In this post, we reconsider whether the Sharpe Ratio should be considered the bedrock of manager performance evaluation.
The very question reminds me of a management technique pioneered by Sakichi Toyoda, the founder of Toyota Industries, and a legend of the Japanese industrial revolution. Toyoda utilized 5 Whys as a means of isolating the source of confusion, or a problem within a complex process. Funny enough, the process works the same way a child may incessantly ask her parent “Why”, until the conversation inevitably veers into an existential dilemma1. The point of the method is to pinpoint the root of a problem in an actionable fashion. Cue our venerable Sharpe ratio:
Child: Why do we use Monthly Sharpe Ratio to evaluate manager skill?
Investor: It allows us to have a normalized value of comparison that denotes risk-adjusted return, the foundation of making all investment decisions around the world.
Child: Why Monthly Sharpe Ratio?
Investor: Uhh, beyond what I told you, largely because…we don’t tend to get or readily digest any better data?
Child: Why don’t you get any better data?
Investor: Umm…because we don’t want to be a burden on our managers?
Child: Why is anything more granular than monthly returns a burden on your managers?
Investor: Uhh…because they don’t…know their daily PnL…or they don’t want to share it with us?
Child: Why wouldn’t they want to share it with you?
Investor: Let me forward you to our customer service line.
Focusing specifically on performance2, the deep roots anchoring Sharpe as the golden standard of alternatives may surprise you. Hedge funds hang their hat on their ability to generate superior Sharpe ratios compared to passive investments. It is their raison d’etre. However, there has been speculation that hedge funds manage returns to smoothen volatility or to “shape their Sharpe”. There is also the oft forgotten issue of incentive allocation equalization which inflates Sharpe versus non-incentive geared investments. Finally, the very notion of Sharpe being misleading as a result of return distribution changes (or hidden, highly asymmetric payoffs such as derivatives in the portfolio) muddles the validity of a statistically insignificant sample set. Ask any convert arb investor in 2008.
I think all of these topics have been well addressed in the public forum. What hasn’t been effectively fleshed out in academia/punditry is the difference between daily Sharpe and monthly Sharpe. We assume (or don’t care) that they are the same in terms of direction, amplitude, distribution, modality, skew or kurtosis. However, there has been very little research done on the topic in the alternatives space, largely because hedge funds by-and-large do not grant this information to their LP (send in the inquisitive child!).
Cue the new era of liquid alternatives. With institutional caliber hedge funds launching long-only, 40Act and UCITS funds, the need for daily performance prints is causing funds to change the way they calculate, scrub, reposit, and print data. Once investors begin analyzing them, they may be disturbed with what they find out.
Just as an example, here are the rolling (90 day / 3 month) gross Sharpe ratios of two sample portfolios we use for research and blogging purposes at Novus. While we’d love to write about our client portfolios, their confidentiality trumps intellectual curiosity. The first is a concentrated long-only fund with ~20 positions and minimal turnover (quarterly rebalance) over an 8 year period:
Without going through the labor of (not so) fancy regressions and variance analyses, we can see the monthly Sharpe is smoothed and leads to a measurably different mean value (.71 for daily vs .84 for monthly) over the 8 year period. Specifically, there are periods where the trend is appreciably different (e.g., 3Q 2012!).
The second example is a few years of hedge fund dummy data; albeit a “fundamental” fund that doesn’t whip its book around much intra-month:
While a good chunk of the output is comparable to our Long-Only fund, the heavily volatile period in October/November 2014 (which we’ve written quite a bit about) yields dramatically different results in inter-month vs monthly Sharpe.
In short, during periods of market duress, or periods where an active manager is potentially taking liberties with moving her book, monthly and daily Sharpes can vary dramatically. If an investor predominantly relies on Sharpe as a tool for gauging risk-adjusted returns (which we wouldn’t suggest), one would hope they would at the very least ask for more granular daily data to better unpack its limitations.
1 – This joke has been brilliantly delivered by Louis CK.
2 – I’ll spare you in this piece from the Novus gospel about why skill sets and risk exposures are far more important to analyze than returns.
To contact the author of this story:
Faryan Amir-Ghassemi | firstname.lastname@example.org