Tuesday, May 10, 2011

Retail Commodity Investors - A Bleeding Indicator?

Retail Commodity Investors – A bleeding indicator?

With the recent “flash crash” in commodities it seemed like a good time to delve into the question of how retail investors react to commodity prices and their investment flows into and out of commodities.  Some of the single commodity investment pools (a.k.a. ETFs) publish their daily issuances and redemptions data.  This allows a perspective into how retail investors are reacting to commodity prices on a daily basis.

Taking UNG as an example, we can observe daily on their website the following information, for example today on 5/10/2011 the report shows:

As of 05/09/2011, subject to change
Pending Trades Shares Created/Redeemed
Units Outstanding 

If we gather this data over time we can create a time series of redemptions and issuances which can be lined up with prices.  Hypothetically treating the entire pool of issuance as a single retail investor, this investor is at the margin buying and selling shares each day in reaction to prices.  The question is: does this hypothetical investor do better or worse than holding their position?  Also is there informational value in the funds flow?

The “as of” date for the pricing of daily holdings suggests that today’s report shows transactions that were done on the prior day 5/9/11.  So the incremental purchase of $2.7m shares was done at yesterday’s closing price for the fund, effectively the last trade price of $10.77.  This would have cost the investor a notional value of $29.079m.

When we track these incremental in/out flows cumulatively over time we can see that the hypothetical incremental investor has had pretty bad timing over most of the last year or so.  The incremental flows cost the investor about $1.4b over 2010, spread pretty steadily over the year.  In other words, investors consistently bought high and sold low in their redemptions and issuances of this ETF.  Intriguingly, this investor, (whether lucky or smart,) has recovered roughly half of this amount over 2011. 

Further when overlaid with the fund’s NAV, we can see that investors have made money exchanging shares when the NAV was falling  in early 2010, lost money while it was relatively stable throughout 2010, and then made money as it has fallen in 2011.

(Note that all these results have been adjusted for the reverse split that took place earlier this year.)

This also helps answer the second question of whether the retail investor flows have informational content regarding the market.  The annualized Sharpe ratio of the investor’s activity above is -0.9.  Not great as an investment strategy, but maybe there is informational content to the investor flow information?

Turns out that if you could know investor flows today and act on it there is a 0.8 Sharpe strategy in a 1 day hold long / short position based on investor flows.  Not a great stand alone strategy but useful as an overlay to a commodity strategy.

Monday, May 2, 2011

Commodities vs. commodity stocks

Just got back from a couple of months vacation in NZ so brief posting this month.

Another common field of mean reversion analysis relates to commodities and commodity related stocks.  In this analysis, I compare the prompt month crude oil price, CL1 to the price of XOP, an ETF that invests in oil & gas exploration & production companies.

As expected, these two prices track each other fairly closely over time:
 A similar analysis as before, performing an ADF test suggests that over its history the ratio of price series has exhibited mean reversion characteristics with a half life of 37 trading days or approximately two months.  This isn’t ideal for trading; it would be preferable if the relationship exerted itself more strongly.  Nevertheless with an ADF test statistic of 4% the evidence seems strong.
This relationship can be generalized to other commodities and commodity equity groupings, providing a guide to over/under valuation of commodity equities in relationship to their corresponding commodity prices.