Various ETF Performances From The Past 2 Years on S&P 500 Down Days

Jan 19, 2011

As we enter a period of higher volatility after a long sustained move up in equities, we ran some statistics on 25 of the largest ETFs in the world to see how they performed on a relative basis when the S&P drops X%.  

In this case, we chose to use 10 S&P pts, which works out to about 0.75%.  How did various ETFs perform on just those particular down days?  

Since the March 2009 low, the S&P 500 has dropped in excess of -0.75% eighty-five times.  The average S&P 500 loss for these 85 days was -1.68%.   Here are some results for 25 ETFs which summarize their performance on just those 85 days:


You can see that among the worst for this period were REITs (VNQ), U.S. financials (XLF),  Brazil (EWZ) and U.S. Small Cap Stocks (IWM).    It makes sense, these are all higher volatility market segments. Technology stands out for doing a bit better than you might have expected.


Bond indexes have low standard deviation and low correlation to U.S. stocks so they generally rose --- but as you can see, very modestly and not nearly enough to offset much of the loss in equities.


Gold, Preferred Stocks and High-Yield Bonds all show losses on average -- but more modest losses given their lower correlations with stocks.   Dividend ETFs (DVY, SDY) were down on all 85 days (85 for 85) -- which isn't a big surprise.   The dividend indexes didn't do particularly well on a relative basis however and lost almost as much as the S&P 500.

Now let's look at todays (Jan 19th) drop in various markets:

At the bottom of the list of todays (1-Day) performance are the exact same ETFs as the first list. 

A couple of ideas here:

1) If you think you are going to get any diversification benefit from owning REITs in a down S&P market, we would tend to doubt that.

2) Bonds don't provide much absolute protection in S&P 500 down days --- but they do obviously serve their purpose of stabilizing a portfolio.

3) Dividend stocks may offer only modest protection in down markets. 

Note that the China ETF (FXI) went up slightly today -- so there is the outlier of the day (of course no conclusion to draw here off 1 day of data).

Summary: it helps to study volatility and down markets. Todays performance was quite consistent with the same relative stats of the last two years.


Update: The first screenshot is now available in application form -- because it runs by auto-loading your user-created portfolios, it needs a login to access the page: ETFreplay Tools Page

Follow us on Twitter:


Comments (6) -

Jan 20, 2011 05:16 #

Chris:  What tool did you use to develop the first screen showing on this blog.

Joe United States

Jan 20, 2011 10:36 #


The tool is available on lower right of the ETFreplay tools page.  It will auto-load the symbols of any portfolio you have created using the drop-menu at the top.   The page then calculates everything dynamically when you submit a request using the 'Get Stats' button.  Use any timeframe you like but be aware that it will have to start calculations on the most recent starting date of any ETF in your portfolio.    So for longer lookbacks -- you should exclude newer ETFs in a portfolio list as it obviously can't run statistics on dates that doesn't exist.

Chris United States

Jan 23, 2011 06:56 #

Hi new member.  We know that because of financial crisis and sovereign, corp, individual debt levels that various asset classes remain painfully correlated as well as individual stocks to the overall market.  And probably more so than anytime in history.  With the asset allocation/moving avg plan I put in our groups pension plan all money moved in and out at the same time.  (emm reits spx commods, efa) Looking forward to utilizing the site for own personal investing.

GRock United States

Jan 23, 2011 10:17 #

You brought up correlation so I thought I would comment on this.

There are a few allocation funds out there.   If you pull up AOR (S&P Balanced Index  ETF), it has a very high correlation (~0.93)to the major stock indexes -- despite being an allocation index (it owns about ~40% bonds)  

Something like Brazil (EWZ) has ~0.80 correlation to the major stock indexes.  So an allocation index actually has a HIGHER correlation than the Brazil ETF.   But while the correlation of Brazil is lower, EWZ does go down a LOT more than the market in corrections.   So in cases like this, correlation per se wasn't indicative of the problem at all.    The problem was not the fact that EWZ went down (both AOR and EWZ dropped) -- its that EWZ went down a LOT.

What you really want to know is not necessarily the correlation but you want methods to help you understand the extent of the risk your portfolio represents and weigh that overall 'exposure' against your return expectations.    

We wanted to create a useful way to do this and not just add yet another correlation coefficient matrix.  

chris United States

Feb 08, 2011 11:11 #

Long duration treasuries, such as TLT, seem to provide an excellent hedge against down days, with the added benefit that you get paid to wait for them. However, as you did not use it in your sampling, I'm wondering if perhaps I'm missing something.

Drew Canada

Feb 08, 2011 12:00 #

We used a sampling of 25 of the largest ETFs in the world --- TLT is not among those though it is in the top 100.   You are right that TLT is definitely one of the premiere 'flight to safety' ETFs when there is real trouble.

The goal of course is to measure return expectations against drawdowns/risk and make a judgment.   Shorter-term bonds are very low risk and very low return.   Given where U.S. treasury yields have been,  TLT combines kind of worst of both worlds ---   above average volatility/risk with very low yield.

Chris United States

Comments are closed