The Illusion of Safety in Diversification
December 23, 2024
9 min read
The Illusion of Safety in Diversification
Reducing risk while increasing return sounds perfect. But what if diversification — our supposed safety net — is holding us back? It's not about owning more, but better. Our original analysis for The Market by NZZ plus the extended investor's view on the mathematics of real concentration.
In 30 seconds — the core thesis
- Markowitz 1952 is the cornerstone of modern portfolio theory — and his concept of diversification is intellectually solid. But its excessive application leads to what Peter Lynch called «diworsification».
- The diversification curve is mathematically clear: from 20-30 carefully selected uncorrelated stocks onwards, additional diversification benefits drop off drastically. More stocks bring more tracking complexity at unchanged risk.
- Real investment success comes from concentration, not from spread. A small number of exceptionally good companies structurally beats a broadly spread mediocrity — that's Markowitz's actual message, not the «buy everything» mentality erroneously derived from it today.
The original analysis — the opening
Reducing risk while increasing return: a thought that sounds too good to be true, doesn't it?
This idea sounds tempting and yet implausible — especially if you've learned that less risk equals more diversification. The concept of diversification has been a cornerstone of modern portfolio theory since Harry Markowitz's work in 1952. It teaches us that spreading investments across a broad spectrum of assets reduces overall risk by minimising the impact of a single investment's failure — the stock-specific risk.
While the premise is solid, I argue that its excessive application can lead to what legendary investor Peter Lynch called «diworsification»: an over-diversified portfolio where the quality of investments is compromised and returns are diluted. The result is the illusion of safety through diversification.
The key to investment success therefore doesn't lie in spreading investments across hundreds of companies. That's also not what Markowitz meant by diversification.
Rather, it's about concentrating on a small number of exceptionally good companies.
→ Read the full article on The Market by NZZ
Chart 1: The dangers of over-diversification — the effect is minimal after 20-30 stocks
Source: NZZ The Market
01The uncomfortable question: why do investors confuse stock count with diversification?
The diversification mathematics is intellectually trivial — and yet systematically misunderstood by the majority of investors. In an NZZ column the Markowitz-Lynch arc can be mentioned, but the depth dimension demands more: why do investors believe 500 stocks are automatically safer than 30 — even when the mathematics shows exactly the opposite?
Three mechanisms explain the structural misinterpretation:
- Linearity assumption. Investors intuitively assume each additional stock reduces risk evenly. Mathematically wrong — the diversification curve is exponentially flat. The first 10-15 stocks reduce risk drastically, then the curve runs only asymptotically flatter.
- Comfort-by-quantity. «I own 500 stocks» feels safer than «I own 25 stocks». This psychological comfort illusion ignores that the 500-stock variant often has 30-40% concentration in the top 10 (cf. Margin of Safety companion). Effective diversification is therefore often lower than in a disciplined 25-stock quality portfolio.
- Abandonment of selection responsibility. «If I own everything, I don't have to decide which.» This mentality is intellectually convenient, but structurally expensive. It hands selection completely over to market mechanisms that systematically over-allocate to the most expensive stocks in bull phases (index inflows mechanic).
What Markowitz really meant in 1952
Markowitz's original work «Portfolio Selection» describes the efficient frontier — the optimal risk-return combination for given stocks. His model shows most diversification benefits are achieved at 15-25 uncorrelated stocks, after that improvements are marginal. Markowitz meant: spread across uncorrelated business models, not across everything. Today's passive «buy everything» mentality is a gross simplification of his original statement. He would have been — historically documented — no fan of modern index concentration in mega-caps.
02The diversification curve — where the benefits really lie
The mathematics of diversification is 70 years old and unchanged. What's been added are empirical studies of the exact shape of the curve. They show a clear pattern:
Diversification benefit by stock count
5
~52% Risk
Too concentrated
10
~36% Risk
Substantial
20-30
~28% Risk
Sweet spot
100
~24% Risk
Marginally more
500+
~22% Risk
Diworsification
Risk value = standard deviation of portfolio as percent of standard deviation of a single stock. Simplified model based on empirical studies.
The important observation: between 25 stocks and 500 stocks, the diversification difference is maybe 4-6 percentage points of risk reduction. That's marginal. What is dramatically different:
Selection discipline. With 25 stocks you can carefully select each one, fundamentally analyse, regularly review. With 500 stocks you accept a blanket mix in which the majority of stocks are structurally mediocre or below-average. The diworsification costs aren't visible in the risk mathematics — but in the long-term return mathematics.
Effective diversification. 25 stocks from various business models, sectors, countries can be effectively more uncorrelated than 500 index stocks with 35-40% top-10 concentration. Real diversification comes from uncorrelated cash flow sources, not from many stock tickers.
The contradiction — and its resolution
If index investors have experienced above-average returns for 15 years, where does the diworsification critique come from? Answer: index returns of the last 15 years aren't an average period, but a historically extreme sample. A bull market with unusually strong mega-cap concentration has over-rewarded passive investors. Mean reversion to mean values is mathematically practically inevitable — only the timing is the question. Whoever understands this builds concentration on quality, not concentration on index fashion.
03What this means for your quality portfolio
The concentration logic is the central implementation question in disciplined quality strategy. Three strategic steps:
| Step | What to do |
|---|---|
| 1. Selection discipline as central activity | 20-30 carefully selected quality businesses structurally beat 500 index stocks. Selection is the actual value contribution of the active investor, not stock count. Invest most analysis time in selection, not market timing. |
| 2. Business model diversification, not just stock diversification | 20 stocks from 12 different business models (software, healthcare, consumer, industrial, utilities etc.) are structurally more uncorrelated than 500 stocks with 30%+ tech concentration. Business model spread beats stock spread. |
| 3. Position sizing as second diversification axis | Without position sizing discipline, even a 25-stock portfolio can be structurally concentrated — if the top 3 together make up 50%. With 4-7% per position, even a 20-stock portfolio remains effectively diversified. |
| Investor profile | Diversification status | What to review |
|---|---|---|
| "I hold an MSCI World ETF" | Nominally broad, effectively concentrated | Calculate effective top-10 concentration, gradually build quality diversification |
| "I hold 5-10 high-conviction positions" | Sub-optimal diversification | Gradually expand to 20-30 without compromising quality standard |
| "I hold 20-30 quality positions" | Structurally optimal | Hold discipline, regularly check position sizing |
| "I hold 100+ stocks" | Diworsification zone | Gradually consolidate to 20-30 with highest quality conviction |
How this concentration discipline complements the other half of quality strategy — namely business quality and valuation discipline — is deepened in our pillar "The Tortoise Problem". The two thematically closely related companions: "Margin of Safety is Razor Thin" and "Valuation Doesn't Matter — Until It Does".
04Three scenarios — how the concentration discussion will resolve
The diversification question isn't a daily topic, but the structural resolution of current market concentration will shape the next investor generation. Three plausible paths:
Bull Case
Index concentration normalises gradually without drama
Over 5-10 years, mega-cap valuations gradually align with broader market valuations. Index investors see 5-7% annual returns over the period — below historical average but positive. Quality investors with concentration discipline deliver 7-10% annually with significantly smaller drawdowns. Both groups make progress, quality is just structurally superior.
Base Case
Valuation reset reveals diworsification costs
A multiple contraction over 18-30 months (cf. Valuation companion) shows for the first time the real costs of passive index concentration. Index investors see drawdowns of 30-40% and very long recovery times. Quality investors with real diversification see 15-25% drawdowns and faster recovery. Over the following 5 years, the structural return difference in favour of real concentration becomes visible. Our base case.
Bear Case
Structural bear phase exposes nominal diversification as illusion
A longer bear phase (12-18+ months) reveals that nominally broad index diversification loses correlation benefits in crises — all mega-caps fall synchronously. Quality portfolios with truly uncorrelated business models maintain differentiation in the crisis. This period will reposition the Markowitz-Lynch discussion — many investors only then discover what «real» vs. «illusory» diversification means.
05What you should review now
An honest diversification inventory of your portfolio takes 60 minutes. Four concrete checks:
1. Calculate effective stock count. Not the nominal stock count, but the effective — weighted by position size. If your top 3 together make up 50%, your effective diversification is far lower than the stock count suggests. Rule of thumb: calculate the sum of (position weight)² — the inverse of that is your effective N.
2. Check business model spread. Write down the 5-7 main business models you're invested in (cloud software, healthcare large caps, consumer brands, etc.). If 80%+ of your positions are concentrated in 2-3 business models, your diversification is structurally weak — even if it's 50 stocks.
3. Correlation reality check. In what kind of drawdown would all your top-10 positions fall simultaneously? If the answer is «all in a tech bear market», you have correlation concentration not visible in diversification mathematics — until it is.
4. Plan consolidation path. If you hold 100+ stocks, plan a gradual consolidation process — not panic-sell, but at next savings amounts top up the 20-30 highest-conviction quality positions. Over 18-24 months the portfolio shifts organically to concentration discipline.
What disciplined investors do
They don't panic-sell their 500-stock indices — that would be reaction. They gradually build a quality concentration that gradually complements or replaces their ongoing index position. They hold selection discipline as central activity — 4-6 new, carefully analysed positions per year, plus periodic review of the existing 20-30. They define position sizing rules that enforce effective diversification — no position above 7%, no business model concentration above 25%. Over 5-10 years the portfolio becomes structurally superior — not through brilliant market forecasting, but through disciplined concentration on the essential. That's Markowitz as he meant it — and Lynch, who sees his diworsification warning historically confirmed.
06Frequently asked questions
Should I sell my index ETFs?
Categorically no. Wholesale sales often trigger taxes and are frequently poorly timed. What's recommended: gradually build real quality concentration — either through reallocation or through additional investments from ongoing savings plans. Over 18-36 months the portfolio shifts organically from index diworsification to disciplined concentration.
What if I don't have time for 25 quality analyses?
An honest answer. If time is missing, professional quality managers (like arvy) are a sensible alternative — they take over the selection discipline and concentration, you keep your lifetime. What doesn't work: a middle path in which you hold 50-100 stocks but can't analyse any. That's the worst position — diworsification costs without concentration benefit.
Are 20-30 stocks enough to diversify in a crisis?
Mathematically yes — provided the 20-30 are from truly uncorrelated business models. 20 cloud software stocks are practically one position in a tech crisis. 20 stocks from healthcare, consumer, utilities, industrial, software, banks are very differently exposed in a crisis. Selection decides, not number alone.
How does arvy construct concentrated quality portfolios?
arvy typically holds 25-35 quality positions from various business models with position sizing discipline (no position above 7%) and valuation filter. The concrete composition and construction logic you find transparently documented in the arvy Quarterly Report Q1 2026.
Further reading — the thematic anchors of this analysis
The Tortoise Problem
Quality concentration as return source
Margin of Safety is Razor Thin
Concentration risk in indices
Valuation Doesn't Matter — Until It Does
Valuation risk at concentration
The Crocodile Will Soon Snap
When mega-cap concentration tips
Why arvy as Partner
Concentrated quality investing in practice
arvy Quarterly Report Q1 2026
Current 25-35 position allocation
More isn't better — better is better
Markowitz's original work from 1952 is more cited than understood today. He didn't describe «buy everything, the market sorts», but «construct effectively uncorrelated portfolios with the minimum necessary stock count». His model isn't designed for today's index mega-cap concentration — it's designed for disciplined selection. Peter Lynch's diworsification critique from the 1980s practically encapsulates Markowitz's actual message: from a certain stock count onwards, more spread becomes a return brake without bringing notable risk benefit.
What separates the disciplined investor from the average is not the willingness to own all 500 stocks of the S&P 500. It's the willingness to carefully select 25-35 of them and consciously not own the rest. This selection discipline is the actual value creation of the active investor — and it's structurally rewarded over 30 years of investing life with above-average risk-adjusted returns. Markowitz would have wanted it that way. Lynch described it that way. Disciplined investors live it that way. The others buy 500 stocks and call it diversification.
Original written by Thierry Borgeat, Co-Founder of arvy, for The Market by NZZ. The extended arvy companion piece reviewed by Patrick Rissi, CFA and Florian Jauch, CFA. Data sources: NZZ The Market, own analyses, Markowitz «Portfolio Selection» (1952), Peter Lynch «One Up On Wall Street» (1989). Last updated: April 2026.
Disclaimer: This article is for general educational purposes and does not constitute personal investment advice. Past performance is no guarantee of future results. Scenarios are assessments, not forecasts. arvy is a FINMA-supervised asset manager with a CISA licence (Art. 24). Imprint & Legal Notice.
