Vanguard correlation matrix

[IndependentlyPoor]

Quote:

Originally Posted by Free To Canoe I find some things surprising. EM to S&P 500 for long periods .62. REIT to S&P 500 for long periods .68. Do I find some things fishy? The Target Retirement funds as a group seem to have that great negative correlation to the equity markets. Even better than the bond funds. No facts here, just feelings. Free to canoe

Quote:

Originally Posted by FUEGO That probably explains why my own analysis produced different results. I looked at correlation of yearly performance and found the REIT fund to be less correlated with most other equity classes than what your results show.

I rechecked the REIT (VGSIX) , Emerging Markets (VEIEX), and S&P500 (VFINX) correlations sort of by hand. By that I mean that I downloaded new data and did each step separately instead of embedded in a program and got the same results.

I am not sure what Free to canoe means by great negative correlations of the Target Retirement funds. For example, using Target 2005 (VTOVS) and Target 2050 (VFIFX), I see the following correlations to Total Stock (VTSMX)
VTSMX/VFIFX: 0.99
VTSMX/VTOVX: 0.78
compared to Total Bond (VBMFX)
VTSMX/VBMFX: 0.56
Am I missing something here, or looking at the data crosseyed?

There are significant differences in the amount of data available for each fund. Here are the number of data points:
VTSMX 3445
VTOVX 1595
VTENX 924
VGSIX 3439
VFINX 5780
VEIEX 3641

For each correlation pair, I used data from inception of the youngest fund to Feb, 10, 2010 (the day I downloaded the dataset). I deleted the following dates from each series because some funds were missing these dates:
{{2001, 11, 22}, {2005, 6, 9}, {2005, 10, 24}, {2005, 10, 25}}

I will continue to check the correlations and post updates if I find anything significant.

[Free To Canoe]

Quote:

Originally Posted by IndependentlyPoor I rechecked the REIT (VGSIX) , Emerging Markets (VEIEX), and S&P500 (VFINX) correlations sort of by hand. By that I mean that I downloaded new data and did each step separately instead of embedded in a program and got the same results.

This is a fine kettle of fish that I got myself into.

What I meant about the Vanguard EM and REIT .62 and .68 correlations to the Vanguard S&P 500 fund being surprising is not that the numbers are inaccurate but that they seem higher than some of the historical numbers used for emerging market and real estate correlations in general (according to my memory).

In the article Emphasizing Low-Correlated Assets: The Volatility of Correlation by William Coaker II in the Journal of Finance 2007 they list the correlations as .50 and .52. Lower but not really surprisingly lower. Recent historical events may have tended to bump them up.

Your recent research has given this untrusting soul some confidence in the Vanguard matrix.

Quote:

Originally Posted by IndependentlyPoor I am not sure what Free to canoe means by great negative correlations of the Target Retirement funds. For example, using Target 2005 (VTOVS) and Target 2050 (VFIFX), I see the following correlations to Total Stock (VTSMX) VTSMX/VFIFX: 0.99 VTSMX/VTOVX: 0.78 compared to Total Bond (VBMFX) VTSMX/VBMFX: 0.56 Am I missing something here, or looking at the data crosseyed?

No, I am looking at the matrix crosseyed. Sorry.

Free to canoe

[Free To Canoe]

Some of the target funds and the short term bond funds have some of the strongest inverse correlations in the matrix.
Specifically, the Target Retirement 2030, Target R. 2040 and Target R. 2050 and the first 5 funds in the matrix. Why are Target R. 2025, 2035 and 2045 somewhat lower than their sister funds?

Free to canoe

[IndependentlyPoor]

I was going cross-eyed looking at the huge correlation matrix so I am trying some other ways of presenting the data.
Here is a pdf of all 5671 correlations, sorted with the negative correlations first. The 4th column is the number of data points included in the correlation. The 5th column is the correlation itself.
Vanguard correlation table.pdf
I found it interesting the Market Neutral Fund (VMNFX) appeares in the first 9 places. The 9th place is also interesting because it is a pair of equity funds with high negative correlation, although the small number of data points casts doubts on its reliability.

The "obvious" VTSMX/VBMFX combination doesn't show up until #1104 with a positive correlation of about 0.56. In fact, this correlation is much worst than the VWUSX/VTSMX combination at #675 with a correlation of about 0.31. Unless there is an error lurking in my correlations, this must be due to the tremendous hit that US Growth took in 2000 and its subsequent non-recovery

Also attached are correlation tables for VTSMX and VBMFX. These are essentially rows (or columns) from the big correlation matrix. I find it interesting (almost impossible to believe) that VTSMX is positively correlated with all the other funds.
Vanguard VTSMX correlation table.pdf
Vanguard VBMFX table.pdf

[freebird5825]

Quote:

Originally Posted by IndependentlyPoor I was going cross-eyed looking at the huge correlation matrix so I am trying some other ways of presenting the data. Here is a pdf of all 5671 correlations, sorted with the negative correlations first. The 4th column is the number of data points included in the correlation. The 5th column is the correlation itself. Attachment 8387 I found it interesting the Market Neutral Fund (VMNFX) appeares in the first 9 places. The 9th place is also interesting because it is a pair of equity funds with high negative correlation, although the small number of data points casts doubts on its reliability. The "obvious" VTSMX/VBMFX combination doesn't show up until #1104 with a positive correlation of about 0.56. In fact, this correlation is much worst than the VWUSX/VTSMX combination at #675 with a correlation of about 0.31. Unless there is an error lurking in my correlations, this must be due to the tremendous hit that US Growth took in 2000 and its subsequent non-recovery Also attached are correlation tables for VTSMX and VBMFX. These are essentially rows (or columns) from the big correlation matrix. I find it interesting (almost impossible to believe) that VTSMX is positively correlated with all the other funds. Attachment 8388 Attachment 8389

Um...first pdf was all white space. I zoomed out to be sure.
2nd and 3rd pdf files both contained data.

[IndependentlyPoor]

Well, well, well. It looks like Adobe Reader has a problem with this file. Perhaps the file is too large. I generally use Apple's PDF reader, Preview, and it works just fine. I'll see if I can't figure out why Adobe Reader chokes on it.

[Midpack]

I continuously calculate the correlation between all my holdings in Excel as part of my tracking. It's not that hard to do.

[IndependentlyPoor]

Dunno what is wrong with the PDF versions and too lazy to figure it out. Here is a plain text version of the first 500 pairs.

Vanguard Correlation Table part 1.txt

[Free To Canoe]

Quote:

Originally Posted by Midpack I continuously calculate the correlation between all my holdings in Excel as part of my tracking. It's not that hard to do.

Interesting concept. Very nice use of colors!

Quote:

Originally Posted by IndependentlyPoor Dunno what is wrong with the PDF versions and too lazy to figure it out. Here is a plain text version of the first 500 pairs. Attachment 8392

Nice to have the raw data. I can get it into Excel from here.

Free to canoe

[IndependentlyPoor]

Warning:

The correlations should polly be calculated using the fractional daily NAV change instead of the raw NAV. I am still thinking about it. Will probably post updated correlations later.

Sorry,

[gcooper]

My VG portfolio is all in the +0.9's. How do I use this correlation chart to get better diversification

[Midpack]

Quote:

Originally Posted by gcooper My VG portfolio is all in the +0.9's. How do I use this correlation chart to get better diversification

Sounds like your holdings are all large to medium stock funds. I shared my actual holdings above. But here is another chart that shows in general what's normally highly correlated and what's not. Best of luck...

[Spanky]

Midpack,
Nice chart. Thanks. It seems that adding national resources, global bonds and high yield bonds provide increased diversification.

[target2019]

Quote:

Originally Posted by Midpack I continuously calculate the correlation between all my holdings in Excel as part of my tracking. It's not that hard to do.

Can you give us a peek behind the scenes, into the data and formulas that generate your chart?
Also, what is the data source for the other chart in this post?
Thanks. Interesting discussion, and I'm all ears.

[photoguy]

Quote:

Originally Posted by IndependentlyPoor Daily, over the longest period possible during which each pair of funds exist, so some of the correlation pairs are over much longer periods than others.

Any reason why you picked a daily timeframe? It would seem to me that for most people who are passive investors, a yearly timeframe would make more sense.

[REWahoo]

photoguy, I doubt you'll get a response. IndependentlyPoor hasn't been heard from for several months.

[photoguy]

Ah thanks. I didn't realize this was an older thread.

[Midpack]

Quote:

Originally Posted by target2019 Can you give us a peek behind the scenes, into the data and formulas that generate your chart? Also, what is the data source for the other chart in this post? Thanks. Interesting discussion, and I'm all ears.

Hopefully I won't regret this and I make no guarantees, but here's the Excel calcs I use to watch actual correlation on my holdings. If someone sees an error, I'd appreciate a heads up. These are all built-in Excel functions, you can research them online if needed. No $ shown and I plugged in an arbitrary AA. Best of luck.

It's not something that I need to act on, I use correlation when I am choosing holdings, and only track after that to make sure the correlation (or lack thereof where desired) is holding up within reason. It would be a mistake to act on these often IMO, there are times when everything will seem to move together for short periods, last meltdown there was no place to hide money except cash and bonds to some extent.

And here's where that other chart came from, but I just Googled for it, I've seen these several times from several sources.

http://www.doughroller.net/wp-conten...ationchart.gif

[Joshua]

I thank you folks for digging up this thread. I needed a reminder about how correlation related to diversification. Since I have just recently finished a statistics course, this actually makes more sense to me than it would have in the past.

[gcooper]

Quote:

Originally Posted by Joshua I thank you folks for digging up this thread. I needed a reminder about how correlation related to diversification. Since I have just recently finished a statistics course, this actually makes more sense to me than it would have in the past.

Okay, now I have this nifty chart, how do I use it?
I have:
VGHAX ADM HEALTH
VASGX LIFE STRAT. GROWTH
VWNFX WINDSOR II
VWELX WELLINGTON
VWINX WELLESLEY
VIMSX MID CAP
VFINX 500 INDEX

What correlation range do I want ?
How do get around choosing a fund that has a good correlation
with one, but not the others?

I'm feeling very dumb