Financial Planning Insights

Clear, thoughtful articles to help you make better decisions with your money, your future
and the people who matter most.

Buying the Haystack: what a question about tech concentration taught me about investing

My account of how I arrived at our one best investment idea

For the past few years, a version of the same question has come up in more client conversations than I can count: aren’t we taking on a huge amount of risk by tracking a global index, when so much of it now sits in a handful of American tech companies?

It’s a reasonable thing to worry about, and I didn’t have as sharp an answer as I wanted. So, I went looking for one, properly, rather than reaching for a stock answer. What follows is the record of that search, the papers I read, the things that surprised me, the moments I questioned the conclusion I was heading towards, and the cross-referencing I did before I trusted it.

It ends up somewhere bigger than the tech question I started with. I’ve consciously written it at length, because the decision it defends, to build portfolios around low-cost index funds rather than picking stocks, sectors, or managers, is probably the single most important idea we hold at Yellowtail. I’d rather you saw the thinking behind it than just the conclusion.

“ “I’d rather you saw the thinking behind it than just the conclusion.”
Dennis Hall, Founder, Yellowtail Financial Planner

Contents

I. The Question I Couldn’t Answer Well Enough
II. Down the Rabbit Hole
III. The Arithmetic That Can’t Be Argued With
IV. Skill, Luck, and the Uncomfortable Truth About Fund Managers
V. Every Generation Has Its Darlings
VI. When the Unsinkable Sinks
VII. The Mind Against Itself
VIII. A Coincidence Too Good to Leave Out
IX. Taking the Other Side Seriously
X. A Few Numbers I Couldn’t Leave Out
XI. How This Came Together
XII. Why This Is Our One Best Investment Idea

I. The Question I Couldn’t Answer Well Enough
It usually comes up unannounced, and never on the meeting agenda. A client will casually mention something they’ve read, or a headline about how much of the S&P 500 now sits in seven or eight technology companies, and ask, offhandedly, whether their portfolio really is as diversified as we’ve always told them it is.

It’s a good question. It deserves more than the reassuring comments I sometimes found myself making, “it’s always been somewhat concentrated,” “the index adjusts for that automatically”, which is true, but still a weaker answer than I wanted to give.

So, I set myself a proper task: understand concentration risk well enough to explain it deeply, not just soothingly. I thought I’d end up with a slightly more technical version of the reassurance I already gave. I didn’t. I ended up somewhere else, which changed and ultimately strengthened how I think about the whole case for indexing, not just the concentration question. This paper is the record of that intellectual journey, including the parts where I wasn’t quite sure I knew where it was heading.

II. Down the Rabbit Hole
My starting point was simple: how unusual is today’s concentration, when compared with history? That question led me to Hendrik Bessembinder, a finance professor who has spent years tracking the lifetime performance of every stock that’s ever traded on the US market since 1926, not performance year by year, but what happened from the day each company first listed to the day it either got taken over, delisted, or is still trading today.

What he found surprised me, mainly because it wasn’t the thing I’d gone looking for. It wasn’t that a handful of companies are dominant right now, I already knew that. The surprise was this: across a full century and roughly 29,000 companies, most individual stocks have made their long-term holders no better off than if they’d simply left the money in Treasury bills, essentially cash.

But there’s a nuance to be acknowledged, because that statement on its own might lead you to think you can’t make money from most listed shares, and that’s simply not true. Holding a share over the entire life of the company is a different proposition to holding it over shorter time periods, where it may deliver ‘better than cash’ returns. It’s the ground that active stock pickers like to rake over to try and beat the market.

Bessembinder’s most recent update, published this year, covers the full hundred years to 2025. American stocks created $91 trillion in wealth above what an investor would have earned in cash over that period. Just 46 companies, well under 0.2% of every company that’s ever listed, accounted for half of that net wealth creation. A larger group, 1,082 companies, accounted for all of it, while the remaining 96% or so of companies, taken together, roughly matched Treasury bill returns. Many of them did considerably worse.

I found that number difficult to reconcile. It reframes the entire question of concentration risk, because it says concentration isn’t a distortion of how markets work, it’s a description of how they’ve always worked.

The market’s return has never really been “the market’s” return, spread evenly across everything in it. It has always been the return of a tiny number of exceptional businesses, carried by an enormous crowd of companies that, over their full lifetimes, added roughly nothing.

I need to be precise about what this does and doesn’t mean, because it’s the part that I’ve subsequently heard misquoted, and I don’t want to be guilty of that myself. This isn’t saying most stocks have a bad year, or a bad decade, because plenty of perfectly ordinary companies have periods of real outperformance, sometimes for years at a stretch, before drifting back or fading away.

What Bessembinder is measuring is lifetime, buy-and-hold performance: if you had bought a company at listing and held it, without trading, right through to the end of its public life, most of the time you would have been better off in Treasury bills. It’s a statement about the full arc of a company’s life, not about any single year along the way, and treating it as the latter is a common and, I think, serious misreading of the research.

Bissembinder’s research goes beyond the US stock market, and I found a degree of comfort in discovering his findings are largely replicated across other global markets. However, the veracity of the underlying data across several different markets, alongside the differing timeframes involved, led me to focus on the US market alone, after all, it was the tech stock concentration that initiated this research.

III. The Arithmetic That Can’t Be Argued With
Once I’d absorbed Bessembinder, another piece of research I already half-knew clicked into place with greater clarity than before: William Sharpe’s 1991 paper, “The Arithmetic of Active Management.” It had formed part of my investment exam syllabus, and I’d always found it a clean, satisfying argument. It needs no assumption about market efficiency, no view on whether prices are ever “wrong,” and no faith in anyone’s clairvoyance. It rests on one simple fact: every share in a market is owned by somebody.

We define a passive investor as someone who holds every security in a market in proportion to its size. Everyone else, anyone overweighting some stocks and underweighting or avoiding others, is, an active investor. Because the passive investor’s holdings simply mirror the market as a whole, and the market as a whole is nothing more than the sum of everyone’s holdings, active and passive, the pound-weighted average return of all active money must equal the pound-weighted average return of all passive money, before costs. It cannot be otherwise; it’s arithmetic, not opinion. Active investing costs more, research, trading, higher fees all must be paid for, so once costs are taken into account, the average actively managed pound must lag the average passively managed pound.

Worked example
Imagine a market of just three stocks, held by three investors who each put everything into one stock: £300 into A, £200 into B, £500 into C. That makes the market 30% A, 20% B and 50% C. A passive investor putting in £2,000 would simply hold the same proportions. If A rises 50%, B falls 60% and C rises 5% over the year, the three active investors collectively make £55 profit on their £1,000, a 5.5% return. The passive investor, holding the same market in the same proportions, also makes exactly 5.5%. The numbers must match, because the passive portfolio is the market. The only difference is that the active investors took on far more individual volatility to arrive at the same place, and, in the real world, would have paid more in fees to get there.

Bessembinder sharpened the argument for me. If half the market’s net wealth creation above Treasury bills rests on 46 companies out of 29,000, then trying to beat the market by picking stocks isn’t just fighting the arithmetic of costs, you’re also fighting genuinely terrible odds of holding the right handful of names in the first place. Sharpe tells you the average active pound must underperform the average passive pound after costs, under his assumptions. Bessembinder then tells you why the deck is so heavily stacked against any individual pound that is actively managed. Of course, the argument has its limits, and we should acknowledge them. Sharpe’s original model assumes a static market portfolio, and real markets aren’t static, new shares are issued, old ones bought back, indices reconstituted, so even passive investors need to trade occasionally. The economist Lasse Pedersen has shown this creates small openings for active managers to add value by providing liquidity around those events. It’s a fair point, and one that has been raised with me on occasion. It doesn’t change the conclusion for the ordinary investor, and as the next chapter shows, decades of real-world fund data confirm that Sharpe’s arithmetic holds up firmly in practice.

IV. Skill, Luck, and the Uncomfortable Truth About Fund Managers
The obvious response to Sharpe is this: fine, the average active manager must lose, so don’t be average, find the skilled ones. This is where Eugene Fama and Kenneth French’s 2010 study earns its place in the argument, because it was designed precisely to separate genuine skill from luck.

Professors Fama and French built a simulation of thousands of fund managers with, by construction, zero real skill, managers who add nothing above or below what chance alone would produce and compared the simulated results against three decades of real US mutual fund data.

If skilled managers exist in any numbers, the real-world distribution of returns should sit meaningfully to the right of the no-skill simulation. It largely doesn’t. Before costs, a handful of the very best real managers do match what a modestly skilled population would produce. The opportunity exists, but that’s different from being able to identify and capture it reliably. Some managers do outperform after costs; most of those top performers, though, still land close to what an efficient passive fund would have delivered anyway.

The high costs of active management show up intact as lower returns to investors – Fama & French, 2010

A natural follow-up question: even if genuine skill is rare, can you identify the skilled managers in advance, and stick with them? This is exactly what S&P Dow Jones Indices tests every year in its SPIVA Persistence Scorecard, and the year-end 2025 edition gets straight to the point on this issue. Across US large-cap funds, 79% underperformed the S&P 500 in 2025 alone, the fourth-worst year in the scorecard’s 25-year history. Of the funds that ranked in the top half in 2021, less than one in twenty stayed in the top half in every subsequent year through 2025. If fund rankings were determined by pure chance, you’d expect slightly better than that just by luck alone; it’s as though they’re deliberately setting out to fail! These findings are a warning against treating past outperformance as sufficient evidence that a manager, or an investor, will continue to outperform; they describe the fund groups and periods actually measured, not a universal law about this year’s winners and next year’s results.

It isn’t only professional fund managers this applies to, either. Terrance Odean and Brad Barber’s study of individual investor trading accounts, provocatively titled “Trading Is Hazardous to Your Wealth,” found that the more frequently individual investors traded, the worse their returns tended to be, the most active traders in their sample underperformed the market by several percentage points a year, largely eaten up by costs and poorly timed decisions. The pattern isn’t confined to professionals with research teams and trading desks. It shows up just as clearly in ordinary people who are convinced that they’ve spotted an edge.

That is the uncomfortable truth underneath the whole “just pick a good manager”, or “trade a bit more actively yourself” idea: today’s outperformance, at every level, is a remarkably poor guide to tomorrow’s outperformance. The manager, or the investor, who beat the market this year is not meaningfully more likely to do it again than one chosen at random.

V. Every Generation Has Its Darlings
The other thing that struck me, once I started looking at market history properly rather than just at today’s headlines, is that extreme concentration in a handful of dominant companies isn’t new, and it isn’t specific to technology. Every generation of investors has had its own version of “a few unstoppable companies at the top of the market,” and every one of those eras eventually rotated on to something else, usually after the previous generation had convinced itself this time was different.

A story that resonated deeply with me was the “Nifty Fifty” of the late 1960s and early 1970s: a group of around fifty blue-chip growth stocks, IBM, Xerox, Polaroid, Coca-Cola, Disney among them, that institutional investors regarded as “one-decision” stocks, safe to buy at almost any price because their quality was beyond question.

By 1972 the top five stocks alone made up roughly 23% of the S&P 500, and the Nifty Fifty as a group traded at an average price-to-earnings ratio of around 42, more than double the broader market. Then the bear market of 1973–74 arrived. The Nifty Fifty fell harder than the market as a whole; Polaroid alone lost more than 90% of its value and never fully recovered, eventually filing for bankruptcy in 2001. Polaroid aside, the companies, by and large, kept being good companies. Coca-Cola and IBM went on to have perfectly respectable decades. What collapsed waswasn’t the businesses, it was the price everyone had agreed to pay for the certainty of owning them.

The pattern repeats with almost eerie regularity if you keep looking. In 1980, on the back of the oil price shocks, the energy sector swelled to nearly 28% of the S&P 500, more than a quarter of the entire US stock market riding on oil and gas companies. It sits at around 3% today. At the peak of the dot-com bubble in March 2000, technology alone reached roughly a third of the S&P 500; Cisco, Lucent, and WorldCom were the “can’t lose” names of that era, and the Nasdaq lost close to 80% of its value over the following two years. In the years before 2008, banks and financial companies reached roughly 20–22% of the S&P 500, propped up by a housing boom that turned out to be built on shakier foundations than almost anyone in the industry wanted to admit at the time.

A quick tour through the market’s previous darlings
The Nifty Fifty, early 1970s. Share of the S&P 500 held by its five largest stocks: roughly 23% in 1972. Average price-to-earnings ratio of the group: about 42. The 1973–74 bear market hit them hardest of all.

Energy sector, 1980. Share of the S&P 500: nearly 28%, driven by the oil price shocks. Around 3% as of December 2025.

Technology and telecom sector, March 2000. Share of the S&P 500: roughly a third, at the dot-com peak. The Nasdaq lost close to 80% of its value over the next two years.

Financials sector, mid-2000s. Share of the S&P 500: roughly 20–22% in the years before the 2008 financial crisis.

Ten largest companies, December 2025. Share of the S&P 500: over 40%, a figure that exceeds even the dot-com peak measured the same way.

I don’t think this tells us what happens next to today’s technology giants specifically, there’s a genuine, serious argument that today’s leaders are different, because they are far more profitable than the dot-com names ever were, with real earnings rather than speculative promises underneath their valuations. But it does tell me that concentration itself is not some unprecedented anomaly that today’s markets have stumbled into. It’s the market doing what it has always done, in every decade I looked at, which is let a handful of the moment’s most exciting companies dominate, right up until, for reasons that were rarely obvious in advance, they don’t.

VI. When the Unsinkable Sinks
Statistics and sector-weight charts are one thing. What shifted my thinking was reading more deeply into what happened to specific companies that occupied the position today’s giants occupy now, and noticing that they failed for three genuinely different reasons, none of which was obvious to the people holding the shares at the peak.

Enron: America’s most innovative company

Through the late 1990s, Enron was a Wall Street darling, Fortune magazine named it America’s Most Innovative Company six years running. By August 2000 its shares hit an all-time high of $90.75. Sixteen months later it was worth 26 cents a share and had filed for what was then the largest bankruptcy in US history. This was fraud, not misfortune: the accounting behind Enron’s growth turned out to be fabricated, and by the time that became public, there was nothing underneath the share price at all.

Nortel: a third of an entire stock market

In the summer of 2000, Canada’s Nortel Networks was so dominant it made up roughly 35–38% of the entire value of Toronto’s stock market index; one company, more than a third of a national index. By 2002 the stock had lost over 95% of its value, and it filed for bankruptcy in 2009 with shareholders left with nothing.

What I find most useful about this one isn’t the collapse itself, it’s what happened to the investors on either side of it. Index investors still suffered through their holding in Nortel; its weighting fell as its share price fell, which reduced the exposure remaining rather than avoiding the losses already taken. The benefit was that they also owned several thousand other businesses, not that the index insulated them from the collapse. Investors who had piled into Nortel directly, because it was the obvious, unmissable stock of its era, had no such cushion, and lost almost everything. A diversified index doesn’t need to identify every corporate failure in advance, but it still participates in the losses when a constituent declines; its advantage is spreading exposure across businesses, not preventing an individual failure from affecting returns.

General Electric: the slow fall of the safest stock in America

Enron and Nortel both collapsed quickly, within a year or two of their peaks. GE’s story is different, and arguably more sobering, because the decline unfolded over nearly two decades, a long, grinding erosion in a company everyone had assumed was permanently, structurally safe. There’s an excellent book published in 2020 telling the whole story titled “Lights Out: Pride, Delusion, and the Fall of General Electric”.

GE was an original 1896 member of the Dow Jones Industrial Average, and for much of the twentieth century was treated as the definition of a blue-chip stock, the kind of holding you didn’t need to thinktwice about. In 2000 it was the most valuable company on earth, worth over $600 billion, run by Jack Welch, who was widely regarded as the best chief executive in America. By 2018, after a slow erosion driven mainly by an opaque and overextended finance arm, GE had lost more than 80% of its value and was removed from the Dow entirely, the last of the index’s original 1896 members to go, after 122 years.

Three companies, three completely different failure modes: fraud that unravelled in months, a market-cap anomaly that unwound over a couple of years, and a slow, two-decade erosion of a company nobody thought needed watching closely because it had always been safe. None of it was obvious in real time to the people holding the shares at the peak, which is precisely the point. A concentrated bet on the biggest, most exciting, or most trusted company of the moment can be permanently impaired for reasons that only become clear afterwards. A diversified index doesn’t need to see any of it coming. It just quietly reflects the change in weighting as it happens, whatever the cause turns out to be.

VII. The Mind Working Against Itself
If the arithmetic is this clear and the history this cautionary, a fair question follows: why does stock-picking, and concentrated sector betting, remain so popular, even among clever, careful, numerate people? I don’t think the answer is a lack of intelligence. I think it’s because human beings are, by nature, poorly wired for exactly the kind of decision index investing asks of us, and recognising that is as much a part of this as the arithmetic.

Start with prospect theory, the framework Daniel Kahneman and Amos Tversky set out in 1979, which later won a Nobel Prize. One of its central findings is that people don’t weigh probabilities in a straight line. Very small probabilities of a large gain get systematically overweighted, valued as though they were more likely than they really are. It’s the same mechanism that sells lottery tickets: a one-in-tens-of-millions chance feels meaningfully more hopeful to most people than the true odds justify. Nicholas Barberis and Ming Huang applied this directly to stock prices in 2008, showing that a stock with a small chance of a very large payoff, the profile of a speculative growth stock or a hot IPO, can become genuinely overpriced precisely because investors overvalue that small chance, accepting a lower expected return in exchange for the dream of the outsized one. Alok Kumar’s 2009 research adds a human detail: it’s overwhelmingly individual investors, rather than institutions, who hold these lottery-like stocks, and the appetite for them tracks closely with the appetite for gambling more generally.

Then there’s overconfidence, which I think is the bias that explains why this affects capable people just as much as anyone else, arguably more. FINRA’s (a US financial regulator, similar in part to the UK’s Financial Conduct Authority) 2024 National Financial Capability Study found that 56% of investors rate their own investment knowledge highly. That group answered more questions correctly on a basic investment knowledge quiz than those who rated their knowledge low, but they also gave more incorrect answers and used “don’t know” far less often; confidence and knowledge were related, but confidence didn’t mean an absence of real gaps. Overconfidence doesn’t feel like overconfidence from the inside. It feels like justified conviction.

Layer on that what researchers call the illusion of control, the tendency to feel more certain about an outcome the more directly involved we are in choosing it, and it becomes easier to see why an index fund can feel intellectually unsatisfying even once you accept the arithmetic. Choosing five stocks yourself, reading the annual reports, following the story, feels like exercising judgement. Buying the whole market feels, in some deep yet irrational way, like giving up on judgement altogether. I think that feeling is a very natural, very human response, and it’s exactly the response the evidence in this paper says to be suspicious of. The discipline isn’t pretending the desire for complexity and control doesn’t exist. Rather, it’s noticing it, and choosing the less satisfying-feeling option anyway, because the numbers are clear about which one serves you better.

VIII. A Coincidence Too Good to Leave Out
While I was in the middle of writing this, something landed in my newsfeed that illustrated the whole thing better than I could have planned it. On 22 July 2026, YouGov asked nearly 4,600 British adults a simple question: take a guaranteed £50,000, or flip a coin for a 50% shot at £1 million? 73% took the guaranteed money. Only 21% took the coin flip, even though it was worth ten times as much on average.

It’s a small thing, but it’s a near-perfect illustration of the probability-weighting point from the previous chapter, and it appeared in the news the same week I happened to be writing about it, what are the odds? I should be clear about what it does and doesn’t show: the survey tells us how people respond to an explicit, labelled gamble, it doesn’t tell us that the same 4,600 people go on to hold concentrated stock positions. I’m using it as an illustration of a documented bias, not as evidence that this specific group behaves a particular way with their investments.

People who would never take that coin flip when the odds are spelled out plainly will happily hold a concentrated stock position, because nobody hands you the odds in black and white when it’s dressed up as a company or a sector rather than a bet. There’s a story instead: a founder, a product, a trend. Strip the story away, and the underlying gamble isn’t so different in shape from the one three in four people just turned down flat.

IX. Taking the Other Side Seriously
If we’re aiming to take a balanced view, then it’s worth engaging with the strongest counterargument here, rather than a weak version of it, because I don’t think intellectual rigour allows for anything less. If Sharpe’s arithmetic requires that all active money, on average, match the passive return before costs, and Fama and French’s own data agrees that a small number of managers genuinely do show statistical evidence of skill before fees, then simple logic says something has to balance the books: some active money must be earning above-average returns before costs, and Bessembinder’s concentration findings confirm there’s real prize money available, in principle, for anyone who can identify tomorrow’s handful of extraordinary winners today. On paper, the opportunity exists.

In practice, it doesn’t survive contact with costs, or with time, or with the history set out in the earlier chapters of this paper. Fama and French’s skilled managers still fail to beat passive returns net of fees, the margin of genuine skill, where it exists, is thinner than the cost of accessing it. The SPIVA persistence data shows that even identifying which manager currently has that edge is close to a coin toss for the next few years running. And picking the individual companies directly carries its own trap, laid out plainly in the previous chapter: Enron, Nortel, and GE were all, at their respective peaks, exactly the kind of company a concentrated bet would have wanted to hold, for three entirely different reasons that only became visible in hindsight. Identifying a winner after the fact is easy. Holding the conviction to identify it in advance, and to keep holding it through every period when it looks like a mistake, is a different matter entirely (I can remember early Amazon investors playing the hokey-cokey, with their in-out-in-out investment strategy). It’s scary, and the previous chapter’s findings on overconfidence and the illusion of control suggest most of us are poorly equipped to manage that well, however capable we are in every other part of our lives.

So, the theoretical door isn’t fully closed. But the odds of walking through it, reliably, identified in advance, net of costs, sustained over the years a real financial plan needs to work, are remarkably small. That’s not a dismissal of the other side. It’s what the evidence, taken as a whole and followed through, shows.

X. A Few Numbers I Couldn’t Leave Out
Along the way, I kept a running list of facts that didn’t make it into the main argument above, but which stuck with me anyway, and which I think say something useful on their own. A few of them are worth sharing properly, rather than leaving on the cutting-room floor.

The one that started this chapter off was a figure I half-remembered from somewhere and went back to verify properly, because I didn’t trust my own memory of it. The best-performing individual stock in US market history, by total return on every dollar invested since 1926, isn’t a technology company, an oil company, or anything exciting at all. It’s Altria Group, the tobacco company formerly known as Philip Morris. According to Bessembinder’s own figures, a single dollar invested in it at the start of 1926 and left entirely alone, with every dividend reinvested, would have grown to $4,421,136 by the end of 2025. That is not a typo.

I find that genuinely remarkable, and not just for the size of the number. Philip Morris was, for most of its public life, about as unglamorous an investment as you could choose: a company selling a product in structural, multi-decade decline, facing continual regulation, litigation, and public disapproval. None of that stopped it compounding into the highest cumulative-return stock in Bessembinder’s study, largely because its unloved reputation kept its share price persistently cheap relative to the enormous, dependable cash it threw off, allowing decades of dividends to be reinvested at a discount. It’s a quietly humbling example of exactly the theme running through this whole paper: the companies that end up mattering most over the long run are very rarely the ones that feel, at the time, like the obvious story to back. Other names near the top of Bessembinder’s all-time cumulative-return list, for what it’s worth, are similarly unglamorous; industrial firms like Vulcan Materials and IBM, not the kind of names that get talked about at dinner parties.

The second number uses a different measure, total net shareholder wealth creation, and answers a question I asked myself while writing the tech concentration chapters earlier in this paper: is Apple still the single largest wealth creator by this measure, or has Nvidia overtaken it? As of Bessembinder’s most recent figures, Apple still holds the top spot, having generated roughly $5.02 trillion in total shareholder wealth since listing, about 5.5% of every dollar of net wealth the entire US stock market has created since 1926. Nvidia is startlingly close behind, at roughly $4.58 trillion, or 5.0% of the total, despite having only been listed since 1999. Almost all of Nvidia’s wealth, roughly $4.51 trillion of it, was created in the nine years between January 2017 and December 2025 alone. Apple took closer to four decades to build a total that Nvidia has all but matched from a standing start within less than a decade.

I don’t think that race is finished, and by the time you read this it may well have been settled one way or the other. What strikes me most isn’t the exact figures, which will be out of date within a year, it’s that the two companies now sit close enough on this measure that the order could plausibly change again. Even the two largest, most durable wealth creators in the market’s history, by Bessembinder’s own accounting, cannot be assumed to hold their ranking indefinitely, which is this paper’s whole argument playing out at the very top of the market.

XI. How This Came Together
I should say plainly: I didn’t do all the reading here alone and credit must go to where credit is due. I had an economics undergraduate with us on a work placement during the Summer, and I set him loose on the original papers, Sharpe, Fama and French, Bessembinder, the SPIVA data, with instructions to read them properly, pull out the exact figures, and check every citation back to its original source rather than relying on my second-hand summaries or half-remembered stats picked up from other people’s articles. Exactly the kind of legwork that’s easy to get subtly wrong when you’re quoting research from memory, and I wanted the numbers in this piece to be ones I could stand behind, not just ones that sounded right.
It’s worth saying, too, that this piece took longer to write than I expected and went through several more drafts than I’d like, precisely because the final version kept turning out to be more complicated than the tidy version I was aiming for. It would have been easier to write five hundred words asserting that index funds are obviously correct and stock-picking is obviously foolish.

I don’t think that version would have been worth your time, and I don’t think it would have been true to how I actually got here, which involved real head scratching, real cross-referencing between papers that don’t always sit together neatly, and a genuine fight with my own instinct, the same instinct described in the chapter on behavioural biases. The thinking and the writing in this paper are mine. The verification beneath came from Tom Bracey the economics undergrad who read everything with a second, careful, and usefully sceptical pair of eyes.

XII. Why This Is Our One Best Idea
Should you be worried about how much of a global index now sits in a small number of American tech companies? I don’t think the answer is “no, don’t worry about it.” Concentration is real, and it’s worth understanding properly, which is the whole point of this paper. But I also don’t think the answer is to try to avoid it by picking your own stocks, steering around the sectors that feel uncomfortably large right now, or hunting for the manager clever enough to do that for you. History says that’s exactly the kind of judgement call that has gone badly, repeatedly, for people who felt certain they’d spotted the risk in time, and the research says the feeling of certainty itself is one of the least reliable signals available to any of us.

The answer, after everything above, is closer to this: concentration is not a flaw in how markets work. It’s a description of how they’ve always worked. A tiny number of exceptional companies drive almost all the long-run gain, in every era, and the companies change even when the pattern doesn’t. Owning the whole market, rather than trying to guess which handful of names will matter next, is how you keep capturing that pattern without having to bet on which sector’s Nortel moment, or GE’s slow decline, is coming next, and without needing to out-think your own overconfidence to do it.

Our preference for indexing is not a claim that concentration risk disappears. It reflects our judgement about the alternatives; alongside how much equity exposure is appropriate within each client’s wider financial plan. The historical concentration of wealth creation doesn’t, by itself, tell us whether today’s largest companies are fairly priced; it tells us that trying to sidestep concentration by picking your own stocks or sectors has a poor track record, which is a different and more answerable question.

This is why, after everything I’ve read, cross-checked, and sat uncomfortably with over the past couple of months, index investing remains what I’d call our one best idea at Yellowtail, not because it’s the most exciting conclusion available, but because it’s the one that survives every angle I’ve tried to attack it from.

Don’t look for the needle in the haystack. Just buy the haystack!
– John C. Bogle

If you’d like to talk through what any of this means for you, we’d welcome the conversation.

This paper explains our general investment approach rather than recommending changes to your individual arrangements. Investments can fall as well as rise, and you may get back less than you invest. Diversification does not eliminate market or concentration risk. Past performance is not a reliable indicator of future results. Returns quoted in US dollars can differ from those experienced by a sterling investor because of exchange-rate movements.