Hedge Fund Investing Chapter 9: Measuring Performance

You would think measuring how well a hedge fund did is simple. Fund went up 10%? Great. Down 3%? Bad. Done.

Nope. Chapter 9 of Mirabile’s book shows that performance measurement is a whole thing. There is no single magic number. You need a combination of measures, and you need to ask whether the data is even reliable.

The Madoff lesson is right there in the opening. Investors saw consistent returns and assumed it was skill. Nobody asked why returns were so steady when the strategy should have produced variable results. There was no real edge. Just fraud.

The Basic Numbers

Let’s start with what most people look at first.

Absolute return is the raw percentage change. You put in $100, now you have $104, that is a 4% absolute return. Simple enough.

Relative return compares your fund to a benchmark. Your fund made 7%? Sounds good until you learn the S&P 500 did 15%. You actually underperformed by 8%. Context matters.

When you have a series of monthly returns, you can calculate either arithmetic or geometric averages. The geometric return is usually lower and more accurate. It accounts for compounding. Use that one.

Standard deviation tells you how much the returns bounce around. If a fund reports the exact same return every month, the standard deviation is zero. No volatility, no risk. It also means something is probably wrong. Every real fund has some variation.

Alpha and Beta: Where the Value Is

This is the core question for any hedge fund investor: am I paying for skill or just for market exposure?

Beta measures how much of your fund’s return comes from the market moving. A beta of 1.0 means your fund just tracks the benchmark. You can get that from a cheap index fund. No need to pay 2-and-20 for it.

Alpha is the good stuff. It is the return that comes from the manager’s actual skill, the part that is not explained by market movements. A positive alpha means the manager is adding value. Negative alpha means they are subtracting it.

The book makes a sharp point here. Some long/short managers run what are basically closet index funds. They pick stocks that track the broad market, borrow money to boost returns, and call it alpha. During calm bull markets it works. When the market turns, investors get crushed. High beta plus low alpha means you are overpaying.

Real hedge funds should produce alpha from both the long and short side. Leverage and short selling generally increase alpha and lower beta. If a fund barely does any short selling, be suspicious.

The Sharpe Ratio and Its Cousin

The Sharpe ratio answers: how much extra return am I getting per unit of risk? Take the fund’s return, subtract the risk-free rate (T-bills), divide by standard deviation. Higher is better. Positive means you are being compensated for risk. Negative means you are not.

The Sortino ratio is smarter. Instead of total volatility, it only counts downside volatility. Nobody complains about upside swings. You care about how bad the bad months are. A high Sortino ratio means low risk of large losses.

Jensen’s alpha measures how much the fund beat what the CAPM model predicted. If the model says you should earn 9% given your beta and market conditions, and you actually earned 11%, your Jensen’s alpha is 2%.

Why Normal Distributions Matter

Here is where it gets nerdy but important.

Most financial models assume returns follow a normal distribution. A nice bell curve. If that is true, you can make reasonable predictions about future performance based on past data. If it is not true, your predictions could be way off.

Two things to check:

Skewness measures whether returns are lopsided. Negative skew means you get steady decent returns most months, then one terrible month wipes out a chunk. Positive skew is the opposite: mostly mediocre months, then one great month saves the year. Both funds might have the same average return, but the experience of being invested in them is completely different.

Kurtosis measures how fat the tails are. Normal kurtosis is around 3. Higher than that means more extreme events than a normal distribution would predict. More blowups. More moonshots. More surprises.

The book mentions several tests you can run: Chi Square for goodness of fit, Kolmogorov-Smirnov (the most common in statistical packages), and Shapiro-Francia for small samples. The point is: do not just assume normality. Test for it.

Stress Testing the Portfolio

Historical returns only tell you what happened before. What about what could happen next?

Smart investors ask managers to stress test their current portfolio. What would happen if interest rates spiked 100 basis points? What if the S&P dropped 20%? What if credit spreads blew out like they did in 2008?

For bond-heavy funds, you want to know the duration (how sensitive the portfolio is to rate changes) and convexity (the non-linear part of that relationship). For equity funds, you want the current beta to the market.

Scenario analysis takes real historical events like October 1987, September 11, or the 2008 crisis, and applies those price movements to today’s portfolio. It gives you a forward-looking view that does not depend on assuming past returns will repeat.

Data Biases You Must Watch For

The data you are looking at might be lying to you. Not on purpose, but structurally.

Survivorship bias is the big one. Hedge fund indices remove funds that close or blow up. Only survivors stay. This makes the strategy look better than it was. That index you are comparing your fund to? Probably overstating average performance.

Valuation and reporting bias show up because many funds self-report returns. Even independent reports can have pricing errors for illiquid securities. The numbers might look smoother than reality.

Does Past Performance Actually Persist?

This is the million-dollar question. And the research is mixed.

Several studies found that performance does persist, but mostly over short time horizons. Agarwal and Naik (2000) found quarterly persistence. Edwards and Caglayan (2001) found it at one- and two-year horizons. Bares and colleagues (2002) saw it over one to three years. Eling (2009) did a meta-analysis and confirmed short-term persistence but not long-term.

There is a catch. Boyson (2003) found persistence was strongest among underperformers. Bad funds keep being bad. Jagannathan and colleagues (2010) found persistence among superior managers but not inferior ones. So alpha generators tend to keep generating alpha, while mediocre managers stay mediocre.

The bottom line from the book: most managers do not stay on top for very long. Today’s stars fade fast. It is not unusual for a portfolio to turn over 20% of its managers in a given year just to maintain performance. Picking a fund based only on past returns is asking for disappointment.

What to Take Away

No single number tells you if a hedge fund is good. You need the full picture: returns, risk-adjusted ratios, alpha versus beta, skewness, kurtosis, stress tests, and an honest look at data biases.

And even after all that, performance persistence is mostly short-term. The fund that crushed it last year might be average next year. Active monitoring and willingness to replace managers is not optional. It is the job.

Previous: Chapter 8 | Next: Chapter 10

This is part of a series retelling of “Hedge Fund Investing” by Kevin R. Mirabile.