Global Minimum Volatility Fund

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
Topic Author
trasmuss
Posts: 235
Joined: Sun Oct 26, 2008 7:10 am

Global Minimum Volatility Fund

Post by trasmuss »

I first noticed this fund when I saw that Vanguard uses it for 20% of their Managed Payout fund. It is a global fund that uses hedging (similar to the International Bond fund) for the international stocks. It is a fairly new fund so it is too early to come to any conclusion concerning its merits. However, in watching it there have been some interesting points:
o It does seem to be a lower volatility fund (especially for an equity fund).
o It offers an admiral version which the World Stock fund doesn't. For such a small fund it is inexpensive.
o It doesn't offer as much diversification as World Stock but that is to be expected if Vanguard is truly looking for lower volatility stocks.
o It uses smaller stocks than World Stock (and you may not recognize many of the names).
o So far in its short history performance does not seem to be suffering for its lower volatility.
o Considering that Vanguard uses it for 20% of the Managed Payout fund, they certainly don't seem to be promoting it (especially compared to their International bond fund).

I hold Total Stock and Total International and have no current plans to add this fund. I will continue to watch it, however. Has anyone else thoughts concerning the potential role of this fund in a portfolio?
grok87
Posts: 10512
Joined: Tue Feb 27, 2007 8:00 pm

Re: Global Minimum Volatility Fund

Post by grok87 »

trasmuss wrote:I first noticed this fund when I saw that Vanguard uses it for 20% of their Managed Payout fund. It is a global fund that uses hedging (similar to the International Bond fund) for the international stocks. It is a fairly new fund so it is too early to come to any conclusion concerning its merits. However, in watching it there have been some interesting points:
o It does seem to be a lower volatility fund (especially for an equity fund).
o It offers an admiral version which the World Stock fund doesn't. For such a small fund it is inexpensive.
o It doesn't offer as much diversification as World Stock but that is to be expected if Vanguard is truly looking for lower volatility stocks.
o It uses smaller stocks than World Stock (and you may not recognize many of the names).
o So far in its short history performance does not seem to be suffering for its lower volatility.
o Considering that Vanguard uses it for 20% of the Managed Payout fund, they certainly don't seem to be promoting it (especially compared to their International bond fund).

I hold Total Stock and Total International and have no current plans to add this fund. I will continue to watch it, however. Has anyone else thoughts concerning the potential role of this fund in a portfolio?
Hi trasmuss,
Agree with your comments on the fund. i would add that the fund is actively managed- i.e. not an index fund.
I do have a small position in the fund. i'd probably be more interested in increasing my position if it were an index fund.

here is a write-up by Larry Swedroe on minimum volatility
http://www.cbsnews.com/news/a-way-to-ge ... less-risk/
i believe Larry concludes that their is nothing extra special about low volatility- you just get exposure to the value premium and the term premium.
Larry Swedroe wrote: They simply trade one risk (market or beta) for others (value and term). The overall risks of a low-beta strategy can easily be replicated with traditional value-oriented stock and bond portfolios.
I don't necessarily disagree with Larry. Larry is very smart. However I would perhaps make just one additional point. A stock investing approach that can capture the (bond) term premium may have some value from a tax perspective.

For simplicity, let's start with a totally taxable investor. If he wants to be 60/40 stocks bonds that may create some tax-inefficiencies for high-tax-bracket investors. Normally bond income is taxed immediately at ordinary income rates, whereas stock capital gains are deferred and qualified dividends are taxed at lower rates. This investor could of course use muni bonds but that may introduce undesirable credit risk etc.

If the minimum volatility stock approach offers exposure to the bond term risk premium, and has lower volatility then instead of the regular 60/40 portfolio one could do say 70/30 using the min vol fund for the stocks. that would perhaps provide the same exposure to bonds but with lower tax drag.
just a thought...
cheers
RIP Mr. Bogle.
Topic Author
trasmuss
Posts: 235
Joined: Sun Oct 26, 2008 7:10 am

Re: Global Minimum Volatility Fund

Post by trasmuss »

Vanguard isn't advertising it as a fund to go with for performance. In fact they say it will likely lag in a bull market. But if it has lower volatility and perhaps doesn't drop as much in the nasty markets it may be a good addition to a portfolio. There must be a reason that Vanguard is including it with such a high percentage in their Managed Payout fund.

It bears watching. And it may be a way for people like me to increase foreign exposure without adding more of the Total International fund. If the fund continues to perform well we will likely hear more about it in the future. And it will be interesting to see if it is ever added to the Life Strategy or Retirement funds.
grok87
Posts: 10512
Joined: Tue Feb 27, 2007 8:00 pm

Re: Global Minimum Volatility Fund

Post by grok87 »

trasmuss wrote:Vanguard isn't advertising it as a fund to go with for performance. In fact they say it will likely lag in a bull market. But if it has lower volatility and perhaps doesn't drop as much in the nasty markets it may be a good addition to a portfolio. There must be a reason that Vanguard is including it with such a high percentage in their Managed Payout fund.
I think you are on to something here. We all know that one of the big risks of a safe withdrawal strategy is a big market drop shortly after one begins the strategy- i.e. shortly after one retires.

So this is the holy grail. A stock fund that has somewhat lower expected return in exchange for markedly lower downside or drawdown risk. I'm a little skeptical that Vanguard has FOUND the answer to this with their new global minimum volatility fund. But I'm willing to give it a shot- like I said I have a small amount in the fund right now.

For a while folks were pushing the idea that dividend stocks were the answer, the holy grail. But let's look at one of the dividend ETFs that has been around for at least 10 years, DVY. http://performance.morningstar.com/fund ... tion?t=DVY

In 2008 DVY lost 33% vs. 39.3% for the average mid-cap value fund (DVY is classified by M* as mid-cap value). So 16% less downside or 6.3% points less downside.
But in terms of 10 year average annual performance, DVY really lagged other mid-cap value funds. It returned 6.97% vs. 8.92% for the category. So it had only 78% of the returns or it lagged by 1.95% points.

So in my book dividend stocks are not really an acceptable tradeoff. You get 84% of the downwide (GOOD) but only 78% of the returns (BAD). I think this demonstrates the point i have often heard made. That a dividend stock strategy is simply another way to capture the small and value premiums but that it is not the optimal way- i.e. it performs less well than other small and value strategies.

So has Vanguard found the holy grail with this new Minimum volatility strategy? only time will tell.

cheers,
RIP Mr. Bogle.
User avatar
steve r
Posts: 1299
Joined: Mon Feb 13, 2012 7:34 pm
Location: Connecticut

Re: Global Minimum Volatility Fund

Post by steve r »

I own this fund for a large share of my portfolio.

Previously, I owned a similar index (ACWV). That said, I strongly encourage you to read the Wiki on the space.
http://www.bogleheads.org/wiki/Low_vola ... ex_returns If still interested, you may also want to read the book by Falkenstein (which is not a good read). Larry's article linked in this story is also good (though I disagree with his conclusion). The performance from 1968-2010 is amazing in my view. You may want to read the articles Larry cites.

STULTZ has written on this space. His ability to explain things in short posts amazes me. You may want to do a search.

You need to understand the difference between minimum volatility and low volatility. Low volatility has outperformed for the last decade as they load up on utilities. I prefer to maintain sector weights over cap weighted indexes. VMNVX (and ACWV) does this. You need to realize that this will under-perform during strong bull market. I also (personally) do not see the need to own everything as is done with a pure cap weighted approach (I do have global indexes with the rest of my portfolio however). So, technically less diversified, but not really as the value of diversification diminishes well before 165 holdings.

The main thing you will find is that not owning the most volatile stocks is probably a good idea, but there is no easy way to do this. Owning the least volatile stocks is "second best" to "not owning the most volatile" in my view. Not owning stocks like Tesla, Krispy Creme, Amazon, Apple, PETS.COM, AOL, Cisco, means you will miss some real winners (thus under perform at time) but also miss some real losers. I still own these winners and losers (I own the market), but make much smaller bets here by loading up on VMNVX. Tilting value, in my view, is not quite as good at doing this specific task (though I know some disagree).

In this space (ONLY) I prefer not owning the index if expenses are low. The problem with the index approach in this space is the twice a year re-balance methodology by rule. When a stock gets volatile - I want out. Moreover, the index approach churns a decent chunk of portfolio on two days a year. And caps the amount of churn on those days regardless of what is now the least volatile portfolio. I prefer baby steps. I have no evidence to support this way is better. That said, I was truly excited when Vanguard came out with VMNVX.

My read of Larry, who I have a lot a respect for and DFA who taught me about value investing ... also enormous respect for ... Yes, there is overlap - enormous amounts of overlap. Yes, statistical factor models shows overlap, and yes, on paper you may replicate results with a value tilt ... ... But I question the direction of the causation/correlation. Plus, there is more to the story and ACWV is more growth than value per M*. Moreover, according some other studies we are now looking at data as far back as the 1960s with solid outperformance (even the 1920s, but data is suspect). Hard to beat this. Lastly, I should point out on this site, which ironically helped me form my opinions, I am definitely in the minority opinion on this space.

I never thought of GROK comment on the benefit of a bond premium in equities, but I think his point is valid. I own this space in Roths, but would consider this fact if in taxable space.
"Owning the stock market over the long term is a winner's game. Attempting to beat the market is a loser's game. ..Don't look for the needle in the haystack. Just buy the haystack." Jack Bogle
Topic Author
trasmuss
Posts: 235
Joined: Sun Oct 26, 2008 7:10 am

Re: Global Minimum Volatility Fund

Post by trasmuss »

Thank you for your input, Steve. I will read up on the sources you mentioned. If I end up investing in this fund it would be in an IRA. Although the fund is small and actively managed it has low expenses. Quite surprising. It will be interesting a year or so from now if others "discover" this fund or if it fades into the horizon like Asset Allocation did (Vanguard also used this for a time).
grok87
Posts: 10512
Joined: Tue Feb 27, 2007 8:00 pm

Re: Global Minimum Volatility Fund

Post by grok87 »

steve r wrote:I own this fund for a large share of my portfolio....

I never thought of GROK comment on the benefit of a bond premium in equities, but I think his point is valid. I own this space in Roths, but would consider this fact if in taxable space.
thanks steve.

when you say a large share, what percentage are you talking? Specifically I guess how much of your equities. As trasmuss points out the vanguard managed payout fund has about 30% of its equities in this fund. From memory I think managed payout has 65% in equities which are split 25/20/20 between Total Stock Market /Total INternational Stock Market/Global Minimum Volatility.

so are you doing 30% of your equities in the global minimum volatility fund? i've got a small position in the fund now but was thinking about getting up to 15-20% of my equities eventually
RIP Mr. Bogle.
scubablue
Posts: 43
Joined: Fri Feb 15, 2013 10:47 am

Re: Global Minimum Volatility Fund

Post by scubablue »

I am also intrigued by this fund and have been thinking of adding a position. Since Vanguard includes the Minimum Volatility in their Managed Payout along with Total Stock and Total International funds, is it safe to say it complements, rather than duplicates, Total Stock and Total International? Thanks for any input!
Topic Author
trasmuss
Posts: 235
Joined: Sun Oct 26, 2008 7:10 am

Re: Global Minimum Volatility Fund

Post by trasmuss »

I would think that the Global Minimum Volatility fund does duplicate holdings in Total Stock and Total International since those two funds hold so many of the marketable securities (especially Total Stock). Perhaps there is a "screen" that measures the duplication that someone can comment on.
User avatar
LadyGeek
Site Admin
Posts: 95466
Joined: Sat Dec 20, 2008 4:34 pm
Location: Philadelphia
Contact:

Re: Global Minimum Volatility Fund

Post by LadyGeek »

Here's an older thread which has some good info: [Vanguard Announces New Low Volatility Fund]
Wiki To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Re: Global Minimum Volatility Fund

Post by grayfox »

Looking at Vanguard Total World Stock Index Fund Investor Shares (VTWSX) there are 6161 stocks.

Top ten holdings are:
1 Apple Inc.
2 Exxon Mobil Corp.
3 Google Inc.
4 Microsoft Corp.
5 Johnson & Johnson
6 Wells Fargo & Co.
7 General Electric Co.
8 Nestle SA
9 Royal Dutch Shell plc
10 JPMorgan Chase & Co.

All the usual large-cap suspects.

In contrast, Vanguard Global Minimum Volatility Fund Investor Shares (VMVFX) only holds 229 stocks.

Top ten holdings are:
1 Vector Group Ltd.
2 Chunghwa Telecom Co. Ltd.
3 BCE Inc.
4 Amdocs Ltd.
5 Capitol Federal Financial Inc.
6 CLP Holdings Ltd.
7 Next plc
8 Colony Financial Inc.
9 National Grid plc
10 Hudson Pacific Properties Inc.

Frankly, I never heard of any of those companies. A couple of financials, a Chinese telecom, and one sounds like a REIT.

So this is basically picking a subset of the total world stock market. But it is stock picking using Modern Portfolio Theory, one form of optimized mean-variance investing.

GMV is a kind of like mean-variance optimization. But GMV does not require mean return estimates which have huge standard errors. GMV only requires covariance estimates which are not nearly as bad. GMV basically tries to be on the tip of the Markowitz bullet without trying to predict what the mean return will be.

There is a paper that looked at many index strategies and GVM fared very well.
Another paper from 2012, Mean‐Variance Investing by Andrew Ang, discusses this subject quite thoroughly
grok87
Posts: 10512
Joined: Tue Feb 27, 2007 8:00 pm

Re: Global Minimum Volatility Fund

Post by grok87 »

grayfox wrote: GMV is a kind of like mean-variance optimization. But GMV does not require mean return estimates which have huge standard errors. GMV only requires covariance estimates which are not nearly as bad. GMV basically tries to be on the tip of the Markowitz bullet without trying to predict what the mean return will be.
i guess i've never really understood that argument. diversification matters the most in a crisis. in a crisis the correlations and covariances all seem to go haywire.
personally i've very suspicious of historical correlations. i tend to follow swensen. i think his key argument is to invest in things that:

1) have a positive expected real return
2) are intrinsically different enough from each other that you have a shot at diversification- u.s. stocks, intl, stocks, real estate, treasuries, tips

personally i think this fund is interesting and worth a small allocation. but i'm not a true believer yet...
RIP Mr. Bogle.
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Re: Global Minimum Volatility Fund

Post by grayfox »

grok87 wrote:
grayfox wrote: GMV is a kind of like mean-variance optimization. But GMV does not require mean return estimates which have huge standard errors. GMV only requires covariance estimates which are not nearly as bad. GMV basically tries to be on the tip of the Markowitz bullet without trying to predict what the mean return will be.
i guess i've never really understood that argument. diversification matters the most in a crisis. in a crisis the correlations and covariances all seem to go haywire.
personally i've very suspicious of historical correlations. i tend to follow swensen. i think his key argument is to invest in things that:

1) have a positive expected real return
2) are intrinsically different enough from each other that you have a shot at diversification- u.s. stocks, intl, stocks, real estate, treasuries, tips

personally i think this fund is interesting and worth a small allocation. but i'm not a true believer yet...
That is just the theory. See referenced paper.

1. Basically, if you think you can estimate means and covariances, you would use plain-old MVO.
2. If you think you can estimate covariances but not means, use GMV
3. If you think you can estimate variance, but not means or correlations, use Risk Parity
4. If you don't think you can estimate anything, use Equal weighted.

5. What about market weight portfolio? I think the argument for it is that it is simple and passive. But it has been about the worst performer of all the index strategies. See other paper.
grok87
Posts: 10512
Joined: Tue Feb 27, 2007 8:00 pm

Re: Global Minimum Volatility Fund

Post by grok87 »

grayfox wrote:
grok87 wrote:
grayfox wrote: GMV is a kind of like mean-variance optimization. But GMV does not require mean return estimates which have huge standard errors. GMV only requires covariance estimates which are not nearly as bad. GMV basically tries to be on the tip of the Markowitz bullet without trying to predict what the mean return will be.
i guess i've never really understood that argument. diversification matters the most in a crisis. in a crisis the correlations and covariances all seem to go haywire.
personally i've very suspicious of historical correlations. i tend to follow swensen. i think his key argument is to invest in things that:

1) have a positive expected real return
2) are intrinsically different enough from each other that you have a shot at diversification- u.s. stocks, intl, stocks, real estate, treasuries, tips

personally i think this fund is interesting and worth a small allocation. but i'm not a true believer yet...
That is just the theory. See referenced paper.

1. Basically, if you think you can estimate means and covariances, you would use plain-old MVO.
2. If you think you can estimate covariances but not means, use GMV
3. If you think you can estimate variance, but not means or correlations, use Risk Parity
4. If you don't think you can estimate anything, use Equal weighted.

5. What about market weight portfolio? I think the argument for it is that it is simple and passive. But it has been about the worst performer of all the index strategies. See other paper.
Thanks grayfox. so i guess this is the paper. i will read it- i promise! :)
http://papers.ssrn.com/sol3/papers.cfm? ... id=2131932

i like your framework. I guess i would add though that i think std. dev./variance is the wrong measure of risk. Academics like it because it makes the math tractable. So they can do lots of fancy (and mostly useless math). But in real life downside risk or drawdown is what matters. tails are fat and not modelled well by normal distributions. In terms of your 4 point framework:

1) I don't think i can estimate means and covariances. so no MVO for me.
2) obviously no GMV for me either
3) risk parity is a bunch of nonsense. basically says bonds are low risk so let's leverage them up. last summer showed the folly of that.
4) i'm partial to equal weighted but don't like the transaction costs
5) mostly i'm tending toward the market weight portfolio these days, at least for stocks. but with some small tilts towards small, value and minimum vol.

cheers,
RIP Mr. Bogle.
columbia
Posts: 3023
Joined: Tue Aug 27, 2013 5:30 am

Re: Global Minimum Volatility Fund

Post by columbia »

Did any of these MV funds exist in 2008?
grok87
Posts: 10512
Joined: Tue Feb 27, 2007 8:00 pm

Re: Global Minimum Volatility Fund

Post by grok87 »

columbia wrote:Did any of these MV funds exist in 2008?
good question. yeah that's exactly my concern that in a crisis none of this stuff will hold up
RIP Mr. Bogle.
User avatar
steve r
Posts: 1299
Joined: Mon Feb 13, 2012 7:34 pm
Location: Connecticut

Re: Global Minimum Volatility Fund

Post by steve r »

grok87 wrote:
steve r wrote: when you say a large share, what percentage are you talking? Specifically I guess how much of your equities. As trasmuss points out the vanguard managed payout fund has about 30% of its equities in this fund. From memory I think managed payout has 65% in equities which are split 25/20/20 between Total Stock Market /Total INternational Stock Market/Global Minimum Volatility.

so are you doing 30% of your equities in the global minimum volatility fund? i've got a small position in the fund now but was thinking about getting up to 15-20% of my equities eventually
I like what the managed payout does, though I am not in withdrawal stage. I am a huge believer in benefit of smoothing out returns for CAPM reasons.

I my view the efficiency frontier as very flat and downward sloping with the most volatile stocks. (curve with least volatile to most volatile). While I agree it is impossible to pick the left most point in a C shaped efficiency curve .... I also view it as impossible to find the peak. Some may quibble with the idea that the curve is downward sloping at some point. I want to be somewhere between the two (left most and peak) ... at this point the slope is not very high. So a little more than a third of equities is here. (all my Roth space). The rest is global equities. I have no desire to own market cap weight of Apple or Google or Microsoft.
grayfox wrote:Looking at Vanguard Total World Stock Index Fund Investor Shares (VTWSX) there are 6161 stocks.

Top ten holdings are:
1 Apple Inc.
2 Exxon Mobil Corp.
3 Google Inc.
4 Microsoft Corp.
In contrast, Vanguard Global Minimum Volatility Fund Investor Shares (VMVFX) only holds 229 stocks.

Top ten holdings are:
1 Vector Group Ltd.

Frankly, I never heard of any of those companies. A couple of financials, a Chinese telecom, and one sounds like a REIT.

So this is basically picking a subset of the total world stock market. But it is stock picking using Modern Portfolio Theory, one form of optimized mean-variance investing.
[/quote]

Good point. I never heard of them either. But none dominate the portfolio. This is exactly what I want. They are sector cap weighted (an important distinction with low vol)
grok87 wrote:
columbia wrote:Did any of these MV funds exist in 2008?
good question. yeah that's exactly my concern that in a crisis none of this stuff will hold up
The non-tradable indexes did fine those years.
"Owning the stock market over the long term is a winner's game. Attempting to beat the market is a loser's game. ..Don't look for the needle in the haystack. Just buy the haystack." Jack Bogle
RNJ
Posts: 863
Joined: Mon Apr 08, 2013 9:06 am

Re: Global Minimum Volatility Fund

Post by RNJ »

I can see where a fund like this might be a good diversifier in the context of a portfolio in theory, in that it is designed to act differently than other asset classes. The question in this regard would seem to be whether it is different enough when you need it to be.

But as a stand-alone holding, if one wants a lower beta portfolio, why not just hold less beta?

What am I missing / misunderstanding?

Thanks!
User avatar
steve r
Posts: 1299
Joined: Mon Feb 13, 2012 7:34 pm
Location: Connecticut

Re: Global Minimum Volatility Fund

Post by steve r »

RNJ wrote:I can see where a fund like this might be a good diversifier in the context of a portfolio in theory, in that it is designed to act differently than other asset classes. The question in this regard would seem to be whether it is different enough when you need it to be.

But as a stand-alone holding, if one wants a lower beta portfolio, why not just hold less beta?

What am I missing / misunderstanding?

Thanks!
Good question.


Conceptually, low vol minimizes volatility, while low beta does the same thing but relative to market returns.

On a practical side, Falkenstein's book uses low beta returns as evidence for min vol. For me personally, I learned of min vol first and think why not min vol (instead of low beta). This blog post claims low vol is not "beta" but your question is on "low beta." http://falkenblog.blogspot.com/2012/11/ ... menon.html You may find it of interest, though it may not answer your question.

Steve
"Owning the stock market over the long term is a winner's game. Attempting to beat the market is a loser's game. ..Don't look for the needle in the haystack. Just buy the haystack." Jack Bogle
stlutz
Posts: 5585
Joined: Fri Jan 02, 2009 12:08 am

Re: Global Minimum Volatility Fund

Post by stlutz »

I think we're overstating the distinction between a minimum variance portfolio and a low beta portfolio. Actually what makes the iShares ETFs distinctive is not so much they they are running minimum variance portfolios but that the deviate from what the math tells them the minimum variance portfolio should be.

If you do the regressions on your own, a minimum variance portfolio will mostly end up with a bunch of utility and consumer staples stocks, just like a low beta portfolio will. What MSCI/iShares then do is impose some overriding criteria that move away from what they expect the minimum variance portfolio to be. For example, they impose sector and turnover constraints. A low-beta fund could do the exact same thing; S&P/Powershares simply chose not to with SPLV.

In backtesting, such constraints actually result in more volatility and lower returns. So, the indexes that these funds are tracking are not based on the best possible results that could be obtained in backtesting. Instead, the backtests built in assumptions for how you would actually want to run a real portfolio that one could sensibly hold a large portion of their assets in.
I can see where a fund like this might be a good diversifier in the context of a portfolio in theory, in that it is designed to act differently than other asset classes. The question in this regard would seem to be whether it is different enough when you need it to be.
The iShares funds explicitly try to have an r-squared of 1 with the overall market. One should not buy these funds with the expectation that they will act differently than the rest of the market. Had they existed, such funds would have gone down in 2008; they simply would have gone down less.

As I've argued before, for the equity-heavy investor who wants to reduce their risk, adding more bonds is the best way to do it. On the other hand, if one already holds at least 50% bonds and wants to de-risk further, these funds are worth considering as a way to accomplish that goal without losing asset-class diversification.

As with all of the "smart beta" strategies out there currently (value, small cap, quality etc.), valuations on these funds are not attractive right now. As such, I think one can invest and expect to get less volatility; I don't think the returns of a low. vol. strategy will be as attractive in coming years as in the past because of the starting valuations. My own view is that future returns will probably resemble an upside-down "U". The most/least risky stocks will not have great returns; the ones in the middle will.

That market forecast is if course worth every dollar you paid for it. :happy

On the Vanguard fund, I think the currency hedging ends up making a pretty big impact on the holdings. If you compare the top 10 holdings with the iShares global fund (ACWV) vs. the Vanguard fund, there is almost no overlap. The top holdings in the iShares fund are almost all US companies. For an unhedged portfolio, this makes sense; currency fluctuation makes international stocks more volatile to US investors. With a hedged portfolio, the VG fund can focus solely on the risk of stocks themselves without regard to what currency they trade in.
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Re: Global Minimum Volatility Fund

Post by grayfox »

grok87 wrote:
i like your framework. I guess i would add though that i think std. dev./variance is the wrong measure of risk. Academics like it because it makes the math tractable. So they can do lots of fancy (and mostly useless math). But in real life downside risk or drawdown is what matters. tails are fat and not modelled well by normal distributions. In terms of your 4 point framework:

1) I don't think i can estimate means and covariances. so no MVO for me.
2) obviously no GMV for me either
3) risk parity is a bunch of nonsense. basically says bonds are low risk so let's leverage them up. last summer showed the folly of that.
4) i'm partial to equal weighted but don't like the transaction costs
5) mostly i'm tending toward the market weight portfolio these days, at least for stocks. but with some small tilts towards small, value and minimum vol.

cheers,
1) Agree with that. The means are all over the place from one period to another. One period AAPL has 50% return. Another period -25%. So the optimum portfolio has 99% in AAPL one period and 0% another. MVO is clearly not practical.

2) Variances and correlations also vary over time, but not as much means. Because the statistics change, you really can't find the actual GMV portfolio for your future holding period. You can only find out what the GMVP was only afterwards. You are hoping that you got something close.

3) I don' think it makes sense to do Risk Parity mixing stocks and bonds together. If you don't think that you can estimate the means of each stock and their correlations, it is the same as assuming all the means (expected returns) and correlations are the same. Obviously a bond has different expected return than a stock, And correlation stock-to--stock and stock-to-bond are clearly different. Assuming they are all the same is clearly wrong. So you would form Risk Parity portfolio with only stocks.

4) If you say you can not estimate anything, then you are assuming that every stock has the same mean and variance, and all the correlations are the same. Then the optimal weighting is 1/N. There should be no transactions costs because theoretically there should be no transactions. You would put 1/N in each and hold until the end of the holding period, like 10 years or 20 years.

Now what is funny is that I had investments in S&P 500 from about early 1980s to about 2002 and the whole time I thought it meant I had 1/500 in every stock. :D I thought the S&P 500 and Fortune 500 were the same thing: the 500 biggest companies in the U.S. by sales, I assumed. I had no idea what market cap weighting meant. It seemed obvious to me that if you invested $10,000 in 100 stocks, you would put $100 in each. Not $1,999 in each of five stocks and and a five cents in the other 96. But that's what cap-weighting does.

During the dotcom bubble I was laughing at people putting so much money in overpriced tech stocks like Microsoft. What fools, I thought. Then I found out that my S&P 500 index fund wasn't 1/500 in MSFT but about 10%. :oops:

It's interesting that what I naively thought you would naturally do, i.e. equal weighting, is what the math says you should do if you have no opinion about means, variances and correlations.
User avatar
steve r
Posts: 1299
Joined: Mon Feb 13, 2012 7:34 pm
Location: Connecticut

Re: Global Minimum Volatility Fund

Post by steve r »

stlutz wrote:I think we're overstating the distinction between a minimum variance portfolio and a low beta portfolio. Actually what makes the iShares ETFs distinctive is not so much they they are running minimum variance portfolios but that the deviate from what the math tells them the minimum variance portfolio should be.

If you do the regressions on your own, a minimum variance portfolio will mostly end up with a bunch of utility and consumer staples stocks, just like a low beta portfolio will. What MSCI/iShares then do is impose some overriding criteria that move away from what they expect the minimum variance portfolio to be. For example, they impose sector and turnover constraints. A low-beta fund could do the exact same thing; S&P/Powershares simply chose not to with SPLV.

In backtesting, such constraints actually result in more volatility and lower returns. So, the indexes that these funds are tracking are not based on the best possible results that could be obtained in backtesting. Instead, the backtests built in assumptions for how you would actually want to run a real portfolio that one could sensibly hold a large portion of their assets in.
Well stated ... it is the constraints.
stlutz wrote: As with all of the "smart beta" strategies out there currently (value, small cap, quality etc.), valuations on these funds are not attractive right now. As such, I think one can invest and expect to get less volatility; I don't think the returns of a low. vol. strategy will be as attractive in coming years as in the past because of the starting valuations. My own view is that future returns will probably resemble an upside-down "U". The most/least risky stocks will not have great returns; the ones in the middle will.
Moreover, as you know, this (mid vol) is the long run sweet spot in terms of return. There is not easy way to play this. I might add, I do not like evaluations of many other assets right now.

stlutz wrote: On the Vanguard fund, I think the currency hedging ends up making a pretty big impact on the holdings. If you compare the top 10 holdings with the iShares global fund (ACWV) vs. the Vanguard fund, there is almost no overlap. The top holdings in the iShares fund are almost all US companies. For an unhedged portfolio, this makes sense;
Is it not the total in U.S. that matters? Both have 50+ percent in U.S./Canada. Any thoughts on what the impact of hedging (apart from direct effect of depreciating dollar good if un-hedged). This part I cannot get my mind around. My only thought is that the rest of my international portfolio is unhedged .... so some hedged may be of value.
grayfox wrote:
2) Variances and correlations also vary over time, but not as much means. Because the statistics change, you really can't find the actual GMV portfolio for your future holding period. You can only find out what the GMVP was only afterwards. You are hoping that you got something close.
+1. Which begs the question, is close "good enough?" For me, I expect it is as I market index cap weight the bulk of my portfolio. But I suspect opinions will vary here. Also, it leads to the question of the benefit of pure "indexing" with constraints, optimization models and the like if all you (or at least all I) hope to get is "good enough." Low cost management may also be good enough (and not saddled with only being able to re-balance twice a year with no more than X percent of portfolio).

I view the last parenthesis point as a weakness of pure indexing in this space (though I like the constraints in terms of sectors, country, etc ... I just hate the twice a year rule of ACWV and limits on what you can move out of ... it seems to me over x percent of your portfolio may no longer be predicted to have min vol in the future. Particular if variances shift over time.
"Owning the stock market over the long term is a winner's game. Attempting to beat the market is a loser's game. ..Don't look for the needle in the haystack. Just buy the haystack." Jack Bogle
grok87
Posts: 10512
Joined: Tue Feb 27, 2007 8:00 pm

Re: Global Minimum Volatility Fund

Post by grok87 »

grayfox wrote: 4) If you say you can not estimate anything, then you are assuming that every stock has the same mean and variance, and all the correlations are the same. Then the optimal weighting is 1/N. There should be no transactions costs because theoretically there should be no transactions. You would put 1/N in each and hold until the end of the holding period, like 10 years or 20 years.

Now what is funny is that I had investments in S&P 500 from about early 1980s to about 2002 and the whole time I thought it meant I had 1/500 in every stock. :D I thought the S&P 500 and Fortune 500 were the same thing: the 500 biggest companies in the U.S. by sales, I assumed. I had no idea what market cap weighting meant. It seemed obvious to me that if you invested $10,000 in 100 stocks, you would put $100 in each. Not $1,999 in each of five stocks and and a five cents in the other 96. But that's what cap-weighting does.

During the dotcom bubble I was laughing at people putting so much money in overpriced tech stocks like Microsoft. What fools, I thought. Then I found out that my S&P 500 index fund wasn't 1/500 in MSFT but about 10%. :oops:

It's interesting that what I naively thought you would naturally do, i.e. equal weighting, is what the math says you should do if you have no opinion about means, variances and correlations.
I'm not opposed to 1/N weighting. But the trouble is if you hold for a year or two then prices change and you are no longer 1/N. So you have to rebalance and that has costs.
Say for example you do 1/N on the S&P 500. Then a couple of years later your friend comes to you and wants to do 1/N. Are you really going to tell him to buy the same portfolio you own now? (which started out 1/N a couple of years ago but now is wildly different due to price changes over those couple of years). Or are you going to tell him to do 1/N based on current market prices. If you tell him to do 1/N based on current market prices, why wouldn't you do that yourself- i.e. rebalance. why wouldn't you eat your own cooking?

And the trouble is, rebalancing has costs. Maybe not a huge issue for the stocks in the S&P 500. But why would one limit oneself to that basket of stocks, picked by some crazy committee? Why not do some broader list of stocks. Maybe not every stock but at least a list that captures 90-95% of the market cap of the whole market. Like the Russell 3000 maybe or something like that. If you do 1/N on that and rebalance then the transaction costs are going to be an issue for those smaller stocks. So it gets complicated.

LIke I said, these days I'm leaning toward market cap approach (total stock market, total international market, etc.) with some small tilts toward small, value, and minimum volatility.
RIP Mr. Bogle.
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Re: Global Minimum Volatility Fund

Post by grayfox »

grok87 wrote:
I'm not opposed to 1/N weighting. But the trouble is if you hold for a year or two then prices change and you are no longer 1/N. So you have to rebalance and that has costs.
Say for example you do 1/N on the S&P 500. Then a couple of years later your friend comes to you and wants to do 1/N. Are you really going to tell him to buy the same portfolio you own now? (which started out 1/N a couple of years ago but now is wildly different due to price changes over those couple of years). Or are you going to tell him to do 1/N based on current market prices. If you tell him to do 1/N based on current market prices, why wouldn't you do that yourself- i.e. rebalance. why wouldn't you eat your own cooking?
With the new investment, the same logic would apply as with the original investment. If your assumption is all stocks have the same distribution of outcomes, you would invest as 1/N.

What about the original investment? Should you rebalance back to 1/N?

If there are transaction costs, then you would never rebalance back to 1/N because it would only reduce the value of the investment with no change in the expected outcome of portfolio. Again assuming that all the stocks have the same expectations, even if stock A doubled and stock B halved, the portfolio expected return is still the same. 0.5 * mu + 0.5 * mu = 0.2 mu + 0.8 mu
Last edited by grayfox on Mon Apr 21, 2014 10:57 am, edited 1 time in total.
grok87
Posts: 10512
Joined: Tue Feb 27, 2007 8:00 pm

Re: Global Minimum Volatility Fund

Post by grok87 »

grayfox wrote:
grok87 wrote:
I'm not opposed to 1/N weighting. But the trouble is if you hold for a year or two then prices change and you are no longer 1/N. So you have to rebalance and that has costs.
Say for example you do 1/N on the S&P 500. Then a couple of years later your friend comes to you and wants to do 1/N. Are you really going to tell him to buy the same portfolio you own now? (which started out 1/N a couple of years ago but now is wildly different due to price changes over those couple of years). Or are you going to tell him to do 1/N based on current market prices. If you tell him to do 1/N based on current market prices, why wouldn't you do that yourself- i.e. rebalance. why wouldn't you eat your own cooking?
With the new investment, the same logic would apply as with the original investment. If your assumption is all stocks have the same distribution of outcomes, you would invest as 1/N.

What about the original investment? Should you rebalance back to 1/N?

If there are transaction costs, then you would never rebalance back to 1/N because it would only reduce the value of the investment with no change in the expected outcome of portfolio. Again assuming that all the stocks have the same expectations, even if stock A doubled and stock B halved.
0.5 * mu + 0.5 * mu = 0.2 mu + 0.8 mu
yeah but your original investment is no longer 1/N. it's basically a somewhat more random allocation to stocks. I'm not sure I buy your argument with the mu's. I think one could use that argument to justify buying and holding just a single stock instead of a broad 1/N portfolio.
so I think if one really believes in 1/N one has to rebalance.
I can see a compromise approach where one does 1/N on the S&P 500 stocks where transaction costs are low. and then holds a market cap weighted portfolio of say small value stocks to get the small and value tilts.
RIP Mr. Bogle.
grayfox
Posts: 5569
Joined: Sat Sep 15, 2007 4:30 am

Re: Global Minimum Volatility Fund

Post by grayfox »

grok87 wrote: yeah but your original investment is no longer 1/N. it's basically a somewhat more random allocation to stocks. I'm not sure I buy your argument with the mu's. I think one could use that argument to justify buying and holding just a single stock instead of a broad 1/N portfolio.
so I think if one really believes in 1/N one has to rebalance.
For all the MPT portfolios, the portfolio weights are only the initial weights. Those initial weights will determine the portfolio's probability distribution of outcomes at the end of the holding period. Of course, the weights will change during the holding period and be different at the end. Classic MPT is a single-period analysis and there is no rebalancing.

For example, if your holding period is 10 years and you choose portfolio weights using MVO optimization, those would be the initial weights. Say 92% A, 5% B, 3% C. At the end it could end up 1% A, 98% B, 1% C. It's irrelevant. You pick the initial weights that gives you some expected distribution of portfolio outcomes at 10 years. After 10 years, you will have some random sample from that distribution.
User avatar
spanky123
Posts: 145
Joined: Sun May 22, 2011 11:15 am

Re: Global Minimum Volatility Fund

Post by spanky123 »

Will this fund be tax-efficient?
stlutz
Posts: 5585
Joined: Fri Jan 02, 2009 12:08 am

Re: Global Minimum Volatility Fund

Post by stlutz »

VG has said that tax efficiency is not a goal of the fund (e.g. expect a high turnover ratio) and that it should be held in a tax-advantaged account.
User avatar
dnaumov
Posts: 495
Joined: Tue Jul 27, 2010 6:04 pm
Location: Finland
Contact:

Re: Global Minimum Volatility Fund

Post by dnaumov »

steve r wrote:I own this fund for a large share of my portfolio.

Previously, I owned a similar index (ACWV).
I took a look at ACWV and noticed it trades at a rather hefty premium to the global total stock market:

ACWV: PE 22.78, PB 4.28
VT: PE 18.3, PB 2.0

Pretty interesting, considering that I previously read that low volatility/beta funds tend to generally have a value tilt. This is most definately not the case here.
User avatar
steve r
Posts: 1299
Joined: Mon Feb 13, 2012 7:34 pm
Location: Connecticut

Re: Global Minimum Volatility Fund

Post by steve r »

dnaumov wrote: I took a look at ACWV and noticed it trades at a rather hefty premium to the global total stock market:

ACWV: PE 22.78, PB 4.28
VT: PE 18.3, PB 2.0

Pretty interesting, considering that I previously read that low volatility/beta funds tend to generally have a value tilt. This is most definately not the case here.
+1 ... Likewise for VMNVX (fund of thread). Low volatility is not value. I have also read this to from some prominent posters / authors.

To be fair, studies using a F/F factor framework find you can explain a good chunk (third to two thirds depending on study and what else you control for) of the low vol premium with other factors including value. Moreover, value investing and low volatility investing both underperform in strong "growth" bull markets such as the late 1990s. But as far as common holdings, this is simply not the case.
"Owning the stock market over the long term is a winner's game. Attempting to beat the market is a loser's game. ..Don't look for the needle in the haystack. Just buy the haystack." Jack Bogle
User avatar
cfs
Posts: 4154
Joined: Fri Feb 23, 2007 12:22 am
Location: ~ Mi Propio Camino ~

Re: Global Minimum Volatility Fund

Post by cfs »

And talking about minimum volatility . . .

Note, I decided to do a search for conversations about this fund after reading Larry's "The Volatility Anomaly Uncovered." As far as Vanguard Global Minimum Volatility Fund, it is hanging there but this is just a baby fund, so, let's check back in December 2033 after it reaches adulthood to talk about past performance.

Thank you and have a Feliz Navidad.
~ Member of the Active Retired Force since 2014 ~
Post Reply