| View previous topic :: View next topic |
| Author |
Message |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Thu Aug 28, 2008 9:52 am Post subject: Fun with efficient frontiers 1972-2007 |
|
|
OK, I am reinventing the wheel for fun and generating efficient frontiers. One of my favorite posts is Updated figures and results on slice and dice vs TSM. I wanted to know what things would look like with international stocks, REITs, and commodities.
Each colored line below represents a portfolio that dilutes a 100% Total Bond Market with some stock portfolio. (OK, I am counting commodities as stocks for now. Please excuse me for that.)
Data are taken from Spreadsheet for backtesting (includes TrevH's data), which means that:
- Returns are net of Investor Shares' expenses for years in which Vanguard funds existed (except for PCRIX, which is from PIMCO).
- Returns are index performance for years in which Vanguard funds did not exist.
- The universe consists of 26 funds.
Efficient
10% PIMCO CommodityRealRet Strat Instl (PCRIX) (0.75%)
60% Vanguard Small Cap Value Index (VISVX) (0.22%)
30% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
SCV40
20% Vanguard Total Stock Mkt Idx (VTSMX) (0.15%)
40% Vanguard Small Cap Value Index (VISVX) (0.22%)
10% Vanguard REIT Index (VGSIX) (0.20%)
30% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
4x25
15% Vanguard 500 Index (VFINX) (0.15%)
15% Vanguard Value Index (VIVAX) (0.20%)
15% Vanguard Small Cap Index (NAESX) (0.22%)
15% Vanguard Small Cap Value Index (VISVX) (0.22%)
10% Vanguard REIT Index (VGSIX) (0.20%)
30% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
SCV20
40% Vanguard Total Stock Mkt Idx (VTSMX) (0.15%)
20% Vanguard Small Cap Value Index (VISVX) (0.22%)
10% Vanguard REIT Index (VGSIX) (0.20%)
30% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
Swensen
42.9% Vanguard Total Stock Mkt Idx (VTSMX) (0.15%)
28.6% Vanguard REIT Index (VGSIX) (0.20%)
28.6% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
Core Four
60% Vanguard Total Stock Mkt Idx (VTSMX) (0.15%)
10% Vanguard REIT Index (VGSIX) (0.20%)
30% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
Target Retirement
80% Vanguard Total Stock Mkt Idx (VTSMX) (0.15%)
20% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
VTSMX
Vanguard Total Stock Mkt Idx (VTSMX) (0.15%)
1972-2007
1972-1980
1981-1989
1990-1998
1999-2007
Observations
As everybody knows, 1990's were strange years when domestic large-cap did extremely well. The ordering of lines are pretty much completely reversed compared to other periods. Other than that, these charts are pretty consistent with Updated figures and results on slice and dice vs TSM.
I looked at those very efficient portfolios that were more efficient than Efficient above. Many of them had a huge bet on Emerging Markets. I don't know if people put such a bet on Emerging Markets starting in 1970's let alone the availability of a low-cost product for Emerging Markets.
The "crosssection" at standard deviation 14% for the 1972-2007 period is as follows:
| Code: | Efficient 13.90%
SCV40 12.92%
4x25 12.73%
SCV20 12.40%
Swensen 12.29%
Core Four 11.70%
Target Retirement 11.16%
VTSMX 10.68% |
Caveats, Disclaimer, etc
The blue dots are not very exhaustive. To cut down the computation time, I used a multiple of 10% allocation for each fund. On top of that, I sampled one in every 25 portfolios.
As mentioned above, I'm mixing index performance and actual fund performance, which may not necessarily be a wise thing to do even though Vanguard index funds tend to have very small tracking errors.
I may be miscalculating something.
Past performance is not a guarantee of future results. |
|
| Back to top |
|
 |
Rodc
Joined: 26 Jun 2007 Posts: 5395
|
Posted: Thu Aug 28, 2008 10:18 am Post subject: |
|
|
It will take a bit to digest all this, but I wanted to be the first to thank you for this work. Very cool. _________________ "all standard caveats apply" |
|
| Back to top |
|
 |
Rodc
Joined: 26 Jun 2007 Posts: 5395
|
Posted: Thu Aug 28, 2008 10:33 am Post subject: |
|
|
Those results are very consistent in nature with my earlier results. This is very reassuring as I was using the Fama-French Benchmark Portfolios and you are using actual mutual funds for the most part and otherwise using live indexes upon which the real funds are based. So two different datasets (although with over-lap in time) led to similar results.
Given that in my work I saw large(ish) differences for the slide and dice portfolios over 20 year periods, but extremely similar results when averaged over 80 years, I might expect the differences you found to narrow with time as well.
Also, in playing with mean variance optimization with other datasets including Emerging Markets I too found that the very most efficient portfolios (in hindsight) included a ton of emerging markets hedged with short term bonds. Not sure I'd count on that repeating though. I agree with your decision not to include them as viable options.
Good work. _________________ "all standard caveats apply" |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Thu Aug 28, 2008 11:17 am Post subject: |
|
|
| Rodc wrote: | | Those results are very consistent in nature with my earlier results. This is very reassuring as I was using the Fama-French Benchmark Portfolios and you are using actual mutual funds for the most part and otherwise using live indexes upon which the real funds are based. So two different datasets (although with over-lap in time) led to similar results. |
Right. Even then I would add that the 1999-2007 period is the only one generated mostly with real data. Here are the data sources for the funds I used in the above portfolios according to Simba's spreadsheet.
| Quote: | MKT = US Cap Weighted Market:
=============================
CRSP Market Decile 1-10 1972-1992
Vanguards Total Stock Market Index Fund 1993-2007
LCB = US Large Blend:
=====================
S&P 500: Standard & Poors 1972-1976
Vanguards 500 Index Fund 1977-2007
LCV = US Large Value:
=====================
Fama and French 1972-1978
Russell 1000 Value Index 1979-1992
Vanguards Value Index Fund 1993-2007
SCB = US Small Blend:
=====================
Ibbotson 1972-1978
Russell 2000 Index 1979-1991
Vanguards Small Cap Index Fund 1992-2007
SCV = US Small Value:
=====================
Ibbotson 1972-1978
Russell 2000 Value Index 1979-1998
Vanguards Small Cap Value Index Fund 1999-2007
REIT = Real Estate:
===================
Nat. Assn. of Real Estate Inv Trusts 1972-1996
Vanguards REIT Index Fund - 1997-2007
EAFE/EM
INTLT - Total International:
============================
MSCI EAFE Index 1972-1987 (Developed Only)
85% EAFE Index
Vanguards Total International Index Fund 1997-2007
ITB = Intermediate Term Bonds:
==============================
Ibbotson 1972
Lehman Brothers 1973-1986
Vanguards Total Bond Index Fund 1987-2007
Commodities
============================
COMM - Commodities/Natural Resources
Collaterallized Chase Index 1972-1990
DJ-AIJ plus T-Bills 1991-1996
DJ-AIG plus Yahoo's IPS 'category' returns 1997-2001 (prior to VIPSX)
DJ-AIJ plus VIPSX - 2001-2002
Pimco PCRIX 2003-2007 |
A fully indexed fairly efficient portfolio is very new in that regard.
| Rodc wrote: | | Given that in my work I saw large(ish) differences for the slide and dice portfolios over 20 year periods, but extremely similar results when averaged over 80 years, I might expect the differences you found to narrow with time as well. |
I was thinking you didn't have very large differences among S&D portfolios.
| Rodc wrote: | Also, in playing with mean variance optimization with other datasets including Emerging Markets I too found that the very most efficient portfolios (in hindsight) included a ton of emerging markets hedged with short term bonds. Not sure I'd count on that repeating though. I agree with your decision not to include them as viable options. |
Yeah. I would take some hint from backtesting but not that much hint.
Thanks. |
|
| Back to top |
|
 |
Rodc
Joined: 26 Jun 2007 Posts: 5395
|
Posted: Thu Aug 28, 2008 12:40 pm Post subject: |
|
|
| Quote: | Rodc wrote:
Given that in my work I saw large(ish) differences for the slide and dice portfolios over 20 year periods, but extremely similar results when averaged over 80 years, I might expect the differences you found to narrow with time as well.
I was thinking you didn't have very large differences among S&D portfolios.  |
Yeah, I just reread my post. The slice and dice portfolios had different returns and sd in different periods, but you are right in that they were pretty close together as a set in each period. The different components showed more variability across different periods. I might also add my slice/dice portfolios were more similar to each other than some of yours are, so it is not so surprising they performed more similarly.
Again, thanks for the work. _________________ "all standard caveats apply" |
|
| Back to top |
|
 |
detifoss
Joined: 13 Oct 2007 Posts: 363
|
Posted: Thu Aug 28, 2008 4:48 pm Post subject: |
|
|
I loved RodC's earlier post in this fashion, and your post is also fantastic - thank you!
I do have a few (simpleton) questions. Sorry for in advance if they are too basic.
As I understand it, the blue dots are representative of 'different' portfolio mixtures. You stated you used a multiple of 10% allocation for each fund, and you sampled one out of every 25 portfolios.
Since the 'universe' for this consists of 26 funds, does that mean that one of the blue dots could conceivably represent 100% EM (even though EM is not in any of the represented portfolios) or 100% REITs or 100% CCF? On the other end of the spectrum can a dot represent a fantastically complicated 26 'fund' slice and dice type portfolio (though I don't know how multiples of 10% would be possible then...). Sorry for my rather naive question, but I want to be able to fully understand the wonderful work that you have done here.
Also, Swensen seems to like EM, somewhere between 5-10% being his desirable allocation i think. Would this (presumably) have shifted his curve up?
Thanks again for doing all of this work, and for any answers (tutorial?) that you are able to provide. I am continually impressed by how much knowledge and altruism exists on this site. |
|
| Back to top |
|
 |
Ilovevolleyball

Joined: 11 Jan 2008 Posts: 333
|
Posted: Thu Aug 28, 2008 5:15 pm Post subject: |
|
|
Hello,
Thank You for your wonderful post. I notice that there are no bonds. I thought that even the most aggressive should have some allocation to bonds like 10%. I was under the impression that the ballast provided in down markets would be worth the drag on returns in up markets.
20% Tips.
20% Total Bond Market.
15% International Index Fund or newer FTSE
5% Emer Markets
35% Total Stock Market.
5% Small Value Fund.
That is what I would be curious about. Small Value Tilt and Slight Emerg Markets Tilt but for a Balanced Portfolio.
If you were willing to crunch those numbers that would be wonderful ... If not I understand. But please consider adding so Bonds to future comparisons... nice to see what bonds do.
Thank You,
Mike |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Thu Aug 28, 2008 5:22 pm Post subject: |
|
|
| detifoss wrote: | | As I understand it, the blue dots are representative of 'different' portfolio mixtures. You stated you used a multiple of 10% allocation for each fund, and you sampled one out of every 25 portfolios. |
Correct.
| detifoss wrote: | | Since the 'universe' for this consists of 26 funds, does that mean that one of the blue dots could conceivably represent 100% EM (even though EM is not in any of the represented portfolios) or 100% REITs or 100% CCF? |
Correct, but those extreme portfolios may be sampled out. Generating one unique portfolio one after another is easy. That's just integer calculation. Backtesting one portfolio takes some time because that involves floating point calculation, so I backtest one portfolio and then skip 24 portfolios before backtesting another portfolio.
| detifoss wrote: | | On the other end of the spectrum can a dot represent a fantastically complicated 26 'fund' slice and dice type portfolio (though I don't know how multiples of 10% would be possible then...). Sorry for my rather naive question, but I want to be able to fully understand the wonderful work that you have done here. |
Right. So, a blue dot can have at most 10 funds in it.
| detifoss wrote: | | Also, Swensen seems to like EM, somewhere between 5-10% being his desirable allocation i think. Would this (presumably) have shifted his curve up? |
Yeah, that's a possibility. I'll put that on my todo list. I think his suggestion is something like:
15% Vanguard Developed Markets Index (VDMIX) (0.22%)
5% Vanguard Emerging Mkts Stock Idx (VEIEX) (0.37%)
I don't know why he splits his international allocation. Maybe he is trying to avoid the "Japan problem"? 5% of a portfolio in EM looks normal these days, but I don't know how that would be considered in 1970's. |
|
| Back to top |
|
 |
detifoss
Joined: 13 Oct 2007 Posts: 363
|
Posted: Thu Aug 28, 2008 5:26 pm Post subject: |
|
|
Hi ilovevolleyball
I will admit I don't understand all of the workings of these graphs but from the OP:
Each colored line below represents a portfolio that dilutes a 100% Total Bond Market with some stock portfolio. (OK, I am counting commodities as stocks for now. Please excuse me for that.)
So as I understand it, Bonds are present nearly everywhere (both on the lines and in the dots). Presumably the further to the right one travels on each colored line, the higher the proportion of equities (in the stated portfolio) is present, relative to the amount of bonds present. Or to put it another way, at the furthest left point from which these curves originate, they are all nearly 100% Total Bond Market and very close to 0% of the studied equity portfolio.
If I am wrong please correct me... |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Thu Aug 28, 2008 6:24 pm Post subject: |
|
|
| Ilovevolleyball wrote: | | I notice that there are no bonds. I thought that even the most aggressive should have some allocation to bonds like 10%. I was under the impression that the ballast provided in down markets would be worth the drag on returns in up markets. |
Sorry if I wasn't clear. There are bonds everywhere. Each line represents a 100% Total Bond Market diluted with some stock portfolio. So the most conservative end (the lower left point of the line) represents a 100% Total Bond Market. The most aggressive end (the upper right point of the line) represents one of the stock-only portfolios above. Intermediate points represent some mix between a stock portfolio and Total Bond Market.
| Ilovevolleyball wrote: | 20% Tips.
20% Total Bond Market.
15% International Index Fund or newer FTSE
5% Emer Markets
35% Total Stock Market.
5% Small Value Fund.
That is what I would be curious about. Small Value Tilt and Slight Emerg Markets Tilt but for a Balanced Portfolio.
If you were willing to crunch those numbers that would be wonderful ... If not I understand. But please consider adding so Bonds to future comparisons... nice to see what bonds do. |
See below:
The big grey triangle indicates where you are. The grey line indicates "what if" scenarios of changing your stock/bond allocation.
Note that TIPS is a new asset class that started its life in 1997. Spreadsheet for backtesting (includes TrevH's data) includes synthesized TIPS, but I don't know how good it is. The performance of TIPS is not just the real yield plus inflation. There are flight to quality, its reversion, possible "insurance" premium, etc. In other words, there are market forces behind the overall performance of TIPS. I simply don't know how well the synthesized TIPS takes these things into account. So, as with any backtesting, take this with a grain of salt. |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Thu Aug 28, 2008 6:25 pm Post subject: |
|
|
| detifoss wrote: | So as I understand it, Bonds are present nearly everywhere (both on the lines and in the dots). Presumably the further to the right one travels on each colored line, the higher the proportion of equities (in the stated portfolio) is present, relative to the amount of bonds present. Or to put it another way, at the furthest left point from which these curves originate, they are all nearly 100% Total Bond Market and very close to 0% of the studied equity portfolio.
If I am wrong please correct me... |
You are correct. Bonds are everywhere. |
|
| Back to top |
|
 |
JustAsking
Joined: 25 Aug 2008 Posts: 162
|
Posted: Thu Aug 28, 2008 6:33 pm Post subject: |
|
|
So what this shows, if I interpret it correctly, is that for a given degree of risk, adding VTSMX to Total Bond produced lower returns than the other portfolios in every decade except the 90s? _________________ "In theory, there is no difference between theory and practice, but in practice there is."
-- Jan L.A. van de Snepscheut (1953-1994), late of CalTech |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Thu Aug 28, 2008 6:45 pm Post subject: |
|
|
| JustAsking wrote: | | So what this shows, if I interpret it correctly, is that for a given degree of risk, adding VTSMX to Total Bond produced lower returns than the other portfolios in every decade except the 90s? |
Right, but things are not so simple. If your stock portion has nothing but Total International or something, that doesn't have very good risk adjusted return. You need some domestic stocks, too. |
|
| Back to top |
|
 |
Roger
Joined: 14 Mar 2008 Posts: 23
|
Posted: Thu Aug 28, 2008 6:58 pm Post subject: |
|
|
PiperWarrior,
This post is Awesome, Thanks for sharing !
What are the chances of running :
40% SP 500
30 % Total International.
20 % Reits
10 Emerging markets
For the longest period charted.
I understand if it is too much work. That is my AA - just curious. All vangaurd funds are ok .
Roger |
|
| Back to top |
|
 |
Roger
Joined: 14 Mar 2008 Posts: 23
|
Posted: Thu Aug 28, 2008 7:03 pm Post subject: |
|
|
Oooops -just realized probably can't even do Emerging markets for that time frame.
Piper - really nice post
Roger |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Thu Aug 28, 2008 7:14 pm Post subject: |
|
|
| Roger wrote: | What are the chances of running :
40% SP 500
30 % Total International.
20 % Reits
10 Emerging markets
For the longest period charted. |
The chance is 100%.
| Roger wrote: | | Oooops -just realized probably can't even do Emerging markets for that time frame. |
There is some data for emerging markets. As Rodc says above, I am not sure if EM (or new countries that will be added to EM) will repeat its nice run into the future. Also, people often take the market capitalization into account to some extent. So, I am not sure if people put 10% in Emerging Markets in 1970's.
The data for Total International takes this into account. It starts with pure EAFE index performance, adds a little bit of EM, and then switches to the real fund performance from VGTSX.
| Roger wrote: | | really nice post |
Thanks! |
|
| Back to top |
|
 |
JustAsking
Joined: 25 Aug 2008 Posts: 162
|
Posted: Thu Aug 28, 2008 7:47 pm Post subject: |
|
|
| PiperWarrior wrote: | | JustAsking wrote: | | So what this shows, if I interpret it correctly, is that for a given degree of risk, adding VTSMX to Total Bond produced lower returns than the other portfolios in every decade except the 90s? |
Right, but things are not so simple. If your stock portion has nothing but Total International or something, that doesn't have very good risk adjusted return. You need some domestic stocks, too. |
OK, but for us unsophisticated investors ... what *do* these graphs show (executive summary)? _________________ "In theory, there is no difference between theory and practice, but in practice there is."
-- Jan L.A. van de Snepscheut (1953-1994), late of CalTech |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Thu Aug 28, 2008 8:18 pm Post subject: |
|
|
| JustAsking wrote: | | OK, but for us unsophisticated investors ... what *do* these graphs show (executive summary)? |
Here is my interpretation. A Target Retirement fund-type allocation was fine, but if an investor had discipline to stick to an asset allocation, adding some REITs and small-cap value helped in the past.
All this is in the past. |
|
| Back to top |
|
 |
InvestorJohn
Joined: 03 Aug 2008 Posts: 87
|
Posted: Thu Aug 28, 2008 9:13 pm Post subject: |
|
|
PiperWarrior,
This post is Awesome, Thanks for sharing !
What are the chances of running : All Vangurd
33.5% Total stock market
16.5% FTSE All-World ex-US Inv
25% TIPS
25% Short-Term Treasury Inv |
|
| Back to top |
|
 |
esf
Joined: 09 Apr 2008 Posts: 13
|
Posted: Thu Aug 28, 2008 9:17 pm Post subject: |
|
|
Piper,
thanks for the excellent work. Am I correct in assuming that the returns are adjusted for expenses, but not for taxes?
I would be interested in knowing where these different portfolios would rank if a portion of the funds or all of the funds had to be held in a taxable account. Would small value provide as much of a boost if it had to be held in a taxable account? |
|
| Back to top |
|
 |
purduepete
Joined: 18 Mar 2008 Posts: 82
|
Posted: Thu Aug 28, 2008 9:28 pm Post subject: |
|
|
Is anybody aware of work that slices and dices across constituent sectors of the US market?
Consumer Staples, Energy, Financials, Health Care, IT etc
If an allocation is maintained to historical averages for the sector and then rebalanced it might result in good performance. For example we would be buying Financials now and selling IT in 99-00 etc in such a scheme. |
|
| Back to top |
|
 |
Rodc
Joined: 26 Jun 2007 Posts: 5395
|
Posted: Thu Aug 28, 2008 9:36 pm Post subject: |
|
|
| purduepete wrote: | Is anybody aware of work that slices and dices across constituent sectors of the US market?
Consumer Staples, Energy, Financials, Health Care, IT etc
If an allocation is maintained to historical averages for the sector and then rebalanced it might result in good performance. For example we would be buying Financials now and selling IT in 99-00 etc in such a scheme. |
Just be careful not to include something like buggy whips. Rebalancing into a dying sector will not do wonders for a portfolio. It is a little like having your own little black hole.
Also not sure what you do when a whole new sector arrives. Imagine setting this up right before software/computer went mainstream, and holding to your original plan, thereby missing this sector entirely. _________________ "all standard caveats apply" |
|
| Back to top |
|
 |
Easy Rhino
Joined: 05 Aug 2007 Posts: 1255 Location: San Diego
|
Posted: Thu Aug 28, 2008 9:44 pm Post subject: |
|
|
| Wow, the 90's were almost completely the precise opposite of every other time period. Bizarre. I wonder if the 50's or 60's looked like that (whenver the "nifty 50" was) |
|
| Back to top |
|
 |
dave.d

Joined: 19 Mar 2007 Posts: 807 Location: Richmond, VA
|
Posted: Thu Aug 28, 2008 10:35 pm Post subject: |
|
|
| Quote: | | So what this shows, if I interpret it correctly, is that for a given degree of risk, adding VTSMX to Total Bond produced lower returns than the other portfolios in every decade except the 90s? |
Lest others be confused: U.S. stocks outperformed bonds by varying amounts in all periods shown except 1999-2007. You can see that where the isolated end of each black line off to the right (100% TSM) is higher on the vertical axis than the point where all the lines come together (which is 100% bonds). It's much clearer to say is that for a given degree of risk, portfolios with the equities sliced and diced (using small, value, small value and commodities) outperformed portfolios where the equities consisted only of total markets, over the whole period 1972-2007 and in each of the subperiods shown except 1990-98.
1972 was the absolute height of the "Nifty 50" large growth boom, comparable in some ways to 2000. Won't that start point cause these to overstate to some degree the benefit of slice & dice?
I'm also trying to imagine if I would have had the discipline to stay on the "Efficient" line through 1990-98. I'm afraid not! |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Thu Aug 28, 2008 11:54 pm Post subject: |
|
|
| InvestorJohn wrote: | What are the chances of running : All Vangurd
33.5% Total stock market
16.5% FTSE All-World ex-US Inv
25% TIPS
25% Short-Term Treasury Inv |
The big grey triangle indicates where you are with the grey line indicating "what if" scenarios of changing the stock/bond allocation. Note that TIPS is synthesized one, except for the last 7 years or so. |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Fri Aug 29, 2008 12:07 am Post subject: |
|
|
| esf wrote: | | Am I correct in assuming that the returns are adjusted for expenses, but not for taxes? |
Correct.
| esf wrote: | | I would be interested in knowing where these different portfolios would rank if a portion of the funds or all of the funds had to be held in a taxable account. |
That would be interesting, but having too many knobs makes things harder. Maybe someday.
| esf wrote: | | Would small value provide as much of a boost if it had to be held in a taxable account? |
Probably not. If you have to put small-cap value in a taxable account, you should carefully choose which product to go with. Also, you might want to evaluate whether it makes sense to put munis in a taxable while sheltering small-cap value in a tax-advantaged account.
If you are concerned about tax efficiency of SV, you might want to consider TSM+SV. TSM+SV allows you to easily control exposure to small-cap value depending on how roomy your tax-advantaged space is. |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Fri Aug 29, 2008 12:13 am Post subject: |
|
|
| Rodc wrote: | Just be careful not to include something like buggy whips. Rebalancing into a dying sector will not do wonders for a portfolio. It is a little like having your own little black hole.  |
Yeah. How about a textile industry? Berkshire Hathaway doesn't count. It used to be one though.  |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Fri Aug 29, 2008 12:17 am Post subject: |
|
|
| Easy Rhino wrote: | | Wow, the 90's were almost completely the precise opposite of every other time period. Bizarre. |
Right. That's why any tilt requires discipline. In fact, even having a broad international stock fund may have required discipline in the 90's.
| Easy Rhino wrote: | | I wonder if the 50's or 60's looked like that (whenver the "nifty 50" was) |
You might find Updated figures and results on slice and dice vs TSM helpful for that time period. |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Fri Aug 29, 2008 12:59 am Post subject: |
|
|
| dave.d wrote: | | 1972 was the absolute height of the "Nifty 50" large growth boom, comparable in some ways to 2000. Won't that start point cause these to overstate to some degree the benefit of slice & dice? |
That's certainly possible, but S&D portfolios get its fair share of slap in 90's. Sorry, 1972 is the first year of the data I have.
| dave.d wrote: | | I'm also trying to imagine if I would have had the discipline to stay on the "Efficient" line through 1990-98. I'm afraid not! |
Sure. The years that really required discipline are only 1998 and 1999. Two years feel like nothing if we are looking at the history, but if you are physically in those two years, you may be wondering if the large cap boom will continue forever. As an old saying goes, investing is simple but not easy. |
|
| Back to top |
|
 |
purduepete
Joined: 18 Mar 2008 Posts: 82
|
Posted: Fri Aug 29, 2008 3:03 am Post subject: |
|
|
| Rodc wrote: | | purduepete wrote: | Is anybody aware of work that slices and dices across constituent sectors of the US market?
Consumer Staples, Energy, Financials, Health Care, IT etc
If an allocation is maintained to historical averages for the sector and then rebalanced it might result in good performance. For example we would be buying Financials now and selling IT in 99-00 etc in such a scheme. |
Just be careful not to include something like buggy whips. Rebalancing into a dying sector will not do wonders for a portfolio. It is a little like having your own little black hole.
Also not sure what you do when a whole new sector arrives. Imagine setting this up right before software/computer went mainstream, and holding to your original plan, thereby missing this sector entirely. |
We seem to be able to adapt and shift our allocation to new classes like TIPS, commodities etc when it makes sense. Maybe the same applies here.
If a new sector is getting big that it is showing up in market-cap weighted indexes, then we need some strategy to deal with it. Of course, we can choose to do what Buffett does with the tech sector - i.e. dont bother with it. |
|
| Back to top |
|
 |
Ilovevolleyball

Joined: 11 Jan 2008 Posts: 333
|
Posted: Fri Aug 29, 2008 3:12 am Post subject: |
|
|
PiperWarrior,
Thank You. I am blown away by the work you did for me.
Statistics was not my best course in college -I took it twice !
(I really did )
I barely understand the Bell Curve and Standard Deviation!
On the efficient frontier... I forget which is the x and which is the y axis so I shall use North South East and West....
The Further North the Higher your return. The Further East the more Volatilty and the Further South, the worse your returns.
The best returns are the highest North and the most West, for those the want to avoid volatility.
The absolute best returns would be whatever was the furthest North.
And for the gray line that you charted for me, the meaning of it would be ... these are the possible places your choice of funds would place you. The Triangle is where I am given the 60% 40% portfolio given in prior post of mine.
Anyone please feel free to educate me....
Thank you again for your kind reply PiperWarrior,
Mike |
|
| Back to top |
|
 |
Valuethinker
Joined: 11 May 2007 Posts: 13437
|
Posted: Fri Aug 29, 2008 9:25 am Post subject: |
|
|
You'd be a brave man to pursue a portfolio 10% commodities, 30% SCV, 60% international.
I suspect all kinds of artefacts of the data in creating that frontier ie it is de facto 'data mining' (I am not accusing you or anyone of that! I should stress).
Therefore it is a dangerous strategy, because it relies on what worked well in the past.
even 37 years of data isn't enough. One needs to go back to theory (that risk and return are highly correlated) and to the economic fundamentals of why different asset classes are/ are not correlated.
The Small Cap Value anomaly is there, to be sure, but it cannot be relied upon (it's the sort of thing one can throw 20% of total wealth at, but probably not 50%). Conversely fixed income returns are near certain either in nominal terms (straight bonds) or more importantly real terms (which is all investors should care about) ie TIPS.
I think the job can be done with:
- Vanguard TSM or equivalent
- an international index fund (large cap)
- a US Treasury ST or IT bond fund
- a TIPS fund
If I was adding classes beyond that, then SCV or even International SCV equities. Municipal bonds if insufficient room in tax exempt accounts.
I am agnostic on property (I think the low correlation with other assets is partly a data artefact ie that valuations move with a lag). Commodities I still don't see the source of economic return. |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Fri Aug 29, 2008 9:41 am Post subject: |
|
|
| Ilovevolleyball wrote: | | On the efficient frontier... I forget which is the x and which is the y axis so I shall use North South East and West.... |
The x axis goes East. The Y axis goes North.
| Ilovevolleyball wrote: | | The Further North the Higher your return. The Further East the more Volatilty and the Further South, the worse your returns. |
Correct.
| Ilovevolleyball wrote: | | The best returns are the highest North and the most West, for those the want to avoid volatility. |
Correct.
| Ilovevolleyball wrote: | | The absolute best returns would be whatever was the furthest North. |
Correct.
| Ilovevolleyball wrote: | | And for the gray line that you charted for me, the meaning of it would be ... these are the possible places your choice of funds would place you. The Triangle is where I am given the 60% 40% portfolio given in prior post of mine. |
Correct.
However, note again that all this is in the past. It may be OK to take a small hint or two from this, but I wouldn't whatever the most efficient and implement it in my portfolio. |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Fri Aug 29, 2008 10:21 am Post subject: |
|
|
| Valuethinker wrote: | | You'd be a brave man to pursue a portfolio 10% commodities, 30% SCV, 60% international. |
No, I wouldn't do that personally. I don't mean to pick on you, but you flipped 30% and 60%. The Efficient line above is:
10% PIMCO CommodityRealRet Strat Instl (PCRIX) (0.75%)
60% Vanguard Small Cap Value Index (VISVX) (0.22%)
30% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
which requires you to be even more brave.
| Valuethinker wrote: | I suspect all kinds of artefacts of the data in creating that frontier ie it is de facto 'data mining' (I am not accusing you or anyone of that! I should stress).
Therefore it is a dangerous strategy, because it relies on what worked well in the past. |
Exactly.
| Valuethinker wrote: | I think the job can be done with:
- Vanguard TSM or equivalent
- an international index fund (large cap)
- a US Treasury ST or IT bond fund
- a TIPS fund
If I was adding classes beyond that, then SCV or even International SCV equities. Municipal bonds if insufficient room in tax exempt accounts. |
That's exactly where I am except that I have a little bit of domestic REITs and domestic small-cap value.
| Valuethinker wrote: | | I am agnostic on property (I think the low correlation with other assets is partly a data artefact ie that valuations move with a lag). Commodities I still don't see the source of economic return. |
I've seen a similar comment from you about the lag. I think you've mentioned something to do with accounting? Could you elaborate a little bit? |
|
| Back to top |
|
 |
stratton

Joined: 04 Mar 2007 Posts: 8219 Location: Puget Sound
|
Posted: Fri Aug 29, 2008 10:29 am Post subject: |
|
|
| PiperWarrior wrote: | 10% PIMCO CommodityRealRet Strat Instl (PCRIX) (0.75%)
60% Vanguard Small Cap Value Index (VISVX) (0.22%)
30% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
which requires you to be even more brave.  |
Unless you do a little throttling back by adding bonds such as:
02% PIMCO CommodityRealRet Strat Instl (PCRIX) (0.75%)
12% Vanguard Small Cap Value Index (VISVX) (0.22%)
06% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
+
80% Some kind of bonds.
I'd still feel more comfortable with Larry Swedroe's version/ratios with a larger percentage in international and EM:
10.5% US SV
10.5% Intl SV
4% EM Value
3% Commodites
72% bonds
I'd also be tempted to toss in some REITs too.
Paul |
|
| Back to top |
|
 |
Valuethinker
Joined: 11 May 2007 Posts: 13437
|
Posted: Fri Aug 29, 2008 11:12 am Post subject: |
|
|
| PiperWarrior wrote: | | Valuethinker wrote: | | You'd be a brave man to pursue a portfolio 10% commodities, 30% SCV, 60% international. |
No, I wouldn't do that personally. I don't mean to pick on you, but you flipped 30% and 60%. The Efficient line above is:
10% PIMCO CommodityRealRet Strat Instl (PCRIX) (0.75%)
60% Vanguard Small Cap Value Index (VISVX) (0.22%)
30% Vanguard Total Intl Stock Index (VGTSX) (0.27%)
which requires you to be even more brave.
|
Mea culpa.
| Quote: |
| Valuethinker wrote: | | I am agnostic on property (I think the low correlation with other assets is partly a data artefact ie that valuations move with a lag). Commodities I still don't see the source of economic return. |
I've seen a similar comment from you about the lag. I think you've mentioned something to do with accounting? Could you elaborate a little bit? |
For any 'private asset' like private equity or real estate, valuations as reported are lagging: buildings or assets are only valued periodically, and that valuation reflects market history rather than prices which are reflective of current market values.
That is to say, what an external valuer does is look at buildings that have been bought and sold with similar characteristics (size, location, lease terms) in the recent past, as a basis for valuing the building (also looks at the lease, rental growth etc.). So it's a backward looking valuation.
Over to a property fund like TIAA RE. The assets in the fund are valued only periodically, so you are buying into those assets at 'stale prices' ie historic valuations. That means in a falling market, you are buying fund units which reflect values which are too high.
Now onto REITs. REIT the market adjusts for this, by pricing its expectations as a premium or discount on assessed NAV. Logically the NAV should eventually catch up. In principle, the REIT market is more efficient (via the transmission mechanism of the discount/ premium to NAV).
However if the value of underlying assets is uncertain (because there are very few comparable transactions going on, eg the credit crunch when big buildings are not being bought and sold) then the problem is still there.
A friend of mine who has looked into this professionally says that much of the apparent lack of correlation between RE as an asset class and equities is a statistical artefact: stock prices are here and now, fully reflecting the market's view of future prospects, and units in private RE funds are historically priced.
The same economic forces and forces of liquidity that cause booms and busts in the stock market, feed through to the RE market over a longer cycle. The instantaneous correlation is low, but the longer run correlation is much higher. |
|
| Back to top |
|
 |
dbr
Joined: 04 Mar 2007 Posts: 5414
|
Posted: Fri Aug 29, 2008 12:28 pm Post subject: |
|
|
| Valuethinker wrote: |
The Small Cap Value anomaly is there, to be sure, but it cannot be relied upon (it's the sort of thing one can throw 20% of total wealth at, but probably not 50%). Conversely fixed income returns are near certain either in nominal terms (straight bonds) or more importantly real terms (which is all investors should care about) ie TIPS. |
One might add that one should throw at least 20% that way if one is going to play that game. Similarly, do these sometimes seen 5% here, 5% there allocations (commodities for example) really make enough difference to worry about? Note the OP planned the exercise with no less than 10% or none. |
|
| Back to top |
|
 |
stratton

Joined: 04 Mar 2007 Posts: 8219 Location: Puget Sound
|
Posted: Fri Aug 29, 2008 12:49 pm Post subject: |
|
|
| dbr wrote: | | Valuethinker wrote: |
The Small Cap Value anomaly is there, to be sure, but it cannot be relied upon (it's the sort of thing one can throw 20% of total wealth at, but probably not 50%). Conversely fixed income returns are near certain either in nominal terms (straight bonds) or more importantly real terms (which is all investors should care about) ie TIPS. |
One might add that one should throw at least 20% that way if one is going to play that game. Similarly, do these sometimes seen 5% here, 5% there allocations (commodities for example) really make enough difference to worry about? Note the OP planned the exercise with no less than 10% or none. |
Larry S. generally suggests 10% of equities. Since the OPs planned exercise is 100% equities this fits in with that theme.
Paul |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Fri Aug 29, 2008 1:24 pm Post subject: |
|
|
| dbr wrote: | | Note the OP planned the exercise with no less than 10% or none. |
Well, the reason is more of computational resources than anything else. The smaller slice I use, the number of combinations explodes. Here are the numbers of combinations over 26 funds with various slice sizes:
20% slices: 30!/25!/5! = 142,506
10% slices: 36!/25!/10! = 6,608,858,256
5% slices: 45!/25!/20! = 3,169,870,830,126
where N! represents a factorial.
I could have used 5% slices, but I would have to sample more aggressively to get the results in a reasonable amount of time. |
|
| Back to top |
|
 |
dbr
Joined: 04 Mar 2007 Posts: 5414
|
Posted: Fri Aug 29, 2008 1:40 pm Post subject: |
|
|
| PiperWarrior wrote: | | dbr wrote: | | Note the OP planned the exercise with no less than 10% or none. |
Well, the reason is more of computational resources than anything else. The smaller slice I use, the number of combinations explodes. Here are the numbers of combinations over 26 funds with various slice sizes:
20% slices: 30!/25!/5! = 142,506
10% slices: 36!/25!/10! = 6,608,858,256
5% slices: 45!/25!/20! = 3,169,870,830,126
where N! represents a factorial.
I could have used 5% slices, but I would have to sample more aggressively to get the results in a reasonable amount of time. |
As an exercise it would be information regarding the sensitivity of the risk adjusted reward to addition of these smaller slices. Fully understand that is not what the original purpose was and the time consumed to obtain this kind of information. |
|
| Back to top |
|
 |
Alex Frakt Site Admin
Joined: 23 Feb 2007 Posts: 5031 Location: Chicago
|
Posted: Fri Aug 29, 2008 3:01 pm Post subject: |
|
|
If you are taking requests
Microcaps. You can use CRSP 10 until BRSIX appears.
Who am I kidding, I just want to see my portfolio. I already took out the EM. If you don't have the international breakdown, let's go with 40% EAFE.
30% VFINX (VTSMX would be fine)
15% Midcap Value
15% Microcap
15% Pacific
15% Europe
10% Int'l Value |
|
| Back to top |
|
 |
carlsworld
Joined: 20 Dec 2007 Posts: 19 Location: Raleigh, NC
|
Posted: Fri Aug 29, 2008 3:16 pm Post subject: |
|
|
| A lot of us use market weights. It would be interesting to see a graph of that. Vanguard Total Stock Index and Vanguard Total Intl Index, never rebalancing between these two. This would probably be trickier to do than the other combinations listed because your software is probably designed to rebalance back to set percentages. |
|
| Back to top |
|
 |
stratton

Joined: 04 Mar 2007 Posts: 8219 Location: Puget Sound
|
Posted: Fri Aug 29, 2008 3:32 pm Post subject: |
|
|
| carlsworld wrote: | | A lot of us use market weights. It would be interesting to see a graph of that. Vanguard Total Stock Index and Vanguard Total Intl Index, never rebalancing between these two. This would probably be trickier to do than the other combinations listed because your software is probably designed to rebalance back to set percentages. |
Considering Total Intl is lacking small cap... It's not the same as a US TSM. To make it equivalent ~15% of the international allocation needs to be small cap.
Paul |
|
| Back to top |
|
 |
caklim00
Joined: 26 May 2008 Posts: 1210
|
Posted: Fri Aug 29, 2008 3:35 pm Post subject: |
|
|
Piper, you rock!
This may be too complicated to run, but if it is possible I would love to see it as it is going to be my equity portion of my portfolio...
27% TSM (VTI)
18% SCV (VBR)
5% CRSP 10 (BRSIX)
5% US REIT (VNQ)
27% Tot Int (VEU)
18% Int Small Cap (I am using DLS, GWX, and DGS equally --> any sort of international small cap estimation would be cool) |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Fri Aug 29, 2008 3:40 pm Post subject: |
|
|
| caklim00 wrote: | | 18% Int Small Cap (I am using DLS, GWX, and DGS equally |
Sorry, I don't have data for international small cap. Would you mind if I substitute Total International for it? |
|
| Back to top |
|
 |
caklim00
Joined: 26 May 2008 Posts: 1210
|
Posted: Fri Aug 29, 2008 3:57 pm Post subject: |
|
|
| PiperWarrior wrote: | | caklim00 wrote: | | 18% Int Small Cap (I am using DLS, GWX, and DGS equally |
Sorry, I don't have data for international small cap. Would you mind if I substitute Total International for it? | Sure, that would be fine. Thanks in advance! |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Sat Aug 30, 2008 1:56 am Post subject: |
|
|
| Alex Frakt wrote: | Microcaps. You can use CRSP 10 until BRSIX appears.
Who am I kidding, I just want to see my portfolio. I already took out the EM. If you don't have the international breakdown, let's go with 40% EAFE.
30% VFINX (VTSMX would be fine)
15% Midcap Value
15% Microcap
15% Pacific
15% Europe
10% Int'l Value |
OK. I don't have mid-cap value, so I used a combination of large-cap value and small-cap value.
30% Vanguard 500 Index (VFINX) (0.15%)
10% Vanguard Value Index (VIVAX) (0.20%)
5% Vanguard Small Cap Value Index (VISVX) (0.22%)
15% Bridgeway Ultra-Small Company Market (BRSIX) (0.65%)
15% Vanguard European Stock Index (VEURX) (0.22%)
15% Vanguard Pacific Stock Index (VPACX) (0.22%)
10% Vanguard International Value (VTRIX) (0.43%)
 |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Sat Aug 30, 2008 2:03 am Post subject: |
|
|
| caklim00 wrote: | This may be too complicated to run, but if it is possible I would love to see it as it is going to be my equity portion of my portfolio...
27% TSM (VTI)
18% SCV (VBR)
5% CRSP 10 (BRSIX)
5% US REIT (VNQ)
27% Tot Int (VEU)
18% Int Small Cap (I am using DLS, GWX, and DGS equally --> any sort of international small cap estimation would be cool) |
I used:
Vanguard Total Stock Mkt Idx (VTSMX) (0.15%)
Vanguard Small Cap Value Index (VISVX) (0.22%)
Bridgeway Ultra-Small Company Market (BRSIX) (0.65%)
Vanguard REIT Index (VGSIX) (0.20%)
Vanguard Total Intl Stock Index (VGTSX) (0.27%)
 |
|
| Back to top |
|
 |
PiperWarrior

Joined: 21 Dec 2007 Posts: 4068 Location: right on course
|
Posted: Sat Aug 30, 2008 2:32 am Post subject: |
|
|
| carlsworld wrote: | | A lot of us use market weights. It would be interesting to see a graph of that. Vanguard Total Stock Index and Vanguard Total Intl Index, never rebalancing between these two. This would probably be trickier to do than the other combinations listed because your software is probably designed to rebalance back to set percentages. |
As a lousy estimate, I started off with domestic/international=40%/60% and went backwards in time to estimate the market capitalization in 1972. In other words, I used the following ratio as the starting point.
Domestic: 40 / (1.1096^36)
International: 60 / (1.1227^36)
The ratio comes out to be about 50:50. (If anyone knows the correct ratio in 1970's, please let me know.)
Basically, I let my program synthesize "Total World Stock Market" first from VTSMX and VGTSX and then did the usual stuff.
So "VTWSX" ended up being a slightly more volatile version of Target Retirement. |
|
| Back to top |
|
 |
JustAsking
Joined: 25 Aug 2008 Posts: 162
|
Posted: Sat Aug 30, 2008 7:02 am Post subject: |
|
|
| Valuethinker wrote: | For any 'private asset' like private equity or real estate, valuations as reported are lagging: buildings or assets are only valued periodically, and that valuation reflects market history rather than prices which are reflective of current market values.
That is to say, what an external valuer does is look at buildings that have been bought and sold with similar characteristics (size, location, lease terms) in the recent past, as a basis for valuing the building (also looks at the lease, rental growth etc.). So it's a backward looking valuation.
Over to a property fund like TIAA RE. The assets in the fund are valued only periodically, so you are buying into those assets at 'stale prices' ie historic valuations. That means in a falling market, you are buying fund units which reflect values which are too high.
Now onto REITs. REIT the market adjusts for this, by pricing its expectations as a premium or discount on assessed NAV. Logically the NAV should eventually catch up. In principle, the REIT market is more efficient (via the transmission mechanism of the discount/ premium to NAV).
However if the value of underlying assets is uncertain (because there are very few comparable transactions going on, eg the credit crunch when big buildings are not being bought and sold) then the problem is still there.
A friend of mine who has looked into this professionally says that much of the apparent lack of correlation between RE as an asset class and equities is a statistical artefact: stock prices are here and now, fully reflecting the market's view of future prospects, and units in private RE funds are historically priced.
The same economic forces and forces of liquidity that cause booms and busts in the stock market, feed through to the RE market over a longer cycle. The instantaneous correlation is low, but the longer run correlation is much higher. |
High enough to justify eliminating REITs from a portfolio? _________________ "In theory, there is no difference between theory and practice, but in practice there is."
-- Jan L.A. van de Snepscheut (1953-1994), late of CalTech |
|
| Back to top |
|
 |
|