Spreadsheet for backtesting (includes TrevH's data)

[edge]

edge,

I updated the links. Can you try it again?

Works fine!

[edge]

Whoops, sorry for the double post.

Thanks again Simba!

[serbeer]

Simba,

Dec. 2007 CPU-I data were released today:
http://www.bls.gov/news.release/cpi.nr0.htm
and I calculated that with Dec index level being 210.036
2007, 2007 Average is 207.342 vs 2006 Average 201.6, thus making official inflation level of 2007 2.85%

also, Morningstar shows BRSIX 2007 return of -5.4 and PCRIX's return of 23.8

So now you can complete backtesting tool for entire 2007--I just did so on my copy.

Thank you for the wonderful tool to play with!

Serbeer

[simba]

Thanks serbeer.

I've updated the spreadsheet and for those interested, You can download the Excel Spreadsheet [rev7c] and OpenOffice version [rev7c]

Rev 7c has updated information for the Commodities returns.

Based on the work done by Jeff (Cb/Gnobility) at http://gnobility.com/Syn_Comm/CCF_1972-2006_san.xls
He calculated the returns as follows:

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

Best Regards,
Simba

[schwarm]

Based on the work done by Jeff (Cb/Gnobility) at http://gnobility.com/Syn_Comm/CCF_1972-2006_san.xls
He calculated the returns as follows:

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

Best Regards,
Simba

Simba,

If you were using index results (pre-PCRIX) would it make sense to take off .75% to estimate PCRIX results? The further back you go, the more sketchy the comparison to PCRIX, but it seems worthwhile to estimate the investable fund's earlier performance.

[Cb]

If you were using index results (pre-PCRIX) would it make sense to take off .75% to estimate PCRIX results? The further back you go, the more sketchy the comparison to PCRIX, but it seems worthwhile to estimate the investable fund's earlier performance.

Schwarm,

Here's a link to the updated Excel spreadsheet that shows how I constructed CCF returns prior to the inception of PCRIX.

http://gnobility.com/ER/CCF_1972-2007_san.xls

I included the 0.74% expense ratio (or it's 'compliment' when combining ^DJC with the Vanguard TIPS fund result which already carry a .20% ER)

I think this is about as good as we can do to assemble a dataset similar to PCRIX. A lot of the credit really shoud go to raddr.

Cb

[schwarm]

If you were using index results (pre-PCRIX) would it make sense to take off .75% to estimate PCRIX results? The further back you go, the more sketchy the comparison to PCRIX, but it seems worthwhile to estimate the investable fund's earlier performance.

Schwarm,

Here's a link to the updated Excel spreadsheet that shows how I constructed CCF returns prior to the inception of PCRIX.

http://gnobility.com/ER/CCF_1972-2007_san.xls

I included the 0.74% expense ratio (or it's 'compliment' when combining ^DJC with the Vanguard TIPS fund result which already carry a .20% ER)

I think this is about as good as we can do to assemble a dataset similar to PCRIX. A lot of the credit really shoud go to raddr.

Cb

So PCRIX level expenses are included in pre-2003 estimates.

Thanks.

[Cb]

If you were using index results (pre-PCRIX) would it make sense to take off .75% to estimate PCRIX results? The further back you go, the more sketchy the comparison to PCRIX, but it seems worthwhile to estimate the investable fund's earlier performance.

Schwarm,

Here's a link to the updated Excel spreadsheet that shows how I constructed CCF returns prior to the inception of PCRIX.

http://gnobility.com/ER/CCF_1972-2007_san.xls

I included the 0.74% expense ratio (or it's 'compliment' when combining ^DJC with the Vanguard TIPS fund result which already carry a .20% ER)

I think this is about as good as we can do to assemble a dataset similar to PCRIX. A lot of the credit really shoud go to raddr.

Cb

So PCRIX level expenses are included in pre-2003 estimates.

Thanks.

Yeah. In fact, I think a nice enhancement to Simba Spreadsheet would be to also subtract Vanguard ER's from the raw index numbers...allowing the user to overide default Investor level ER's with Admiral or ETF ER's.

I figured the 0.74% that PCRIX carries is just too stiff to ignore (though PIMCO appears to be doing a terrific job managing the fund. Try comparing Yahoo monthly returns for PCRIX to ^DJC plus VIPSX sometime. You don't see 0.74% tracking error.)

So far.

Cb

[serbeer]

Simba,
I was looking into the way you calculate Sortino Ratio and noticed that in the version 07c of the spreadsheet, there are problems with encompassing 2007 data in some formulas.

For example, Sortino Ratio calculations in Returns_72_07!F105 refer to
AH$62:AH$96 range, while 2007 portfolio return is in AH97. You probably should have used AH$62:AH$102 range to allow for expansion.

Same with both nominal and real data of STDDev, Sharpe ratio, C-US, C-Intl, etc formulas, and with other tabs as well...


Also, a question:
If I were trying to calculate Sortino for DRO, I'd probably be forced to make Target Return another parm on Porfolio spreadsheet, so that the user can calculate Sortino based on that...

You seem to be using Average ofr all returns for target return on Sortino ratio. I assume you and perhaps Gummy if he was helping with it had your reasoning, wander what it was. Sortino was supposed to be Target Return-specific...

Thanks,
Serbeer

[simba]

Serber - curiousinvestor informed me of the errors a couple of days ago. I have uploaded the revised version (rev7d).

Those interested can download the Excel Spreadsheet [rev7d] and OpenOffice version [rev7d]

As far as the sortino ratio is concerned, I am not sure I follow you. I am using the TBill returns as the Riskfree return.

Cb - I do mean to post a topic to see what ER's I should be deducting from the raw Index Fund returns.


Regards,
Simba

[RobertH]

Simba,

You asked for clarification on the issue with the Sortino ratios. Check out this thread, where Sortino ratios were calculated using your spreadsheet. The ratios for Gibson's Medium Return portfolios don't look right to me.

And thanks for the great work on this spreadsheet.

Robert

P.S. You might consider adding the unhedged global bond returns shown in the thread linked to above, for everyone's benefit.

[simba]

Simba,

You asked for clarification on the issue with the Sortino ratios. Check out this thread, where Sortino ratios were calculated using your spreadsheet. The ratios for Gibson's Medium Return portfolios don't look right to me.

And thanks for the great work on this spreadsheet.

Robert

P.S. You might consider adding the unhedged global bond returns shown in the thread linked to above, for everyone's benefit.

Robert - I uploaded a new version rev-7e of the spreadsheet. I added the unhedged global bond returns as well. Thanks to Stratton for providing the returns.

You can download the Excel Spreadsheet [rev7e] and OpenOffice version [rev7e]

The biggest change is earlier I was using the average of T-Bill returns as the MAR(minimum acceptable returns) to calculate Sortino Ratios. In the latest version, it is a modifiable parameter. Also provided a way to deduct Expense Ratios (ER's can be modified) from synthetic/benchmark/index returns.

Disclaimer/Caution/Caveat Emptor: Use this spreadsheet for Entertainment purposes. Please do not change your asset allocation purely based on this spreadsheet/historical performance. This was a fun project and please treat it as such.

Best Regards,
Simba

[stratton]

I'm getting errors trying to download this. "Requested header could not be found." So the file is probably:

1. Not there.
2. You have it open and its locked.
3. We don't have privs to get it.
4. File name/location not correct in url

Paul

[simba]

Paul/Cb,

I corrected the links. Try downloading it again.

Thanks

-Simba

[Cb]

Thanks Simba - got it this time!

I know adding the option to subtract appropriate ER's from the pre-fund inception data must have taken quite a bit of effort.

Cb 8)

[MossySF]

The formula in the "Compare 5 portfolios" section appear to be slightly different from the main section. For example:

1972-2007 TSM = 10.96/17.24
Compare 5 TSM = 11.08/17.25

1985-2007 TSM = 12.22/16.03
Compare 5 TSM = 12.29/15.69

[simba]

Thanks Simba - got it this time!

I know adding the option to subtract appropriate ER's from the pre-fund inception data must have taken quite a bit of effort.

Cb 8)

Thanks Cb - it did take a bit.

The formula in the "Compare 5 portfolios" section appear to be slightly different from the main section. For example:

1972-2007 TSM = 10.96/17.24
Compare 5 TSM = 11.08/17.25

1985-2007 TSM = 12.22/16.03
Compare 5 TSM = 12.29/15.69

The comparison section does not deduct the ER's. Hence the difference.

Back to the drawing board.

I am thinking it may be easier to just update the data and deduct the ERs from the index returns.

The next question would - how much should I deduct? Obviously the ERs back in 70's and 80's were higher. The current setup assumes the same ER. If we modify the index returns data then we can deduct different ERs.

What say ye?

Best Regards,
Simba

[Cb]

Simba - regarding how much to deduct - we're probably splitting hairs, but if you're going to eliminate the option to specify an ER I think maybe subtracting current Vanguard Investor ER's would be appropriate... transaction costs were higher in the past, but current Admiral Funds or ETF's are a tad lower...

Cb

[etc06]

For the Sharpe calculation when comparing 5 portfolios for 1972-2007, the divisor is the average (row 197) instead of STDEV (row 198) like you have for 1985 - 2007.

Thanks for making this spreadsheet, it's fun to see how portfolios would've done.

[Willy]

Thanks to all who contributed input and, especially, to Simba for this great tool!!

[jmFightSpam]

Hi,

Simba, TrevH, or to whom it may concern:

Where did you get the data from to populate the spreadsheet?

Thanks.

[financialguy]

Hi,

Simba, TrevH, or to whom it may concern:

Where did you get the data from to populate the spreadsheet?

Thanks.

There is a data sources tab in the spreadsheet that explains for each asset class.

[jmFightSpam]

Hi,

Simba, TrevH, or to whom it may concern:

Where did you get the data from to populate the spreadsheet?

Thanks.

There is a data sources tab in the spreadsheet that explains for each asset class.

Thank you financialguy and simba (who sent me a PM).

It looks like the link to

http://www.tamasset.com/other/AC2705.xls

is not working anymore. Do you know where I can find this spreadsheet?

Thanks.

[nathank]

This thing is AMAZING! Thanks

[Webfoot]

What does nominal and real mean or MAR?

This is a truly amazing piece of work.

[mikenz]

Real returns are returns adjusted for inflation (return after inflation). Nominal is the unadjusted return.

No idea about MAR. Median Annualized Return? March?

[simba]

nathank/webfoot - Thanks for the compliments.

As I've said numerous times, use this ONLY for entertainment purposes. Do not alter your AA based on historical returns.
As Paul says - "Don't chase the ghosts of past returns"

Having said that the I've updated the spreadsheet.

Those interested can download the Excel Spreadsheet [rev7f] or the OpenOffice version [rev7f]

The returns for the single portfolio should now match the returns from the comparison section.

Best Regards,
Simba

[simba]

What does nominal and real mean or MAR?

This is a truly amazing piece of work.

Webfoot - MAR (Minimum acceptable Return) [in the spreadsheet] is used in calculating Sortino Ratio. which measures the risk-adjusted returns of a portfolio.

The Sortino ratio is similar to the Sharpe ratio, except it uses downside deviation for the denominator instead of standard deviation.

Sortino Ratio S = {R-T}/{DR}
where
R=portfolio return
T = target or required rate of return (T was originally known as the minimum acceptable return, or MAR)
DR=downside risk.

In the earlier revisions of the spreadsheets, I was using the average of T-Bills for calculating the Sortino Ratio. But decided to use a "modifiable" parameter so one can change the target ROR (rate of return).

Regards,
Simba

[mbrasher1]

What a great tool! Just for fun, I tried to see what the highest Sharpe ratio I could get on the 1972-2007 data. I found that a portfolio of:

30% SCV
20% EM
10% REIT
40% Commodities

Produces a Sharpe ratio of .82

A high allocation to commodities really bumps up the Sharpe ratio of a portfolio.

[Random Musings]

mbrasher1 wrote:

What a great tool! Just for fun, I tried to see what the highest Sharpe ratio I could get on the 1972-2007 data. I found that a portfolio of:

30% SCV
20% EM
10% REIT
40% Commodities

Produces a Sharpe ratio of .82

A high allocation to commodities really bumps up the Sharpe ratio of a portfolio.

Fun to play, but past performance is just that, in the past. With commodities being such a hot topic, I expect disappointment in this arena coming up relatively soon (next few months, at most). If one would run this model 1972-2000, commodities allocation would probably be more muted.

RM

[Cb]

mbrasher1 wrote:

What a great tool! Just for fun, I tried to see what the highest Sharpe ratio I could get on the 1972-2007 data. I found that a portfolio of:

30% SCV
20% EM
10% REIT
40% Commodities

Produces a Sharpe ratio of .82

A high allocation to commodities really bumps up the Sharpe ratio of a portfolio.

Fun to play, but past performance is just that, in the past. With commodities being such a hot topic, I expect disappointment in this arena coming up relatively soon (next few months, at most). If one would run this model 1972-2000, commodities allocation would probably be more muted.
RM

...as would the SCV & EM...

[james22]

Something I've found interesting through backtesting is that combining the *riskiest* stocks with the *safest* bonds produces some really neat results

Yup, I agree. One of my favorite asset allocations is something along the lines of 70% TIPs and 30% international small-company stocks (better yet, international emerging small-company stocks). Huge tracking error relative to any common benchmark (obviously), but it has a very interesting risk-return profile.

- DDB

+1

[DP]

Hi,
Thanks so much for posting this spreadsheet! Very helpful.

I came in with respect for the Permanent Portfolio (search these forums for separate discussion on this) but thinking it was a little conservative and perhaps a little out of balance. I was thinking that a portfolio equally balanced among diverse (as much as possible) asset classes would outperform. The spreadsheet confirms: equal allocations to US Stocks, Int''l Stocks, REIT, Commodity, LT US Bonds, Global Bonds does very well, outperforming well known allocations such as Coffeehouse, Permanent Portfolio, and others. Following this a logical breakdown for US stocks would be Small Cap Value and LC Growth (arguably the best performing sub-class and the most uncorrelated subclass), and a similar breakdown for Int'l stocks would be Emerging Markets and Int'l Value. This yields a portfolio with a sortino of nearly 10 since 1985 and substituting Tips for Global bonds, a sortino of 1.81/sharpe .79 since 72.

One other comment, I have seen disparaging remarks on the outlook for commodities and rarely see this incorporated into a portfolio. Being virtually the only asset class with negative correlation and yet strong historical returns it would seem essential to include this in any balanced portfolio, and to exclude it based on expectations of future performance seems to go against the whole concept - creating a balanced portfolio precisely because it is so difficult to predict what asset class will outperform in the future.

Don

[edge]

Hi,

Do not fall victim to overanalysis of past performance and shifting correlations. I don't think commodities have performed well except in the very recent past. Their historically low/negative correlation may be useful and it does make sense in many ways.

If you include them, they should be part of portfolio insurance against a unique set of risks. The estimated real return from commodities is very low. Do not expect the current trend of performance to continue. That would similar to expecting housing ni 2006 to continue its extraordinary above historical trend return path.

Hi,
Thanks so much for posting this spreadsheet! Very helpful.

I came in with respect for the Permanent Portfolio (search these forums for separate discussion on this) but thinking it was a little conservative and perhaps a little out of balance. I was thinking that a portfolio equally balanced among diverse (as much as possible) asset classes would outperform. The spreadsheet confirms: equal allocations to US Stocks, Int''l Stocks, REIT, Commodity, LT US Bonds, Global Bonds does very well, outperforming well known allocations such as Coffeehouse, Permanent Portfolio, and others. Following this a logical breakdown for US stocks would be Small Cap Value and LC Growth (arguably the best performing sub-class and the most uncorrelated subclass), and a similar breakdown for Int'l stocks would be Emerging Markets and Int'l Value. This yields a portfolio with a sortino of nearly 10 since 1985 and substituting Tips for Global bonds, a sortino of 1.81/sharpe .79 since 72.

One other comment, I have seen disparaging remarks on the outlook for commodities and rarely see this incorporated into a portfolio. Being virtually the only asset class with negative correlation and yet strong historical returns it would seem essential to include this in any balanced portfolio, and to exclude it based on expectations of future performance seems to go against the whole concept - creating a balanced portfolio precisely because it is so difficult to predict what asset class will outperform in the future.

Don

[DP]

Hi,

I don't think commodities have performed well except in the very recent past. .... The estimated real return from commodities is very low.

I don't see this. From the spreadsheet, average annual return since 72 and average annual return by decade:

PCRIX avg: 13.2, 70's: 21.8, 80's: 6.71, 90's: 8.4, 2000's: 18.0
REIT avg: 14.1, 70's: 18.2, 80's: 11.5, 90's: 16.4, 2000's: 16.7
VTSMX avg: 12.4, 70's: 7.7, 80's: 17.1, 90's: 18.3, 2000's: 3.5
CPU avg: 4.7, 70's: 7.6, 80's: 5.6, 90's: 3, 2000's: 2.8

Based on this data, commodity outperformance has not been unique nor has the REIT outperformance, and both appear to have produced positive real returns by decade higher than stocks. What am I missing?

Do not expect the current trend of performance to continue.

I made that mistake in the late 1990's. I may have mostly missed the bear but I also missed a lot of gains also. I can't say that I have no expectations regarding the future but now I do my best to invest that way.

Don

[zhiwiller]

Sorry for resurrecting an ancient thread, but has anyone updated Simba's spreadsheet with 2008's returns? It should make for some interesting backtests.

[stratton]

Sorry for resurrecting an ancient thread, but has anyone updated Simba's spreadsheet with 2008's returns? It should make for some interesting backtests.

I did a quick look, but I think its "locked" in certain rows. It might help if I read the instructions page.

Paul

[grok87]

Sorry for resurrecting an ancient thread, but has anyone updated Simba's spreadsheet with 2008's returns? It should make for some interesting backtests.

I did a quick look, but I think its "locked" in certain rows. It might help if I read the instructions page.

Paul

I guess I sort of assumed Simba would be doing it- anyone know what happened to him. Looks like his last post was in May '08- did he just drift away or what...
Let me know if you need any help with any of the data...
cheers,

[Paul Douglas Boyer]

I have attempted to make this spreadsheet web-enabled by importing it into Google Spreadsheets. It appears to work. I did detect some bugs in the formulas that I tried to fix.

I have added data for 2008 from https://flagship.vanguard.com/VGApp/hnw/FundsByType.

I also added Gold so that we can try calculating the Harry Browne Permanent Portfolio. Not sure if I picked the right bond funds.

Here's the link:

https://spreadsheets.google.com/ccc?key=pOjc3ot10vgs0eml-DJZKcw&newcopy

I believe you need a Google account to open a Google Docs spreadsheet. This link will open the template as a new document that you own. This means that you will not get updates automatically. That's too bad.

Let me know if it works for you. Let me know of better ideas.

[grok87]

I have attempted to make this spreadsheet web-enabled by importing it into Google Spreadsheets. It appears to work. I did detect some bugs in the formulas that I tried to fix.

I have added data for 2008 from https://flagship.vanguard.com/VGApp/hnw/FundsByType.

I also added Gold so that we can try calculating the Harry Browne Permanent Portfolio. Not sure if I picked the right bond funds.

Here's the link:

https://spreadsheets.google.com/ccc?key=pOjc3ot10vgs0eml-DJZKcw&newcopy

I believe you need a Google account to open a Google Docs spreadsheet. This link will open the template as a new document that you own. This means that you will not get updates automatically. That's too bad.

Let me know if it works for you. Let me know of better ideas.

THanks Paul! You're the best...
I'll create a google account and check it out.
cheers,

[Paul Douglas Boyer]

Also note that http://www.icarra.com gives you the ability to input your portfolio and calculate returns, standard deviations, and Sharpe ratios.

And you can input any funds, stocks, whatever. I don't think it has a good proxy for Gold, however. GLD only goes back a couple of years.

Have fun. And remember, history doesn't predict.

[DP]

Hi,

Let me know if it works for you. Let me know of better ideas.

It works for me. Thanks so much for updating the spreadsheet! And Google Docs was a great idea! I've never used it before but it is very easy to use, at least for the basics ... which is all I know anyways.

Don

[Cb]

Paul,

Have you got an Excel version of the spreadsheet?

If so, I could host it...

thanks,

Cb

[DaleMaley]

Paul,

Have you got an Excel version of the spreadsheet?

If so, I could host it...

thanks,

Cb

I tried to export the google docs spreadsheet to an excel file and download it, but looks like formulas are messed up. Can you post the Excel version?

I am not a big fan of Google Docs........the site was broken for about a 4 week period over Christmas...I guess there is a reason they call it a Beta version.

[Cb]

Paul,

Have you got an Excel version of the spreadsheet?

If so, I could host it...

thanks,

Cb

Can you post the Excel version?

Yep...see your PM's...

Cb

[simba]

Sorry Folks, I have been out of pocket for sometime.

I've updated the spreadsheet and for those interested, You can download the Excel Spreadsheet [rev7h] and OpenOffice version [rev7h]


This includes the 2008 returns and some minor updates.

Best Regards,
Simba

[grok87]

Sorry Folks, I have been out of pocket for sometime.

I've updated the spreadsheet and for those interested, You can download the Excel Spreadsheet [rev7h] and OpenOffice version [rev7h]


This includes the 2008 returns and some minor updates.

Best Regards,
Simba

Thanks Simba- good to have you back!
cheers,

[Bob]

Many thanks Simba

[grayfox]

What happened to gold? I thought gold was added to the spreadsheet so that Harry Browne permanent portfolio could be compared.

[stratton]

What happened to gold? I thought gold was added to the spreadsheet so that Harry Browne permanent portfolio could be compared.

Craigr has gold bullion returns here: http://crawlingroad.com/blog/2008/12/22 ... /#more-299

Paul