A manager who beat the market may simply have taken more risk. This lesson shows how a planner judges managers with the Sharpe, Treynor, Jensen, information and Sortino measures, and which one fits the role a fund plays in the client's portfolio.
Sharpe and Treynor ratios
Both divide the return above the risk-free rate (the excess return) by a measure of risk:
Sharpe ratio = (Rp - Rf) / SDp Treynor ratio = (Rp - Rf) / Bp
Rp is the portfolio's return, Rf the risk-free rate, SDp its standard deviation and Bp its beta. Sharpe charges for total risk, so it fits a fund that is all or most of what the client holds. Treynor charges only for market risk, so it fits a fund that is one of many in a well-diversified portfolio, where the fund's own diversifiable risk washes out. Neither number means much alone; compare it with the benchmark's figure over the same period. Compare managers net of fees and over several years, since one year of any ratio is mostly luck.
Jensen's alpha
Jensen's alpha = Rp - [Rf + Bp x (Rm - Rf)]
Rm is the market's return. The bracket is the return the capital asset pricing model predicts for the fund's beta, so alpha is the return above or below that prediction. A positive alpha means the manager added value for the market risk taken. Like Treynor, alpha relies on beta, so it is only as good as the fund's fit to the market index (its R-squared).
Information ratio
Information ratio = (Rp - Rb) / tracking error
Rb is the benchmark's return, and tracking error is the standard deviation of the differences between the fund's and the benchmark's returns. It asks how consistently an active manager beats the chosen index for each unit of departure from it. It suits a planner who hires a manager to run one asset class against a stated benchmark.
Sortino ratio
Sortino ratio = (Rp - MAR) / downside deviation
MAR is the client's minimum acceptable return, and downside deviation measures only returns below the MAR. Clients usually don't mind upside surprises, so for a retiree drawing income, or for strategies with lopsided returns, the Sortino ratio separates good volatility from bad. Two funds with the same standard deviation can have very different downside deviations if one fund's large moves are mostly gains.
Worked comparison
The risk-free rate is 3%, the market and benchmark returned 9%, and the market's beta is 1.
| Measure | Manager A: 10.5%, SD 15%, beta 1.1, tracking error 3% | Manager B: 9.5%, SD 11%, beta 0.75, tracking error 2% |
|---|---|---|
| Sharpe | 7.5 / 15 = 0.50 | 6.5 / 11 = 0.59 |
| Treynor | 7.5 / 1.1 = 6.82 | 6.5 / 0.75 = 8.67 |
| Jensen's alpha | 10.5 - (3 + 1.1 x 6) = 0.90% | 9.5 - (3 + 0.75 x 6) = 2.00% |
| Information ratio | 1.5 / 3 = 0.50 | 0.5 / 2 = 0.25 |
Manager B wins on three measures, but Manager A beat the benchmark more consistently per unit of tracking error. Which matters depends on the job the client is hiring the manager to do.
| Client situation | Measure that fits |
|---|---|
| The fund is the client's whole portfolio | Sharpe |
| The fund is one sleeve of a diversified portfolio | Treynor or Jensen's alpha |
| The manager is judged against a set benchmark | Information ratio |
| The client fears falling below a target return | Sortino |
On the exam
- Subtract the risk-free rate (or MAR, or benchmark) before dividing.
- Treynor and alpha need a meaningful beta; with a low R-squared, use Sharpe.
- Information ratio divides by tracking error, not by the fund's standard deviation.
- Sortino divides by downside deviation, measured below the client's MAR.