Bogleheads

CyberBob

So I noticed at Morningstar that balanced funds always seem to be compared to both their category and the Dow Jones Moderate Portfolio. The graph usually looks something like this:

I noticed that the funds often don't meet the return of that index (green line), but I wasn't really sure what that index consisted of, so I set off to investigate.

The Dow Jones Portfolio Indexes are listed here. I won't bore you with the details of them but rather cut to the reason for my post.

The interesting thing about those indexes is how they are constructed. For example, for the U.S. stock allocation they don't simply use something like the Dow Jones Wilshire 5000 index, but rather represent U.S. stocks with six subindexes:

Large-Cap Growth
Large-Cap Value
Mid-Cap Growth
Mid-Cap Value
Small-Cap Growth
Small-Cap Value

And International stocks were represented with:

Europe
Pacific
Emerging Markets

A global stock allocation simply being equal parts of all nine of the above subindexes, which would give you a 2/3 U.S. and 1/3 International allocation.
This really intrigued me. With all of the talk about overweighting small-cap or value, I haven't seen much on evenly-weighting sectors. Evenly weighting by stock, such as the evenly-weighted S&P 500 index comes up occasionally, but this is difficult due to the churning required to constantly rebalance.

So here was an apparently completely agnostic allocation that just equally weighted the six major defined sectors in the U.S. and the three major defined regions in the rest of the world. Would this evenly-weighted everything type of index make for a good possible allocation?
Using the corresponding Vanguard funds, the combination of the six U.S. sectors gives an interesting Morningstar X-Ray of:
10 10 13
14 15 20
06 06 07

Compared to the more top-heavy Total Stock Market Index:
24 23 25
06 06 08
03 03 03

I fired up my OpenOffice Calc spreadsheet, and used Simba's Backtest Spreadsheet (available in .ods or .xls) to find that indeed, this evenly-weighted type of stock allocation beat the return of both the Couch-Potato Portfolio and a portfolio consisting of Total Stock Market Index (2/3) and Total International Stock Index (1/3). (Note: all portfolios are balanced and contain 40% bonds).

Here are the graphs for the two time periods used in Simba's spreadsheet:

I'm thinking this even-weighting by sector allocation has promise because, if regularly rebalanced, it would seem to eliminate the possibility of a bubble in any particular sector.
Has anyone else ever read about or thought about using a sector-agnostic allocation like this?

Bob

DaveTH · Joined: 05 Apr 2007 Posts: 2447

Bob, I think that is what Paul Merriman has been advocating for years. He recommends equal-weight allocations to 10 equity asset classes (Domestic & International).

Blackhawkzone · Joined: 06 Mar 2007 Posts: 295 Location: Chicago

the problem with total stock market and total international stock index is that they are basically just large cap funds with very little mid and small cap.

I guess what you typed is a variation of merriman's portfolios with midcap added without value overweighted

bob90245 · Joined: 19 Feb 2007 Posts: 4174

HearDoc · Joined: 31 May 2007 Posts: 453 Location: New England

sterjs · Joined: 25 Mar 2007 Posts: 281

Any "strategy" that overweights midcaps(like this one) will crush everything in backtests for 1985-2007.

bob90245 · Joined: 19 Feb 2007 Posts: 4174

fundtalker123 · Joined: 27 Feb 2007 Posts: 285

Bob, such analysis is fun, but you're not really thinking of going over to the "dark side" are you? Wink

(i.e. non market cap weighting US market) Maybe you should just take a cold shower and re-read Bogle's latest book? Well, ok, Bogle says you can mess around a tilting maybe 15% of equity to small and value (or perhaps some other asset class you believe might outperform. Growth stocks or REITS, maybe? Wink

). Don't get too fancy.

grok87 · Joined: 27 Feb 2007 Posts: 3365

Here's what I do for my equity allocation (roughly):

Large Cap 1/3
Mid/Small Cap 1/3
Intrenational 1/3

IMHO putting twice as much in mid/small cap as large cap is a bit extreme.

cheers
grok

bob90245 · Joined: 19 Feb 2007 Posts: 4174

Les · Joined: 10 Mar 2007 Posts: 1706 Location: Northern Calif.

Bob (Cyberbob), thanks for the interesting post. It seems like your objective was to get 33/33/33 in L/M/S but instead wound up with 33/49/19. Was wondering what funds (or proxies in Simba's data) you used to construct the US portfolio and why not revise the mix to get closer to the objective? Or was it because of the international components that you did not get the 33/33/33 and if so was the US part giving you close to 33/33/33? Also what VG funds and allocations would fit this portfolio?
_________________
Les

CyberBob · Posted: Fri Jul 06, 2007 9:36 pm Post subject:

nick22 · Joined: 04 Mar 2007 Posts: 803 Location: Ohio

Fun with CyberBob,
That is interesting. I just have a slight tilt domestically with a total equity breakdown of 30% US Large, 10% US medium, 10% US Small, 20% Europe, 10% Pacific, 10% EM, 10% Int small, 10% Int value. So I guess I overweight the US large and Europe relative to a balanced pie somewhat.
_________________
Nick22

CyberBob · Posted: Fri Jul 06, 2007 9:51 pm Post subject:

Looking at a Periodic Table of Returns, where everything appears to just be a random jumble, also might be an argument for an agnostic, evenly-weighted portfolio.

Bob

Dino · Joined: 25 Feb 2007 Posts: 155

Thanks Bob for the interesting post.
Grok, you mentioned your 1\3 LC, 1\3 Mid & SC and 1\3 International. It seems to me the international portion in most VG funds is heavily weighted to LargeCap, so doesn't that count as part of your LC portion and thus your LC portion is much more than 1\3 total??

Dino

Les · Joined: 10 Mar 2007 Posts: 1706 Location: Northern Calif.

Bob, thanks very much for the replies. I'll have to think about this a little and maybe play with the allocations in Simba's spreadsheet. I assume I can use the Boglehead's spreadsheet from Gummy for this.

I find it quite interesting that the 33/33/33 graph is almost identical to the 33/49/19 one. This brings up the question of just how sensitive portfolio allocations are in the mid/small area. If the answer is that it doesn't matter that much then this would help with need to rebalance (if nothing more).
_________________
Les

Trev H · Joined: 02 Mar 2007 Posts: 1573

.
Good Morning Bogleheads...

When I first got my hands on CRSP Data... I did a crunch looking at different combinatins of Large+Small vs Large+Mid+Small vs Mid Caps Only...

The CRSP Data is "BLEND" so the results should be similar to the comparisons by Bob... for example instead of holding LV & LG... hold Large Blend... or MV and MG... Mid Blend... same for Small.

I did a comparison in 20 year slices 1927-2005 and for the full span.

The portfolio of 100% Mid Caps or 1/3 Each Large/Mid/Small did well in each period.... always putting you on the high-to-high/mid end of the scale as far as possible portfolio success for the period.

100% Mid Caps or 1/3 Each also outperformed 50/50 Large/Small in each period and for the full span.

Your best bet for having the worst performing portfolio was to be dominated by Large Caps... which we all know that TSM is exactly that.

Now that is a look at (performance possibilities) without regards to risk... but the table also includes standard deviation so you can see at least that measure of risk in each combination.

Trev H

grok87 · Joined: 27 Feb 2007 Posts: 3365

CyberBob · Posted: Sat Jul 07, 2007 1:24 pm Post subject: Re: Morning...

Les · Joined: 10 Mar 2007 Posts: 1706 Location: Northern Calif.

Trev, thanks for the data. If you happen to have it, I'd be interested in seeing a similiar analysis for high stress periods like 1929-1938 and 1966-1975. Basically I'd be interested in the periods which test investor resolve. Am interested because I'm always trying to balance greed/fear. I guess your 75/25 is similiar to TSM's 70/20/10.

Bob, interesting point about trying to avoid blend's skewing to growth or value. Is there a historical period you are thinking of? If we characterize the late 90's to 2000 as a large cap growth bubble then it would be interesting to compare a rebalanced portfolio of L/M/S blends versus a rebalanced value/growth L/M/S portfolio. Did rebalancing growth/value in L/M/S improve portfolio performance much between say 1995 - 2005? I would guess that it might have in the large category but having an equal weighting of small growth may have subtracted from the benefit. I know that VTSMX has had periods of growth bias and also of value bias.
_________________
Les

Les · Joined: 10 Mar 2007 Posts: 1706 Location: Northern Calif.

I should add my data on the VTSMX (total stock market) fund from M* as it may be of interest here:

bob90245 · Joined: 19 Feb 2007 Posts: 4174

CyberBob · Posted: Sat Jul 07, 2007 6:12 pm Post subject:

Les · Joined: 10 Mar 2007 Posts: 1706 Location: Northern Calif.

For Feb 2006 I have data characterizing some recommended portfolios of interest for this thread:

unclemick · Posted: Sat Jul 07, 2007 7:16 pm Post subject:

And of course all will faithfully rebalance without any shred of emotion through all investment climes(like a postman with the mail) as the decades roll on - right? Smile

My hat is off to those who can do it. I failed over a modestly short (1966 - 2006) period.

Love the data/graphics presented here though. Bogle on!

heh heh heh - Chickenheartedness and Target Retirement for this cat.

A little tongue in cheek razz. Don't fall tooo much in love with past numbers - trust me I was an en - ga - neer in a prior lifetime - lefthanded at that.

MiRa55 · Joined: 07 Jul 2007 Posts: 3

I weighted each of these as shown

VIVAX 19.00%
VIGRX 21.00%
VMVIX 3.00%
VMGIX 0.00%
VISVX 27.00%
VISGX 30.00%

And got the following equity style box

33.64% 32.68% 33.68%

33.59% 10.69% 12.77% 10.12%
33.30% 11.54% 9.46% 12.30%
33.11% 11.41% 10.45% 11.25%

The column and row totals are above and to the left. This demonstrates to get equal weighting in each of the 9 boxes requires some gymnastics.

Are these weightings close to what you used above?

Hope this post works. It's my first time on board.

CyberBob · Posted: Sun Jul 08, 2007 10:28 am Post subject: Re: Hi Bob,

IlliniSigEp · Joined: 01 Mar 2007 Posts: 174 Location: Chicago, IL

50/50 vs. Blend rebalancing bonus would depend on the underlying index being used. There would be no rebalancing bonus using Vangaurd's MSCI Value/Growth indices 50/50 vs. blend. MSCI split the equity universe into 50% of the market cap in growth, the other half in value. By owning 50/50 of MSCI, you end up with the total market cap.

-Dave

ajbibi · Joined: 07 May 2007 Posts: 195

The size of the Mira55 drop indicates that anything beyond market weights probably involves taking on different/higher levels of risk (not just "high standard deviation" risk). This also suggests why historical data mining is dangerous. My guess is that anything beyond market weights -- whether value/small tilt or equal weights -- buys extra return at the price of higher risks of different types even if backtesting seems to suggest that it is "safe."

This is the point that Sharpe wanted to make in his comments on Fama/French (quoted in another post).

Random Musings · Joined: 22 Feb 2007 Posts: 2410 Location: Pennsylvania

I also get nervous when I see the "best" portfolio combination backtested over a period of time - and the effects that recency have in choosing the winner.

Kinda like the current efficient frontier combination du jour. You know it's going to change. I expect the MIRa55 to underperform the equally balanced or 33 portfolio's over the next 20 years; why not? - it did at the beginning of the data set looked at.

RM

Les · Joined: 10 Mar 2007 Posts: 1706 Location: Northern Calif.

Bob, you might want to show your plots with semilog scale for the y-axis. This would help to display the nasty 1973-1974 market.
_________________
Les

CyberBob · Posted: Sun Jul 08, 2007 4:55 pm Post subject:

springwater · Joined: 11 Mar 2007 Posts: 268

For a domestic equity allocation, if you put 50% in VTI Vanguard Total Market, 25% in IJJ S&P 400 Value Mid Cap, and 25% in IJS S&P 600 Small Cap Value, you get a pretty balanced portfolio with a slight value tilt in X-Ray.

12 12 12

14 10 10

12 11 7

I don't know what mix of funds will acheive this in International. Probably a mix of VEU, VWO, DLS, and GWX will get you close.

modal · Joined: 20 Feb 2007 Posts: 1080 Location: USA

My 401K is equally balanced...

20% Large Cap Index
20% Mid Cap Value (Non Index)
20% Small Cap Value (Non Index)
20% International Value (Non Index)
20% Pimco Total Return Bonds

Sunny Sarkar · Joined: 02 Mar 2007 Posts: 1360 Location: Dallas-Ft.Worth

This thread looks like S&D vs TSM revisited. The only new point this raises vis-a-vis a TSM vs S&D debate is whether the S&D-ers were right in eliminating mid caps. The four corners hypothesis was, I think, that mids will stay true to their name by returning midway between large and small, thus not making any difference. Bob's post puts that hypothesis to the test.

I always liked the following 10x10 portfolio:
10% each of LG, LV, SG, SV, Dev, Pac, EM, RE, TBM, TIPS
but the rebalancing and tracking error would drive me crazy.