Monday, January 5, 2009

ETF Risk in Review

As I sat down to write this article relying on last Friday's closing ETF Rewind (still in beta), at first glance I have to admit that the annual Sharpe Ratios made no sense to me whatsoever. How could Financials (XLF) be in the middle of the pack! Surely I had checked and rechecked the math!

Then came the "aha" moment (sometimes we need to relearn old lessons, no?). Traditional risk-reward measurements may produce absolutely nonsensical results in broad negative return environments -- but, I'm getting ahead of myself.

By now you have undoubtedly read the thousand-and-one blogs on 2008 relative ETF return rankings, but how did they fair on a "risk-reward basis"? This is the focus of the table below, as follows:


If one thinks of each ETF as a miniature strategic portfolio in and of itself, the savvy investor may consider not only the raw returns of that portfolio, but also how much return it provided for each unit of risk. The statistical out-takes above cover a number of risk and risk-reward measures, including: Beta, Historic Volatility, Sharpe, Sortino and Omega ratios, as further described here (the nightly ETF Rewind file covers nearly 60 key statistics).

Getting back to my opening comments, let's focus on the Sharpe Ratio, which is defined as Excess Returns (subject asset returns less risk-free returns)/ Risk (the standard deviation of the subject asset returns). In a positive return environment, a security featuring relatively higher returns and lower risk receives a higher ranking. In the "biz", results approaching or exceeding +2.0 are considered especially good.

Now let's imagine that ratio in a negative return environment... more volatile securities are rewarded rather than punished, receiving relatively higher/closer to zero rankings (think, for example, -10%/1 risk unit versus -10%/2 risk units)! The simple solution to this problem? Modify the traditional calculation by raising the volatility denominator for all assets with negative returns to the power of -1. Last year that just about covered the waterfront -- unlike prior recessionary bear-market periods, there was nowhere to hide in 2008 (alright, except cash)!

The table above is sorted by Sortino Ratio, which is similar in construction to the Sharpe, but only penalizes returns for downside volatility, as similarly modified to accommodate last year's net negative return environment. Do any of the results surprise you? Last year's deflationary recession trade is well described and all is right in the world again!

The Momentum Persistence effect remains largely unexplained by Efficient Market Hypothesis proponents. The 2007 year-end risk-reward review provided key forward looking clues towards both the intermediate and ultimate outcomes for 2008. Take a close look at last year's table, recalling that early on stagflation remained a primary concern even as the full scope of the financial industry woes was yet to emerge.

You'd better believe that I'll be tracking this year's list and its constituent rank evolution over this and shorter time frames very closely.

5 comments:

SnoopyJC said...

How would something like ETJ fare, which is risk-managed, but suffers by being a closed-end fund (and can trade at a premium or discount to it's NAV)?
--joe

Jeff Pietsch CFA Esq said...

Hey Snoopy, I'll run it and post the results. Best, Jeff

Jeff Pietsch CFA Esq said...

Hi again Snoopy - ETJ gained almost 6% during the subject 250-day period, which is in its favor right off the bat. Shall I guess you knew that! Within the sample, it would have been the only one with a postive Sharpe. It's Sortino also would have been top of the heap. Neat little fund.

http://finance.yahoo.com/q?s=ETJ

guillaume robinet said...

Hello,


If an ETF had a negative return during a year like 2008, his Sharpe Ratio will then be negative as well, correct? how to interpret it?

Guillaume

jgpietsch said...

Correct because the return in the numerator is negative. Perhaps you mean you would like me to elaborate on this section?:

Now let's imagine that ratio in a negative return environment... more volatile securities are rewarded rather than punished, receiving relatively higher/closer to zero rankings (think, for example, -10%/1 risk unit versus -10%/2 risk units)! The simple solution to this problem? Modify the traditional calculation by raising the volatility denominator for all assets with negative returns to the power of -1. Last year that just about covered the waterfront -- unlike prior recessionary bear-market periods, there was nowhere to hide in 2008 (alright, except cash)!