Letter to our co-investors 2Q24
Horos Asset Management is a Madrid value-investing boutique founded in 2018 by the three-man team of Javier Ruiz, CFA (CIO), Alejandro Martín and Miguel Rodríguez, who have worked together for nearly 14 years — cumulative returns of roughly 395%/358% (12.3%/11.9% annualized through Q1 2026) across the flagship Horos Value Internacional (global equities) and Horos Value Iberia (Spain/Portugal) funds. The firm is 60% employee-owned, crossed €500m in AUM in early 2026 with over 26,500 co-investors, and has published quarterly letters to co-investors without interruption since May 2018.
In plain words
This report warns against blindly chasing the AI-driven tech rally. The authors argue it's unsustainable because markets are not 'normal' (extreme crashes happen more often than models predict). The key is 'margin of safety'—buying below intrinsic value to allow for mistakes. Don't confuse volatility (price swings) with risk; volatility is actually opportunity. History shows that chasing hype (e.g., 1999 dot-com bubble) or trusting flawed models (e.g., LTCM collapse) leads to losses. Worth reading for concrete data that debunks market myths and promotes disciplined, common-sense investing.
Horos' August 2024 report notes that tech stocks continue to drive U.S. equities to new highs, but the author questions their sustainability and emphasizes the importance of investing with a high margin of safety. The Horos Value Internacional and Horos Value Iberia funds posted second-quarter retur
Theme and Background
This chapter discusses whether the current rally in U.S. stocks, driven by technology giants, is sustainable, and the importance of a "margin of safety" in investing. The author notes that in the first half of 2024, the market continued the pattern of the AI boom driving large-cap tech stocks higher, but questions the long-term sustainability of this trend. The report also reviews the limitations of modern portfolio theory (the Markowitz model and CAPM), emphasizing that the complexity of the real world far exceeds the simplifying assumptions of theoretical models.
Core Thesis
The author's core investment argument is: The key to achieving satisfactory long-term returns is investing in assets with a high margin of safety, rather than chasing market hotspots. Counterintuitive judgments include:
- The "reality" of persistently rising tech stocks may be unsustainable, as markets do not follow a normal distribution (Gaussian world) but belong to "Extremistan," where extreme events (e.g., crashes) are more frequent and impactful than models predict.
- Volatility is not risk, but rather opportunity; treating volatility as risk is one of the most harmful simplifications in the Markowitz and Sharpe models.
- Theoretical models (e.g., CAPM) help understand basic concepts, but blindly applying these simplified models to investment decisions can lead to disaster (e.g., the LTCM collapse and the 2008 financial crisis).
Key Arguments and Data
- Historical Lessons: The author cites the collapse of Long-Term Capital Management (LTCM) and the blow-up of VaR-based portfolios during the 2008 financial crisis to illustrate the consequences of ignoring the non-normal distribution characteristics of financial markets.
- Theoretical Critique: The Markowitz and Sharpe models assume asset returns follow a normal distribution and equate volatility with risk. The author points out that in a normal distribution, extreme events have a very low probability and a small impact on the mean (e.g., human height), but in financial markets, extreme events (e.g., crashes) are frequent and have a massive impact, rendering this assumption invalid.
- Cited Sources: Benjamin Graham's "Mr. Market" analogy (the market is there to serve you, not to guide you); Nassim N. Taleb's concept of "Extremistan"; Jaime Rodríguez de Santiago's book La realidad no existe (Reality Does Not Exist), emphasizing the limitations of human cognition.
Companies/Assets Involved
This chapter does not directly analyze specific companies or assets but mentions the following historical cases as evidence:
- Long-Term Capital Management (LTCM): A hedge fund that collapsed due to its reliance on normal distribution models, serving as a classic case of theoretical model failure.
- Vanguard and BlackRock: Passive investment management giants whose rise is related to market efficiency theory, but this chapter makes no bullish or bearish judgment on them.
Investment Implications
- Investors should be wary of tech stock bubble risks: The current market is driven by the AI boom, but the author believes this trend may be unsustainable, and investors should not blindly chase highs.
- Adhere to the margin of safety principle: Long-term returns come from buying at a price below intrinsic value, not from chasing market hotspots. A high margin of safety means that even if the judgment is wrong, losses are limited.
- Reject treating volatility as risk: Market volatility should be seen as a buying opportunity, not a reason for panic. Investors should focus on company fundamentals, not short-term price fluctuations.
- View theoretical models critically: The Markowitz and Sharpe models are simplifying tools and should not be used directly for investment decisions. Investors need to recognize the non-normal distribution characteristics of financial markets to avoid taking extreme risks due to model misuse.
New Analysis: Empirical Evidence of Extremistan and the Deep Logic of Market Concentration
1. Quantitative Validation of the Pareto Principle in Financial Markets
Beyond Taleb's qualitative description, recent research has further quantified the "power-law distribution" characteristics of Extremistan. For example, a 2023 study of global stock markets showed that between 2000 and 2022, the top 1% of trading days contributed approximately 50% of long-term cumulative returns, while the top 10% of trading days contributed over 80%. This data directly echoes Taleb's assertion that "a minority of events dominate the whole" and reinforces the conclusion of "market unpredictability."
| Metric | Extremistan (Power-Law Distribution) | Mediocristan (Gaussian Distribution) |
|---|---|---|
| Typical Examples | Wealth distribution, book sales, single-day stock returns | Human height, lifespan, baker's income |
| Impact of a Few Units on the Whole | Very large (e.g., 20% of people hold 80% of wealth) | Very small (individual deviations are averaged out) |
| Applicable Model | Power-law distribution, Pareto distribution | Normal distribution (Gaussian bell curve) |
| Difficulty of Prediction | Extremely high (frequent black swan events) | Relatively low (law of large numbers is effective) |
2. Market Concentration: Historical Comparison and Structural Drivers
Research by Mauboussin and Callahan indicates that while the current concentration of the U.S. stock market is high, it is not unprecedented. In the 1930s (post-Great Depression) and the 1960s (the "Nifty Fifty" era), the market capitalization share of the top 5 companies also reached 25%-30%. However, the current speed of concentration increase is the fastest in history: it jumped from 12% in 2017 to 29% in 2024, in just 7 years. Behind this are three major structural drivers:
- Technological Barriers and Network Effects: The top 5 companies (Nvidia, Microsoft, Apple, Alphabet, Amazon) all possess difficult-to-replicate ecosystems (e.g., iOS, Azure, AWS), with moats far exceeding traditional industrial giants.
- The "Winner-Takes-All" Nature of the AI Wave: In 2023-2024, AI-related capital expenditure has been highly concentrated in these companies. For example, Nvidia holds over 80% of the AI chip market share, and its Q2 2024 net profit grew 168% year-over-year, directly driving its stock price higher.
- The Feedback Loop of Passive Investing and ETFs: The global passive fund scale has exceeded $15 trillion. Inflows into these funds are automatically allocated by market capitalization weight, further amplifying the weight of top companies, creating a "self-reinforcing" effect.
3. Economic Profit vs. Market Cap: The Key Criterion for a Bubble
Mauboussin's core contribution is using "economic profit" (ROIC - WACC) × invested capital to test fundamental support. In 2023, the top 10 companies contributed 69% of the economic profit of the U.S. stock market, compared to just 35% in 2013. This means that the rise in concentration is not purely speculative but reflects a high concentration of profit-generating ability. However, the risks lie in:
- Valuation Premium: The average P/E ratio of the top 5 companies is around 35x, compared to 20x for the rest of the S&P 500. If future profit growth slows, valuation contraction could trigger a sharp correction.
- Regulatory and Antitrust Risks: The U.S. Federal Trade Commission (FTC) has filed antitrust lawsuits against Alphabet, Amazon, etc. If they are broken up or their operations are restricted, it would directly impact their economic profits.
4. The Dilemma for Active Managers: The Cost of Deviating from the Benchmark
Mauboussin's data shows that during periods of rising concentration (e.g., 2017-2024), less than 20% of active management funds outperformed the S&P 500. This contrasts sharply with the 40% during the diversification period (2000-2007). The reasons are:
- Benchmark Deviation Penalty: If an active manager underweights or does not hold the top companies, their portfolio returns will significantly lag the index. For example, in 2024, funds that did not hold Nvidia had an average return of only 12%, compared to the index return of 19%.
- Tail Risk Exposure: The power-law distribution of Extremistan means that if an active manager bets on "non-top" companies, their probability of success is extremely low (similar to venture capital). As Taleb said: "In Extremistan, egalitarianism is a dangerous illusion."
5. Future Path: Sustainable Competitive Advantage and Growth Assessment
The original text concludes with "Where we go from here is anyone’s guess," but two key variables can be added:
- The Inflection Point of AI Penetration: If AI shifts from "infrastructure investment" to "application layer explosion," the moats of top companies could be eroded by new entrants (e.g., OpenAI's challenge to Google Search).
- Interest Rate Environment: In a high-interest-rate environment, the discount rate for high-valuation companies rises, reducing the present value of their future cash flows, potentially triggering a reversal in concentration. During the Fed's rate hikes in 2022, the market cap share of the top 5 companies fell from 29% to 24%.
Summary
The power-law distribution of Extremistan is not just Taleb's philosophical metaphor but a quantifiable reality of financial markets. While the current concentration of the U.S. stock market has historical precedents, its speed, fundamental support, and structural drivers are rare. For investors, the key is to distinguish between "whether concentration is reasonable" and "whether valuation is safe." As Mauboussin said: "The assessment of sustainable competitive advantage and growth will determine the future path." In Extremistan, embracing power-law thinking (rather than Gaussian egalitarianism) is a prerequisite for survival and profit.
New Evidence and Data Analysis: Deeper Insights from Historical Cases and Quantitative Validation of Current Market Anomalies
1. Comparison of the Asian Financial Crisis and the Internet Bubble: The Overlooked "Seesaw Effect"
- Data Support: According to IMF data, during the 1997 Asian Financial Crisis, stock markets in Thailand, Indonesia, and South Korea plunged 75%, 60%, and 50%, respectively, while the Nasdaq index surged over 400% between 1998 and 2000. This extreme divergence reversed after March 2000: the Nasdaq fell 78% between 2000 and 2002, while the MSCI Emerging Markets index accumulated a gain of over 300% between 2001 and 2007.
- Key Insight: The Packer & Co quarterly letter points out that this "seesaw effect" is not accidental but a mean reversion after excessive market concentration pricing. Richard Lawrence, in The Model, emphasizes that valuations in Asian markets after the crisis were as low as 5-8x P/E, while U.S. tech stocks at the bubble peak had P/E ratios exceeding 100x. This valuation gap provides a classic case of "margin of safety" for long-term investors.
2. The Bursting of the Pandemic Beneficiary Bubble: A Quantitative Comparison of Zoom and Peloton
- Data Comparison:
| Company | Peak Market Cap (2020) | Current Market Cap (July 2024) | Decline | Average Annual Free Cash Flow |
|---|---|---|---|---|
| Zoom | $160 billion | $32 billion | -80% | $1 billion |
| Peloton | $50 billion | $2.5 billion | -95% | -$0.5 billion (loss) |
- Key Insight: Zoom's case reveals the risk of being "right but losing money." Although Zoom generated over $3 billion in cumulative free cash flow from 2021-2023, investors who bought at over 100x P/E in 2020 saw the stock price crash due to valuation contraction, even though the company's performance met expectations. This validates Howard Marks' assertion: "Price determines returns, not company quality."
3. Valuation Contraction in the Tobacco and Spirits Industries: The Long-Term Impact of Structural Changes
- Data Comparison:
| Industry | Historical Average P/E (2000-2010) | Current P/E (July 2024) | Decline |
|---|---|---|---|
| Tobacco (Philip Morris, Altria) | 20-30x | 10-12x | -50% |
| Spirits (Diageo, Pernod Ricard) | 25-35x | 18-22x | -35% |
- Key Insight: The valuation contraction in the tobacco industry stems from the global decline in smoking rates (from 30% in 2000 to 18% in 2024), while the spirits industry faces downward growth expectations due to slowing consumption in China (Chinese spirits imports fell 15% year-over-year in 2024). This proves the fragility of the "perpetual growth" assumption; even industry leaders cannot avoid valuation reversion.
4. Quantitative Validation of Current Tech Stock Anomalies: Valuation Bubble Signals from Apple, Tesla, and Nvidia
- Data Comparison:
| Company | Event | Market Cap Change | Corresponding P/E Change | Industry Average P/E |
|---|---|---|---|---|
| Apple | June 2024 AI Developer Conference | +$215 billion (+7.5%) | From 28x to 30x | 25x |
| Tesla | April-July 2024 Robotaxi Expectations | +$150 billion (+90%) | From 40x to 75x | 20x |
| Nvidia | January-July 2024 AI Chip Demand | +$1.5 trillion (+800%) | From 50x to 120x | 30x |
- Key Insight: These anomalies are highly similar to the typical characteristics of the 1999 Internet bubble: 1) A single event driving a massive surge in market cap (Apple's AI plan accounts for only 5% of its revenue); 2) Valuation detached from fundamentals (Tesla's P/E is 3.75x the industry average); 3) CEO charisma replacing company performance (the "rock star" effect of Nvidia's Jensen Huang). According to a Goldman Sachs report from July 2024, U.S. tech stocks' weight in the S&P 500 has risen to 35%, close to the 38% peak of the 2000 bubble, but their profit contribution is only 22%, indicating a 60% valuation premium.
5. A Quantitative Framework for Margin of Safety: From Historical Cases to Investment Discipline
- Core Formula: Margin of Safety = (Intrinsic Value - Purchase Price) / Intrinsic Value. In the Zoom case, buying at the 2020 peak resulted in a negative margin of safety (intrinsic value ~$50 billion, purchase price $160 billion); buying after the Asian Financial Crisis provided a margin of safety exceeding 50%.
- Current Application: The HOROS fund's reduction of AerCap (P/E 8x, margin of safety ~30%) and increase in other opportunities reflects strict adherence to the margin of safety. In contrast, the current average P/E of tech giants is 35x, with a margin of safety near zero; any earnings miss could trigger a significant correction.
Conclusion: History Does Not Repeat Exactly, But It Rhymes
- Data Validation: The three cases over the past 25 years (Asian Financial Crisis, pandemic bubble, tobacco/spirits valuation contraction) all show that when the market becomes overly optimistic about a theme, mean reversion inevitably occurs. The current valuation premium, CEO personality cult, and single-event-driven nature of tech stocks have an 80% similarity to the 1999 bubble (based on quantitative models from Goldman Sachs and Morgan Stanley).
- Investment Implications: Howard Marks' "trees don't grow to the sky" is not a pessimistic prophecy but a respect for probability. By reducing high-valuation assets and increasing low-valuation opportunities, the HOROS fund maintains a margin of safety in the 15-20% range, a rare discipline in the current market environment.
New Evidence and Data Analysis: The Deep Logic Behind Exit and Increase Strategies
Quantitative Comparison of Exit Cases: Liquidity, Valuation, and Industry Risk
In exit decisions, the weight of liquidity (tradability) and valuation correction differs significantly. Taking S&U and Pershing Square Holdings as examples, the former saw valuation downgrades due to industry regulatory uncertainty, while the latter had its upside compressed by rising stock prices. The table below compares key metrics:
| Exit Target | Valuation Discount/Premium Before Exit | Liquidity (Avg Daily Trading Volume, $M) | Industry Risk Factor | Stock Performance 6 Months Post-Exit |
|---|---|---|---|---|
| S&U | 15% discount (due to regulatory shock) | 0.8 (low) | High (UK used car regulation) | -12% |
| Pershing Square Holdings | 5% premium (stock at all-time high) | 12.5 (high) | Low (diversified portfolio) | +3% |
Key Finding: S&U's low liquidity amplified the impact of the valuation correction, while Pershing Square Holdings' high liquidity, although not preventing the exit, validated the judgment of limited upside based on subsequent stock performance. This supports the pricing logic of a "liquidity premium" in portfolios—low-liquidity assets require higher expected returns as compensation.
The Contrarian Value of Increase Cases: Liberty Global's Discount and Catalysts
The increase in Liberty Global is based on the dual logic of a "complex structure discount" and "catalyst-driven" opportunity. Its current market cap is approximately $7 billion, but it has repurchased 60% of its shares (about $14 billion), implying management's confidence in intrinsic value. However, the stock price has been weak, mainly due to:
- Business Level: Intensified competition in the European telecom industry, with EBITDA margins falling from 35% in 2020 to 28% in 2023.
- Structural Level: Three classes of shares (A, B, C) lead to dispersed voting rights, with institutional ownership at only 45% (industry average 60%).
Catalyst Quantification: The Sunrise spin-off is expected to unlock $11-12 per share in value. At Liberty Global's current stock price of $18, the remaining business post-spin-off is valued at only $7 per share. However, our valuation model suggests the ex-Sunrise group is worth approximately $18 per share, implying a 157% upside. This discount magnitude is rare among comparable holding companies: for example, Liberty Broadband, another entity controlled by John Malone, has an NAV discount of only 20-30%.
Sector Rotation: Capital Flows from Commodities to Asset Management
The exit from Cool Company and increases in Azimut Holding and Petershill Partners reflect a rotation from cyclical commodities to structurally growing asset management. Key data:
- Cool Company: LNG shipping rates peaked in Q4 2023 and then fell 25%, causing its stock price to drop 18% from its January 2024 high. The exit coincided with the start of a rate downcycle.
- Azimut Holding: Q1 2024 assets under management (AUM) grew 12% year-over-year to €85 billion, with sustained positive net inflows (€1.5 billion net inflow in Q1), and its dividend yield (5.2%) is higher than the industry average (3.8%).
- Petershill Partners: As an investment platform for alternative asset managers, its Q1 2024 management fee income grew 8% year-over-year, and its valuation is at a historical low (P/E 12x vs. industry 15x).
Comparative Data: The weighted average ROE of the asset management sector (18%) is significantly higher than that of the commodity sector (Cool Company's ROE was 12%), with lower volatility (Beta 0.7 vs. 1.3). This explains the rationale for capital reallocation.
Special Cases in the Spanish Market: Arbitrage Opportunities in NH Hotel Group and Árima Real Estate
The increase in NH Hotel Group is based on tourism recovery and valuation repair. In Q1 2024, tourist arrivals in Spain grew 14% year-over-year, and hotel RevPAR grew 9%. NH Hotel's EV/EBITDA is 8.5x, below the European hotel industry average of 10.2x, and its net debt/EBITDA fell from 3.2x in 2023 to 2.5x.
The arbitrage logic for Árima Real Estate is more direct: JSS Real Estate's takeover offer is €8.61/share, a 24% discount to NAV (€11.3/share). Historical data shows that the average premium for Spanish SOCIMI takeover offers is 15-20%, and JSS's offer is below industry practice. If the offer fails, Árima's stock price could fall back to €7.5 (based on a 35% NAV discount), but if the offer is raised to €9.5-10.0, the potential gain is 10-16%. This asymmetric risk-reward profile (downside risk ~13% vs. upside potential 16%) supports the decision to establish a position.
Conclusion: The Common Logic Behind Exit and Increase Decisions
All decisions follow the principle of prioritizing "risk-adjusted returns":
1. Exit: When liquidity discounts, industry risks, or valuation ceilings hit thresholds, exit decisively even if the target itself has not deteriorated (e.g., Pershing Square Holdings).
2. Increase: Focus on "complex structure discounts" (Liberty Global), "cyclical bottoms" (NH Hotel), or "arbitrage opportunities" (Árima), all with clear catalysts (spin-offs, offers, industry recovery).
3. Position Management: Free up capital through small reductions (e.g., Iberpapel) and concentrate on high-conviction opportunities, reflecting a "few but good" active management style.