'Don't put all your eggs in one basket' is the most common investment advice, but do you know how many baskets are actually effective? What is systematic risk? Starting from Modern Portfolio Theory (MPT) and the Efficient Frontier, this article breaks down the practical applications of correlation coefficients and Beta, using the 2026 AI stock correction to illustrate the trap of 'false diversification.'
NI Editorial Team
Comprised of senior wealth management, global markets, and fintech professionals
"Diversification reduces risk" is only half true. To understand its limitations, you must first grasp the structure of risk.
Also called company-specific risk or diversifiable risk. Sources include:
These risks are individual events, unrelated to the overall market. By holding enough uncorrelated assets, this type of risk can be effectively diluted.
Also called market risk or non-diversifiable risk. Sources include:
This type of risk affects all assets. No matter how many stocks you hold, you cannot escape. In 2008, whether tech, consumer, or financial stocks, all dropped over 40% — that is the power of systematic risk.
Core Conclusion: Diversification can only eliminate "unsystematic risk" and is virtually ineffective against "systematic risk." This is why even holding 100 stocks still results in heavy losses during the 2008 financial crisis.
---
In 1952, Harry Markowitz published a groundbreaking paper in The Journal of Finance, later called Modern Portfolio Theory (MPT), earning him the 1990 Nobel Prize in Economics.
His core insight: Investors should not focus solely on individual asset returns, but on the risk-return relationship of the entire portfolio.
Markowitz proposed the "Efficient Frontier" concept: among all possible portfolio combinations, there exists a curve representing the set of portfolios with the highest return at each risk level — or the lowest risk at each return level.
Portfolios on the Efficient Frontier are "efficient"; those below it mean that at the same risk level, returns are insufficient and should be adjusted.
This concept tells us: stock selection cannot focus only on individual stock quality — you must evaluate how it contributes to the overall portfolio's risk-return structure when added.
---
This is the most frequently asked question. Academic research provides relatively clear answers.
The classic Evans & Archer (1968) study showed:
Subsequent research (such as Statman's 1987 revision) suggests that considering transaction costs and taxes, the optimal number of holdings is approximately 30-40.
This is the most common misconception. In Q1 2026, the AI sector corrected over 20%, and many investors holding 20 "tech stocks" found they all dropped together — because the correlation coefficient among these 20 stocks was close to 1, providing zero diversification effect.
Real Example: Holding TSMC, MediaTek, Wistron, Quanta, and Inventec — these appear to be 5 stocks, but they are all highly exposed to the same "AI supply chain" risk factor. When NVIDIA drops, all 5 fall. This is a classic case of "false diversification."
---
What truly determines diversification effectiveness is the Correlation Coefficient between assets, ranging from -1 to +1.
r = +1: Perfect positive correlation — Two assets rise and fall together; zero diversification effect
r = 0: Uncorrelated — Two assets move independently; holding both reduces volatility
r = -1: Perfect negative correlation — One rises while the other falls; theoretically can completely hedge risk
U.S. stocks vs. Taiwan stocks: Approximately +0.55-0.70, moderate positive correlation (stronger when tech weighting is high)
Stocks vs. Government bonds: Traditionally approximately -0.2 to -0.4, offering some hedging function. But 2022's "stocks and bonds falling together" reminds us that in high-inflation environments, this negative correlation can disappear.
Stocks vs. Gold: Approximately -0.1 to +0.1, nearly uncorrelated, though gold's safe-haven properties typically emerge during crises.
Dollar vs. Commodities: Approximately -0.3 to -0.5; dollar depreciation typically benefits raw materials.
Important Reminder: Correlations are not fixed! During market crises (e.g., 2008, March 2020), previously low-correlated assets often crash simultaneously — academically termed "Correlation Breakdown." Diversification's most vulnerable moment is precisely when you need its protection most.
---
Since systematic risk cannot be diversified away, investors need a metric to measure how sensitive their portfolio is to market fluctuations — that is Beta.
Beta = 1: Asset moves in lockstep with the market (e.g., broad market ETF)
Beta > 1: Asset is more volatile than the market; e.g., Beta = 1.5 means when the market drops 10%, the asset is expected to drop 15%
Beta < 1: Asset is less volatile than the market; e.g., utility stocks Beta ~0.3-0.5
Beta < 0: Asset moves inversely to the market (e.g., some gold mining stocks under specific conditions)
If your overall portfolio Beta is 1.3, your risk exposure exceeds the market average. When you believe downside market risk exists, you can add low-Beta assets (Treasuries, cash, utility stocks) to reduce overall Beta, achieving "risk adjustment."
---
Truly effective diversification should not be limited to "buying more stocks" — it should span asset classes:
Equities (growth engine): Taiwan stocks + U.S. stocks + small allocation to emerging markets. The moderate positive correlation between Taiwan and U.S. stocks means diversifying between just these two is insufficient.
Bonds (stabilizer): Government bonds perform well during non-inflationary crises, serving as the traditional tool for hedging equity downside risk. Recommended maturity should not be too long (under 5-10 years) to control interest rate risk.
Real assets (inflation hedge): Gold, REITs, or commodity ETFs can provide defense in inflationary environments.
Cash or money market instruments (flexibility): Maintain 10-20% in cash for opportunistic buying during significant market corrections.
Many believe "buying U.S. stocks" diversifies Taiwan stock risk. But in 2022, Taiwan and U.S. stocks fell in tandem; during the 2024 AI stock crash, both were simultaneously hit. When global risk appetite declines simultaneously, international diversification effects diminish significantly.
---
Diversification is not a "set it and forget it" strategy. Market fluctuations cause asset weights to drift from the original allocation, requiring periodic rebalancing.
For example: you set 60% stocks, 30% bonds, 10% gold. After a year of strong stock market gains, stock weight may become 75%, meaning your portfolio bears more systematic risk than originally planned. Rebalancing involves selling some stocks and buying bonds and gold to return to the original allocation.
Academic research shows that quarterly or semi-annual rebalancing can, over the long term, significantly reduce portfolio volatility while achieving similar returns.
從零開始學投資理財