can absorb the psychological cost of being early when it does - of holding a view that has not yet been rewarded and may not be for some time.

What I take from this paper is not a formula. It is a restatement of something I have believed for a long time, now backed by a century of evidence: patience is the most underpriced asset in markets.

Not patience as passivity. Not owning everything and waiting. Patience as the active, disciplined, behaviourally-informed willingness to hold a concentrated position, often as part of a diversified portfolio, through the periods when it is uncomfortable and in doing so to stay invested long enough for the compounding to do its work. Most investors exit too early. Not because they are wrong about the company. Because the emotional cost of staying eventually exceeds the intellectual argument for it.

The market created USD91 trillion for the patient and the concentrated. It delivered minus 6.87% at the median.

That arithmetic is available to almost everyone. Almost no one captures it.

Rebecca Lewis is Partner & Co-CEO at Arisaig Partners. The views expressed are personal.