Abstract
Portfolio investment theory is critical to global financial markets. However, the specific actual combination effect is affected by factors such as weight selection and asset selection. Based on the Markowitz model, we use a more reasonable weight selection method to establish an asset portfolio. This study demonstrates that strategically incorporating risk-free assets significantly lowers overall portfolio risk and improves risk-adjusted returns compared to portfolios consisting solely of risky assets.
Introduction
While the Markowitz model is a cornerstone of modern portfolio theory, it faces two major hurdles in practice: the extreme complexity of calculating covariance matrices for large numbers of assets and the difficulty of determining the "best" weights for each asset. Investors often struggle to balance the desire for high returns with the necessity of risk diversification. This research provides a more accessible way to use these mathematical models to find an "optimal" portfolio—one that offers the highest possible return for every unit of risk taken, as measured by the Sharpe ratio.
Methods
The researchers employed a multi-step analytical approach:
- Markowitz Mean-Variance Model: Used to establish the relationship between expected return and risk (variance). The goal is to find the "efficient frontier" where risk is minimized for a target return.
- Sharpe Ratio Evaluation: This served as the primary metric for efficiency. By dividing the excess return of a portfolio by its standard deviation, the authors identified optimal combinations.
- Weight Optimization Loop: To address weight selection complexity, a simplified simulation was used. Weights were divided between risky assets ($W_a$) and risk-free assets ($W_b$). $W_a$ was tested in intervals of 0.1 and equally distributed among chosen risky assets to find the peak Sharpe ratio.
- Comparative Analysis: The study compared portfolios without risk-free assets ("Portfolio_nf") against those including risk-free assets ("Portfolio_f") to quantify diversification benefits.
Datasets
The study utilized historical financial data sourced from Yahoo Finance, covering the period from January 1, 1968, to December 1, 1982. The model focused on four core variables: eqmrkt (equity market), valmrkt (value market), mkt (general market), and tbill (Treasury bills).
Key Insights and Ideas
- The "Risk-Free" Stabilizer: Adding risk-free assets (T-bills) disproportionately lowers risk compared to the reduction in returns. This leads to a higher Sharpe ratio, indicating a better risk-adjusted return.
- Diversification Beats Individual Assets: The standard deviation of a diversified portfolio (0.049) was lower than that of any single asset in the portfolio (the lowest individual was 0.055), validating the principle of diversification.
- The Dilution Trade-off: While adding T-bills reduced total cumulative return in some cases (e.g., from 154.80% to 143.46%), it created a more stable yield by shielding the portfolio from market volatility.
- Weighting Strategy: Using historical performance to determine weights, rather than simple equal weighting, significantly optimized portfolio impact.
Key Results
- Risk Reduction: Including risk-free assets reduced the portfolio's standard deviation from 0.049 to 0.047.
- Optimal Combinations: The loop-based optimization identified two "peak" portfolios achieving a Sharpe ratio of 0.8946: 20% Windsor Fund + 80% T-bills, and 70% Windsor Fund + 30% T-bills.
- Asset Performance: The
eqmrktasset showed the highest individual cumulative return (194.846%) but also the highest volatility, highlighting the need for the Markowitz approach to balance its variance.
Limitations and Future Work
The authors acknowledge several simplifications made to keep the model computationally feasible:
- Simplified Weighting: The use of 0.1 intervals is a heuristic that might miss the absolute mathematical optimum.
- Static Assumptions: The model assumes normal distribution and independent, identically distributed (i.i.d) data, which may not hold during extreme market events. Autocorrelation analysis (ACF) was used to inspect these assumptions.
Future work should focus on model optimization to handle more complex, real-world investing situations and a broader variety of asset classes beyond the ones tested.
Summary updated