18.1 Risk-Adjusted Return Measures: Sharpe, M-Squared, Treynor & Jensen's Alpha

18.1 Risk-Adjusted Return Measures: Sharpe, M-Squared, Treynor & Jensen's Alpha

Evaluating investment performance purely on nominal or absolute returns is fundamentally flawed. A portfolio generating a 15% return by taking extreme leverage or concentrated speculative risks may represent inferior management compared to a portfolio delivering 12% through disciplined, low-volatility execution. Performance measurement provides the analytical framework to evaluate returns in the context of the risk assumed, isolating genuine investment skill from passive market exposure.

                     Risk-Adjusted Performance Measurement Architecture
                                              │
         ┌────────────────────────────────────┼────────────────────────────────────┐
         │                                    │                                    │
   Total Risk Metrics                 Systematic Risk Metrics             Benchmark-Relative Metrics
(Standard Deviation: σ)                     (Beta: β)                       (Tracking Error: TE)
  • Sharpe Ratio                       • Treynor Ratio                     • Information Ratio (IR)
  • Modigliani-Modigliani (M²)         • Jensen's Alpha (α)                • Active Return / Active Risk
  ► Standalone / Total Wealth          ► Sub-Allocations / Diversified     ► Mandate / Manager Selection

---

1. Total Risk-Adjusted Measures: Sharpe Ratio & Modigliani-Modigliani ($M^2$)

The Sharpe Ratio

Introduced by Nobel laureate William F. Sharpe in 1966 (originally termed the reward-to-variability ratio), the Sharpe Ratio measures the excess return earned per unit of total risk, as quantified by the annualized standard deviation of returns ($\sigma_p$):

$\text{Sharpe Ratio} = S_p = \frac{R_p - R_f}{\sigma_p}$

Where:

• $R_p$ = Realized annualized return of the portfolio
• $R_f$ = Risk-free rate of return over the measurement period (typically short-term Treasury bill yield)
• $\sigma_p$ = Annualized standard deviation of portfolio returns
• $R_p - R_f$ = Portfolio excess return (risk premium earned above cash)

   Expected Return E(R)
     ▲
     │                                              / Capital Allocation Line (CAL)
     │                                            /   Slope = Sharpe Ratio (Sp)
     │                             .-''''''-.   / 
     │                         .-'          `● Portfolio P (Rp, σp)
     │                       .'             /│
     │                     .'              / │
     │     GMVP ───────●                 /   │
     │                  `.              /    │ Excess Return = Rp - Rf
     │  Risk-Free (Rf) ──●─────────────/─────┼──────────
     │                   │            /      │
     └───────────────────┴───────────/───────┴───────────────────► Total Risk (σ)
                         0          /        σp

Key Characteristics and Applications:

1. Total Risk Focus: Because standard deviation encompasses both systematic (market) and unsystematic (idiosyncratic) risk, the Sharpe ratio is the mandatory metric when evaluating an investor's entire portfolio or undiversified holdings.
2. Geometric Interpretation: In mean-variance space ($E(R)$ vs. $\sigma$), the Sharpe ratio represents the slope of the Capital Allocation Line (CAL) connecting the risk-free asset to the risky portfolio. A steeper slope denotes superior return efficiency per unit of volatility.
3. Ex-Post vs. Ex-Ante: Ex-ante, the Sharpe ratio guides optimal asset allocation (identifying the tangency portfolio). Ex-post, it evaluates realized historical performance.

Practical Limitations of the Sharpe Ratio:

• Asymmetric Return Distributions: The Sharpe ratio assumes normally distributed returns. For hedge funds, private equity, or option strategies exhibiting negative skewness (fat left tails) or excess kurtosis, the Sharpe ratio significantly understates downside tail risk.
• Negative Excess Returns: When $R_p < R_f$, the Sharpe ratio becomes negative. In this scenario, ranking portfolios becomes misleading: a portfolio with higher volatility will produce a Sharpe ratio closer to zero (e.g., $-2% / 20% = -0.10$) than a lower-volatility portfolio (e.g., $-2% / 5% = -0.40$), falsely making the riskier fund appear superior.

---

The Modigliani-Modigliani ($M^2$) Measure

Developed by Franco Modigliani and Leah Modigliani in 1997, the Modigliani-Modigliani measure ($M^2$) translates the Sharpe ratio into intuitive percentage-return units. It quantifies the return the portfolio would have achieved if it had been leveraged or deleveraged (by borrowing or lending at the risk-free rate) to match the exact volatility of the market benchmark ($\sigma_m$).

$M^2 = R_f + S_p \cdot \sigma_m = R_f + \left( \frac{R_p - R_f}{\sigma_p} \right) \sigma_m$

Alternatively, $M^2$ excess return over the market is expressed as:

$M^2 - R_m = S_p \cdot \sigma_m - (R_m - R_f) = (S_p - S_m) \cdot \sigma_m$

   Return R
     ▲                                               CAL (Portfolio P)
     │                                              /
     │                                            .● Leveraged P* (M² Return, σm)
     │                                          .' /│
     │                             Portfolio P ●  / │
     │                                       .'│ /  │
     │                                     .'  │/   │
     │                     Benchmark (M) .●────┼────┼────────────
     │                                 .' │    │    │
     │  Risk-Free (Rf) ───────────────●   │    │    │
     └────────────────────────────────────┴────┴────┴────────────► Standard Deviation (σ)
                                         σm    σp   σm (Standardized)

Why Institutional Consultants Use $M^2$:

• Direct Percent Comparability: While board trustees or private clients often struggle to interpret a Sharpe ratio of $0.68$ vs. $0.54$, an $M^2$ showing that "Portfolio A earned $13.2%$ vs. Benchmark $11.0%$ at equal risk" is immediately clear.
• Identical Ranking to Sharpe: Because $M^2$ is a linear transformation of the Sharpe ratio ($M^2 = R_f + S_p \cdot \sigma_m$), portfolio rankings under $M^2$ and Sharpe are mathematically identical.

Worked Example: Calculating Sharpe and $M^2$

An endowment board evaluates Fund Alpha against the S&P 500 Index. Risk-free rate $R_f = 3.0%$.

• Fund Alpha: $R_p = 14.0%$, $\sigma_p = 22.0%$
• S&P 500 Index ($M$): $R_m = 10.0%$, $\sigma_m = 16.0%$

1. Sharpe Ratios: $S_{\text{Alpha}} = \frac{14.0\% - 3.0\%}{22.0\%} = \frac{11.0\%}{22.0\%} = 0.500$ $S_{\text{Market}} = \frac{10.0\% - 3.0\%}{16.0\%} = \frac{7.0\%}{16.0\%} = 0.4375$
2. Modigliani-Modigliani ($M^2$): $M^2_{\text{Alpha}} = 3.0\% + (0.500 \times 16.0\%) = 3.0\% + 8.0\% = 11.00\%$
3. Interpretation: Fund Alpha delivered superior risk-adjusted return ($S = 0.500 > 0.4375$). Deleveraging Alpha to match the market's 16% volatility yields an $M^2$ return of $11.00%$, generating $1.00%$ of pure risk-standardized outperformance over the benchmark's $10.00%$.

---

2. Systematic Risk-Adjusted Measures: Treynor Ratio & Jensen's Alpha

When an investment strategy is not held in isolation but rather serves as a component within a well-diversified master portfolio, unsystematic risk is diversified away. In this context, total risk ($\sigma_p$) is no longer the relevant risk metric; performance must be standardized against systematic market risk (Beta: $\beta_p$).

The Treynor Ratio

Developed by Jack Treynor in 1965, the Treynor Ratio measures excess return earned per unit of systematic risk ($\beta_p$):

$\text{Treynor Ratio} = T_p = \frac{R_p - R_f}{\beta_p}$

Where:

• $\beta_p = \frac{\text{Cov}(R_p, R_m)}{\sigma_m^2} = \rho_{pm} \cdot \frac{\sigma_p}{\sigma_m}$ = Portfolio Beta relative to the market benchmark

   Expected Return E(R)
     ▲
     │                                              / Security Market Line (SML)
     │                                            /   Slope = Market Risk Premium
     │                             Portfolio P  .●
     │                                        .' │
     │                     Benchmark (M)    .●───┼───────────
     │                                    .' │   │ Excess Return = Rp - Rf
     │  Risk-Free (Rf) ──────────────────●   │   │
     └───────────────────────────────────────┴───┴───────────────► Systematic Risk (β)
                                            1.0  βp

Key Characteristics:

• Systematic Risk Focus: Evaluates return per unit of market sensitivity. It assumes the investor can eliminate idiosyncratic risk through broad multi-asset diversification.
• SML Slope Comparison: In CAPM space ($E(R)$ vs. $\beta$), the Treynor ratio represents the slope of the line connecting $R_f$ to portfolio $P$. A portfolio with $T_p > (R_m - R_f)$ plots above the Security Market Line (SML).
• Sharpe vs. Treynor Relationship: The relationship between Sharpe and Treynor is governed by the portfolio's correlation with the market ($\rho_{pm}$), or coefficient of determination ($R^2 = \rho_{pm}^2$): $S_p = T_p \cdot \left( \frac{\beta_p}{\sigma_p} \right) = T_p \cdot \left( \frac{\rho_{pm}}{\sigma_m} \right)$
  • If a portfolio is perfectly diversified ($\rho_{pm} = 1.0$), Sharpe and Treynor rankings are identical.
  • If a portfolio is poorly diversified ($\rho_{pm} \ll 1.0$), it will score relatively well on Treynor (which ignores unique risk) but poorly on Sharpe (which penalizes uncompensated unique risk).

---

Jensen's Alpha ($\alpha_p$)

Developed by Michael Jensen in 1968, Jensen's Alpha measures the absolute excess return generated by a portfolio above the return predicted by the Capital Asset Pricing Model (CAPM), given the portfolio's systematic risk:

$\alpha_p = R_p - E(R_p)_{\text{CAPM}} = R_p - \left[ R_f + \beta_p (R_m - R_f) \right]$

   Realized Return R
     ▲                                               Security Market Line (SML)
     │                                              /
     │                                            .● Portfolio P (Realized Rp)
     │                                          .' │ ▲
     │                                        .'   │ │ +αp (Jensen's Alpha: Manager Skill)
     │                                      .●─────┼─▼───────────────────────────
     │                                    .' │     │ CAPM Expected Return E(Rp)
     │                     Benchmark (M) .●  │     │
     │                                 .' │  │     │
     │  Risk-Free (Rf) ───────────────●   │  │     │
     └────────────────────────────────────┴──┴─────┴─────────────► Systematic Risk (β)
                                         1.0      βp

Time-Series Regression Estimation:

In institutional practice, Jensen's Alpha is estimated via ordinary least squares (OLS) regression using the Security Characteristic Line (SCL):

$R_{p,t} - R_{f,t} = \alpha_p + \beta_p (R_{m,t} - R_{f,t}) + \epsilon_{p,t}$

Where:

• $\alpha_p$ = Regression intercept (Jensen's Alpha)
• $\beta_p$ = Regression slope coefficient (Systematic Beta)
• $\epsilon_{p,t}$ = Residual error term (idiosyncratic return variation)

Interpretation & Statistical Significance:

• $\alpha_p > 0$: The manager generated superior risk-adjusted returns, outperforming the SML (value added through security selection or market timing).
• $\alpha_p = 0$: Performance is exactly fair compensation for the systematic risk assumed.
• $\alpha_p < 0$: The manager underperformed expectations, failing to generate sufficient return to cover fees and systematic exposure.

[!IMPORTANT] An institutional consultant must never rely on point estimates of $\alpha_p$ alone. The consultant must evaluate the regression $t$-statistic ($t_\alpha = \hat{\alpha} / \text{SE}(\hat{\alpha})$) and associated $p$-value. A positive alpha of $+2.5%$ with a $t$-statistic of $1.15$ ($p > 0.05$) is statistically indistinguishable from zero (luck), whereas an alpha of $+1.8%$ with $t = 2.45$ ($p < 0.05$) demonstrates statistically significant active skill.

---