Here's a statement of the obvious: The opinions expressed here are those of the participants, not those of the Mutual Fund Observer. We cannot vouch for the accuracy or appropriateness of any of it, though we do encourage civility and good humor.
From the Article: "...real estate was the best overall performer among the 12 individual asset classes, with its combination of 5.2% standard deviation and 11.56% average three-year annualized return.
Commodities has an undeserved reputation as a very risky asset class. So it is intriguing to note that the standard deviation for the PowerShares DB Commodity Tracking ETF (DBC) was lower than for U.S. large stocks, while it had a dramatically higher average three-year return of 12.71% — the highest of the 12 asset classes being considered."
"Commodities has an undeserved reputation as a very risky asset class. So it is intriguing to note that the standard deviation for the PowerShares DB Commodity Tracking ETF (DBC) was lower than for U.S. large stocks, while it had a dramatically higher average three-year return of 12.71% — the highest of the 12 asset classes being considered. (Note that when an ETF in this analysis did not have a 15-year performance history, I used the performance of the underlying index, minus an implied expense ratio.)"
I was looking at that too, trying to find it. Have no idea. Where did he get the idea that this asset class was tops out of all the other asset classes? Looks like it was a very poor performer.
I can only use ARYVX as an example but last year was very good until the last couple of months where it gave it all back. This year also had been good so far. Hopefully it's not a repeat.
I think his 12.71%, which I didn't try to recalculate, is achieved by taking annualized return for 1999-2001, then 2000-2002 all the way to 2011-2013 and averaging the 13 results. Of course it says nothing about the volatility one had to live with during each of those 3 year periods.
Edit: Sorry, I guess the standard deviation does tell you something about volatility but is he measuring the standard deviation of the asset class over the whole time or just the standard deviation of the returns over the 5 periods he considered?
He also mentions: "...an ETF in this analysis did not have a 15-year performance history, I used the performance of the underlying index, minus an implied expense ratio."
Also, I think it's more challenging to find choices of investments (either mutual funds or ETFs) in Commodities for individual investors to invest in.
Overall, I very much like the analyses and interpretations produced by Professor Craig Israelsen. I recently referenced his website on this Board. He does a splendid job with his portfolio diversification policy.
On a rare occasion, Israelsen wanders off the conventional investment wisdom reservation; sometimes he generates an enigmatic analysis that tends to baffle rather than inform us amateur investors. In his referenced article and Table, his volatility calculation is one such instance; it is a strange Standard Deviation definition.
It is so strange that I decided to do a few check calculations to insure that I was properly interpreting what the base input data really were. I did the verification calculations for the Large US Cap and the Commodities categories.
Israelsen’s quoted Standard Deviation is indeed based on the average annual returns from the 5 period returns that he listed earlier in the Table. Why?
I’m not sure I understand the meaningfulness of this volatility measure. Since it’s the deviation from these 5 data periods, the Mean of that 5 data point input must also be calculated. I’m equally puzzled about the significance of that average.
Most importantly, by constructing a Mean and a Standard Deviation in this manner, Israelsen is weighting the annual returns data by including the most recent 1-year annual return data 5 times in the calculation, but only using the 15-year old annual return data once in the analysis.
I don’t basically disagree with some weighting function that emphasizes the most recent outcomes (exponential decay is frequently used), but the Israelsen method seems a bit ad hoc. It seems arbitrary. Buyer beware.
I’m sure Professor Israelsen can defend his approach. I’m just not familiar with the logic or the empirical data that support his formulation. That’s my shortcoming. I really would appreciate an explanation.
Comments
"Commodities has an undeserved reputation as a very risky asset class. So it is intriguing to note that the standard deviation for the PowerShares DB Commodity Tracking ETF (DBC) was lower than for U.S. large stocks, while it had a dramatically higher average three-year return of 12.71% — the highest of the 12 asset classes being considered. (Note that when an ETF in this analysis did not have a 15-year performance history, I used the performance of the underlying index, minus an implied expense ratio.)"
@bee and MFOers:
Have no idea what he's talking about. Can someone explain this to me?
Here's Morningstar's performance data for DBC:
Where did he get the idea that this asset class was tops out of all the other asset classes?
Looks like it was a very poor performer.
Edit: Sorry, I guess the standard deviation does tell you something about volatility but is he measuring the standard deviation of the asset class over the whole time or just the standard deviation of the returns over the 5 periods he considered?
"...an ETF in this analysis did not have a 15-year performance history, I used the performance of the underlying index, minus an implied expense ratio."
Also, I think it's more challenging to find choices of investments (either mutual funds or ETFs) in Commodities for individual investors to invest in.
Overall, I very much like the analyses and interpretations produced by Professor Craig Israelsen. I recently referenced his website on this Board. He does a splendid job with his portfolio diversification policy.
On a rare occasion, Israelsen wanders off the conventional investment wisdom reservation; sometimes he generates an enigmatic analysis that tends to baffle rather than inform us amateur investors. In his referenced article and Table, his volatility calculation is one such instance; it is a strange Standard Deviation definition.
It is so strange that I decided to do a few check calculations to insure that I was properly interpreting what the base input data really were. I did the verification calculations for the Large US Cap and the Commodities categories.
Israelsen’s quoted Standard Deviation is indeed based on the average annual returns from the 5 period returns that he listed earlier in the Table. Why?
I’m not sure I understand the meaningfulness of this volatility measure. Since it’s the deviation from these 5 data periods, the Mean of that 5 data point input must also be calculated. I’m equally puzzled about the significance of that average.
Most importantly, by constructing a Mean and a Standard Deviation in this manner, Israelsen is weighting the annual returns data by including the most recent 1-year annual return data 5 times in the calculation, but only using the 15-year old annual return data once in the analysis.
I don’t basically disagree with some weighting function that emphasizes the most recent outcomes (exponential decay is frequently used), but the Israelsen method seems a bit ad hoc. It seems arbitrary. Buyer beware.
I’m sure Professor Israelsen can defend his approach. I’m just not familiar with the logic or the empirical data that support his formulation. That’s my shortcoming. I really would appreciate an explanation.
Best Wishes.