TSP G Fund/Back Tester?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
Topic Author
rich126
Posts: 287
Joined: Thu Mar 01, 2018 4:56 pm

TSP G Fund/Back Tester?

Post by rich126 » Mon Feb 11, 2019 2:35 pm

Are there any portfolio back testing/analyzer tools that incorporate the TSP G fund? I'm guessing the answer is no since it isn't publicly available but I've been wrong before. I did a quick search and didn't find anything but thought I'd check here.

I found historical returns over at https://www.tspdatacenter.com/annual_rates going back to 1988.

Thanks.

User avatar
Taylor Larimore
Advisory Board
Posts: 27836
Joined: Tue Feb 27, 2007 8:09 pm
Location: Miami FL

Re: TSP G Fund/Back Tester?

Post by Taylor Larimore » Mon Feb 11, 2019 2:41 pm

rich126:
Jack Bogle: "The biggest mistake investors make is looking backward at performance and thinking it’ll recur in the future."
Here's more: viewtopic.php?f=10&t=156573

Best wishes.
Taylor
"Simplicity is the master key to financial success." -- Jack Bogle

megabad
Posts: 966
Joined: Fri Jun 01, 2018 4:00 pm

Re: TSP G Fund/Back Tester?

Post by megabad » Mon Feb 11, 2019 8:58 pm

It is calculable. Yield for G is the weighted average yield of all outstanding treasury securities with beyond 4 year maturity. Historically I think this yield falls pretty close to 10 yr treasury rates (what I use when modeling). That's why G is so amazing. You get a 10 yr interest rate with a short term security.

hiddenace
Posts: 11
Joined: Mon Apr 20, 2015 12:54 pm

Re: TSP G Fund/Back Tester?

Post by hiddenace » Tue Feb 12, 2019 12:15 am

I think you can back test using "cash" on portfolio visualizer. The G fund pays a bit more than cash does, but it behaves similarly in that principal doesn't go down. It's like a supercharged savings account, but interestingly the TSP website shows the total bond (F fund) still outperforms.

I even made an Excel sheet with fund price data on the TSP's website stretching back to 2003. The F fund had higher returns, but I think it was skewed by interest rates being low for most of the past two decades...

I split the difference in my fixed income allocation between the F (defationary hedge) and G fund (inflation hedge).

User avatar
White Coat Investor
Posts: 13717
Joined: Fri Mar 02, 2007 9:11 pm
Location: Greatest Snow On Earth

Re: TSP G Fund/Back Tester?

Post by White Coat Investor » Tue Feb 12, 2019 12:35 am

I'd just use cash in your models, like Prime MMF.
1) Invest you must 2) Time is your friend 3) Impulse is your enemy | 4) Basic arithmetic works 5) Stick to simplicity 6) Stay the course

azanon
Posts: 2076
Joined: Mon Nov 07, 2011 10:34 am
Location: Little Rock, AR

Re: TSP G Fund/Back Tester?

Post by azanon » Tue Feb 12, 2019 8:21 am

I think the greatness of G fund is slightly overstated because duration impact on returns is actually zero-sum long-term. Math experts check me, but I would actually assume that if one "dollar cost averaged" into two hypothetical funds - one the G fund, and the other a hypothetical treasury with "average duration of 4 or more years" (so same yield), the latter would end up with more money because the volatility of the latter while DCA would actually help.

Topic Author
rich126
Posts: 287
Joined: Thu Mar 01, 2018 4:56 pm

Re: TSP G Fund/Back Tester?

Post by rich126 » Tue Feb 12, 2019 11:31 am

Last night I saw something on Portfolio Visualizer where it implied you can upload data and create your own "ticker". I just need to create an actual login for the site. I may try that this weekend. Not something I'm going to give a lot of weight to but it is nice to see how things in the past turned out and I saw about 30 years of data for the G fund.

Some of the stuff I read tried to say it is like an intermediate bond fund but when I compared returns to those funds, I didn't see any real correlation and I think it was also true for cash.

Thanks.

User avatar
Ketawa
Posts: 2009
Joined: Mon Aug 22, 2011 1:11 am
Location: DC

Re: TSP G Fund/Back Tester?

Post by Ketawa » Tue Feb 12, 2019 11:38 am

azanon wrote:
Tue Feb 12, 2019 8:21 am
I think the greatness of G fund is slightly overstated because duration impact on returns is actually zero-sum long-term. Math experts check me, but I would actually assume that if one "dollar cost averaged" into two hypothetical funds - one the G fund, and the other a hypothetical treasury with "average duration of 4 or more years" (so same yield), the latter would end up with more money because the volatility of the latter while DCA would actually help.
I don't think this is correct. Given two investments with equal returns, I would much rather have the one with no volatility.

G Fund returns might not be as good as the F Fund, but due to lower volatility, one can hold it and have a safer portfolio supporting slightly higher withdrawals, or hold less of it and have higher expected returns with similar overall portfolio volatility, for two examples.

azanon
Posts: 2076
Joined: Mon Nov 07, 2011 10:34 am
Location: Little Rock, AR

Re: TSP G Fund/Back Tester?

Post by azanon » Tue Feb 12, 2019 11:41 am

Ketawa wrote:
Tue Feb 12, 2019 11:38 am
azanon wrote:
Tue Feb 12, 2019 8:21 am
I think the greatness of G fund is slightly overstated because duration impact on returns is actually zero-sum long-term. Math experts check me, but I would actually assume that if one "dollar cost averaged" into two hypothetical funds - one the G fund, and the other a hypothetical treasury with "average duration of 4 or more years" (so same yield), the latter would end up with more money because the volatility of the latter while DCA would actually help.
I don't think this is correct. Given two investments with equal returns, I would much rather have the one with no volatility.

G Fund returns might not be as good as the F Fund, but due to lower volatility, one can hold it and have a safer portfolio supporting slightly higher withdrawals, or hold less of it and have higher expected returns with similar overall portfolio volatility, for two examples.
I'm pretty sure my math is correct, actually, but was offering to be double-checked. Specifically, on the fund with duration (that can lose value) - or just volatility in general, you're going to automatically buy more shares when rates rise, and less shares when rates fall. A lot of "portfolios" profit that way, by making a bonus return on the volatility when either rebalancing, or in this case, because of duration swings.

Now you may prefer G fund over F fund (I do to, actually), but that doesn't have anything to do with my comment that duration is zero-sum, long-term.

User avatar
Ketawa
Posts: 2009
Joined: Mon Aug 22, 2011 1:11 am
Location: DC

Re: TSP G Fund/Back Tester?

Post by Ketawa » Tue Feb 12, 2019 12:08 pm

azanon wrote:
Tue Feb 12, 2019 11:41 am
Ketawa wrote:
Tue Feb 12, 2019 11:38 am
azanon wrote:
Tue Feb 12, 2019 8:21 am
I think the greatness of G fund is slightly overstated because duration impact on returns is actually zero-sum long-term. Math experts check me, but I would actually assume that if one "dollar cost averaged" into two hypothetical funds - one the G fund, and the other a hypothetical treasury with "average duration of 4 or more years" (so same yield), the latter would end up with more money because the volatility of the latter while DCA would actually help.
I don't think this is correct. Given two investments with equal returns, I would much rather have the one with no volatility.

G Fund returns might not be as good as the F Fund, but due to lower volatility, one can hold it and have a safer portfolio supporting slightly higher withdrawals, or hold less of it and have higher expected returns with similar overall portfolio volatility, for two examples.
I'm pretty sure my math is correct, actually, but was offering to be double-checked. Specifically, on the fund with duration (that can lose value) - or just volatility in general, you're going to automatically buy more shares when rates rise, and less shares when rates fall. A lot of "portfolios" profit that way, by making a bonus return on the volatility when either rebalancing, or in this case, because of duration swings.

Now you may prefer G fund over F fund (I do to, actually), but that doesn't have anything to do with my comment that duration is zero-sum, long-term.
I made a spreadsheet to model this. Here are the results.

Portfolio with no volatility, return = 1.995% (GEOMEAN of 1% and 3% returns)

Code: Select all

No Volatility             1.995%

Year          Added             Starting Value         Return    Final Value
       0     $10,000                    $10,000           $200        $10,200
       1      $1,000                    $11,200           $223        $11,423
       2      $1,000                    $12,423           $248        $12,671
       3      $1,000                    $13,671           $273        $13,944
       4      $1,000                    $14,944           $298        $15,242
       5      $1,000                    $16,242           $324        $16,566
       6      $1,000                    $17,566           $350        $17,916
       7      $1,000                    $18,916           $377        $19,294
       8      $1,000                    $20,294           $405        $20,698
       9      $1,000                    $21,698           $433        $22,131
Portfolio with volatility, return in even years 1.000% and odd years 3.000%

Code: Select all

Volatility Return Even Years      1.000%
Volatility Return Odd Years       3.000%

Year          Added             Starting Value         Return    Final Value
       0     $10,000                    $10,000           $100        $10,100
       1      $1,000                    $11,100           $333        $11,433
       2      $1,000                    $12,433           $124        $12,557
       3      $1,000                    $13,557           $407        $13,964
       4      $1,000                    $14,964           $150        $15,114
       5      $1,000                    $16,114           $483        $16,597
       6      $1,000                    $17,597           $176        $17,773
       7      $1,000                    $18,773           $563        $19,336
       8      $1,000                    $20,336           $203        $20,540
       9      $1,000                    $21,540           $646        $22,186
Portfolio with volatility, return in even years 3.000% and odd years 1.000%

Code: Select all

Volatility Return Even Years      3.000%
Volatility Return Odd Years       1.000%

Year          Added             Starting Value         Return    Final Value
       0     $10,000                    $10,000           $300        $10,300
       1      $1,000                    $11,300           $113        $11,413
       2      $1,000                    $12,413           $372        $12,785
       3      $1,000                    $13,785           $138        $13,923
       4      $1,000                    $14,923           $448        $15,371
       5      $1,000                    $16,371           $164        $16,535
       6      $1,000                    $17,535           $526        $18,061
       7      $1,000                    $19,061           $191        $19,251
       8      $1,000                    $20,251           $608        $20,859
       9      $1,000                    $21,859           $219        $22,077
I played around with some other inputs and found the same thing. There was no advantage to having volatility in the portfolio. The portfolio with volatility did better with higher return in the odd years because the total portfolio is slightly larger in the odd years. It did worse with the higher return in even years because the total portfolio is slightly smaller in even years.

User avatar
Ketawa
Posts: 2009
Joined: Mon Aug 22, 2011 1:11 am
Location: DC

Re: TSP G Fund/Back Tester?

Post by Ketawa » Tue Feb 12, 2019 12:13 pm

Here are some examples with gains and losses, same results.

Code: Select all

No Volatility             2.225%

Year          Added             Starting Value         Return    Final Value
       0     $10,000                    $10,000           $223        $10,223
       1      $1,000                    $11,223           $250        $11,472
       2      $1,000                    $12,472           $278        $12,750
       3      $1,000                    $13,750           $306        $14,056
       4      $1,000                    $15,056           $335        $15,391
       5      $1,000                    $16,391           $365        $16,756
       6      $1,000                    $17,756           $395        $18,151
       7      $1,000                    $19,151           $426        $19,577
       8      $1,000                    $20,577           $458        $21,035
       9      $1,000                    $22,035           $490        $22,525


Volatility Return Even Years     -5.000%
Volatility Return Odd Years      10.000%

Year          Added             Starting Value         Return    Final Value
       0     $10,000                    $10,000          -$500         $9,500
       1      $1,000                    $10,500         $1,050        $11,550
       2      $1,000                    $12,550          -$628        $11,923
       3      $1,000                    $12,923         $1,292        $14,215
       4      $1,000                    $15,215          -$761        $14,454
       5      $1,000                    $15,454         $1,545        $16,999
       6      $1,000                    $17,999          -$900        $17,099
       7      $1,000                    $18,099         $1,810        $19,909
       8      $1,000                    $20,909        -$1,045        $19,864
       9      $1,000                    $20,864         $2,086        $22,950
       

Volatility Return Even Years     10.000%
Volatility Return Odd Years      -5.000%

Year          Added             Starting Value         Return    Final Value
       0     $10,000                    $10,000         $1,000        $11,000
       1      $1,000                    $12,000          -$600        $11,400
       2      $1,000                    $12,400         $1,240        $13,640
       3      $1,000                    $14,640          -$732        $13,908
       4      $1,000                    $14,908         $1,491        $16,399
       5      $1,000                    $17,399          -$870        $16,529
       6      $1,000                    $17,529         $1,753        $19,282
       7      $1,000                    $20,282        -$1,014        $19,268
       8      $1,000                    $20,268         $2,027        $22,294
       9      $1,000                    $23,294        -$1,165        $22,130

azanon
Posts: 2076
Joined: Mon Nov 07, 2011 10:34 am
Location: Little Rock, AR

Re: TSP G Fund/Back Tester?

Post by azanon » Tue Feb 12, 2019 12:19 pm

So duration is zero-sum long term. Thanks for confirming. I think my main point was to say that, as a means of implying that G fund isn't quite as great as some think it is. I certainly didn't intend to endorse adding duration as a benefit or intentionally, so if someone took it that way, that's not what I intended to say.

User avatar
tadamsmar
Posts: 8082
Joined: Mon May 07, 2007 12:33 pm

Re: TSP G Fund/Back Tester?

Post by tadamsmar » Tue Feb 12, 2019 12:41 pm

There is an analyzer tool that incorporates the G Fund: www.financialengines.com

At least it supported the G Fund back when I use to use around 2000. I will tell you a bit about it circa 2000. There was a free version for federal employees. Vanguard has or had some sort of access deal for their clients. I don't know if there is a free version for everyone.

The free version allowed you to test AAs. The paid version (which I used for a while) will create/optimize AAs.

One drawback is that it only worked up until your specified retirement age. It simulated annuitization after that.

The company was founded by Nobel prize winner Bill Sharpe, but I think it's under new management now.

I have never found another analyzer tool or backtest that included the G Fund. Of course, the L funds are the results of an analysis that is limited to only TSP funds.
Last edited by tadamsmar on Tue Feb 12, 2019 12:54 pm, edited 1 time in total.

User avatar
Ketawa
Posts: 2009
Joined: Mon Aug 22, 2011 1:11 am
Location: DC

Re: TSP G Fund/Back Tester?

Post by Ketawa » Tue Feb 12, 2019 12:52 pm

azanon wrote:
Tue Feb 12, 2019 12:19 pm
So duration is zero-sum long term. Thanks for confirming. I think my main point was to say that, as a means of implying that G fund isn't quite as great as some think it is. I certainly didn't intend to endorse adding duration as a benefit or intentionally, so if someone took it that way, that's not what I intended to say.
I agree, being able to "zero out" the volatility of a bond fund caused by its duration generally does not result in higher returns. People shouldn't be expecting excess returns from the G Fund. However, you also said:
azanon wrote:
Tue Feb 12, 2019 8:21 am
...I would actually assume that if one "dollar cost averaged" into two hypothetical funds - one the G fund, and the other a hypothetical treasury with "average duration of 4 or more years" (so same yield), the latter would end up with more money because the volatility of the latter while DCA would actually help.
This is a different claim and I was countering it. I did not incorporate a separate "yield return" and "duration return" component, but my example is equivalent to a scenario where both funds have equal yield and varying return due to duration.

1.000% and 3.000% example: Years with 1.000% return have -0.976% due to duration. Years with 3.000% return have 0.985% due to duration. This would be an overall 0.000% duration return.

-5.000% and 10.000% example: Years with -5.000% return have -7.068% due to duration. Years with 10.000% return have 7.606% due to duration. This would be an overall 0.000% duration return.

User avatar
tadamsmar
Posts: 8082
Joined: Mon May 07, 2007 12:33 pm

Re: TSP G Fund/Back Tester?

Post by tadamsmar » Tue Feb 12, 2019 1:48 pm

"PALO ALTO, Calif., June 12, 2000 – Financial Engines today announced the expansion of the Financial Engines Investment AdvisorSM service to include the Thrift Savings Plan or TSP, the defined contributions retirement plan for Federal government employees. This unique service will initially be offered for free to 2.4 million employees of the Federal government who participate in the TSP, enabling them to have access to institutional quality personalized investment advice to help them choose among the fund alternatives available in their TSP account in order to plan for retirement. Financial Engines is the first online investment advisor to extend its market-leading advice service to all federal employees."

https://financialengines.com/education- ... l-engines/

azanon
Posts: 2076
Joined: Mon Nov 07, 2011 10:34 am
Location: Little Rock, AR

Re: TSP G Fund/Back Tester?

Post by azanon » Tue Feb 12, 2019 2:02 pm

Ketawa wrote:
Tue Feb 12, 2019 12:52 pm
azanon wrote:
Tue Feb 12, 2019 8:21 am
...I would actually assume that if one "dollar cost averaged" into two hypothetical funds - one the G fund, and the other a hypothetical treasury with "average duration of 4 or more years" (so same yield), the latter would end up with more money because the volatility of the latter while DCA would actually help.
This is a different claim and I was countering it. I did not incorporate a separate "yield return" and "duration return" component, but my example is equivalent to a scenario where both funds have equal yield and varying return due to duration.

1.000% and 3.000% example: Years with 1.000% return have -0.976% due to duration. Years with 3.000% return have 0.985% due to duration. This would be an overall 0.000% duration return.

-5.000% and 10.000% example: Years with -5.000% return have -7.068% due to duration. Years with 10.000% return have 7.606% due to duration. This would be an overall 0.000% duration return.
I asked someone to check my math (which is an implication that I knew I could be incorrect). You initially didn't, but later did. I admit I was only receptive to the latter, since that's what I requested. Thanks for taking the time to do so. Hope that clarifies.....

User avatar
Ketawa
Posts: 2009
Joined: Mon Aug 22, 2011 1:11 am
Location: DC

Re: TSP G Fund/Back Tester?

Post by Ketawa » Tue Feb 12, 2019 2:44 pm

rich126 wrote:
Tue Feb 12, 2019 11:31 am
Last night I saw something on Portfolio Visualizer where it implied you can upload data and create your own "ticker". I just need to create an actual login for the site. I may try that this weekend. Not something I'm going to give a lot of weight to but it is nice to see how things in the past turned out and I saw about 30 years of data for the G fund.

Some of the stuff I read tried to say it is like an intermediate bond fund but when I compared returns to those funds, I didn't see any real correlation and I think it was also true for cash.

Thanks.
I wasn't even aware of this! It's very useful.

I uploaded data for the G Fund since 2003 using https://www.tspdatacenter.com/ for 2003 through August 2011, and my own records of the TSP loan rate (equal to the G Fund return) since. I might be able to make it longer if I can access a site with older loan rates. Regardless, since 2003:

CAGR: 3.07%
Stdev: 0.34%
Sharpe Ratio: 6.33
Sortino Ratio: 48.98

Now I'll be able to use it in backtesting asset allocations.

gtwhitegold
Posts: 339
Joined: Fri Sep 21, 2012 1:55 pm

Re: TSP G Fund/Back Tester?

Post by gtwhitegold » Tue Feb 12, 2019 3:10 pm

Ketawa wrote:
Tue Feb 12, 2019 2:44 pm
rich126 wrote:
Tue Feb 12, 2019 11:31 am
Last night I saw something on Portfolio Visualizer where it implied you can upload data and create your own "ticker". I just need to create an actual login for the site. I may try that this weekend. Not something I'm going to give a lot of weight to but it is nice to see how things in the past turned out and I saw about 30 years of data for the G fund.

Some of the stuff I read tried to say it is like an intermediate bond fund but when I compared returns to those funds, I didn't see any real correlation and I think it was also true for cash.

Thanks.
I wasn't even aware of this! It's very useful.

I uploaded data for the G Fund since 2003 using https://www.tspdatacenter.com/ for 2003 through August 2011, and my own records of the TSP loan rate (equal to the G Fund return) since. I might be able to make it longer if I can access a site with older loan rates. Regardless, since 2003:

CAGR: 3.07%
Stdev: 0.34%
Sharpe Ratio: 6.33
Sortino Ratio: 48.98

Now I'll be able to use it in backtesting asset allocations.
All of the share prices since 02 June 2003 are available on the TSP website. Note that the TSP treats each fund as a total return which adds dividends and interest to the share price as they occur.

https://www.tsp.gov/InvestmentFunds/Fun ... index.html

carol-brennan
Posts: 130
Joined: Thu Dec 27, 2018 3:19 pm

Re: TSP G Fund/Back Tester?

Post by carol-brennan » Tue Feb 12, 2019 3:29 pm

I really like what L Income fund is doing at the moment: going from 20 percent securities to 30 percent over the next 10 years, so a "rising glidepath," in other words. Most of it is G.

Topic Author
rich126
Posts: 287
Joined: Thu Mar 01, 2018 4:56 pm

Re: TSP G Fund/Back Tester?

Post by rich126 » Tue Feb 12, 2019 3:43 pm

Ketawa wrote:
Tue Feb 12, 2019 2:44 pm
rich126 wrote:
Tue Feb 12, 2019 11:31 am
Last night I saw something on Portfolio Visualizer where it implied you can upload data and create your own "ticker". I just need to create an actual login for the site. I may try that this weekend. Not something I'm going to give a lot of weight to but it is nice to see how things in the past turned out and I saw about 30 years of data for the G fund.

Some of the stuff I read tried to say it is like an intermediate bond fund but when I compared returns to those funds, I didn't see any real correlation and I think it was also true for cash.

Thanks.
I wasn't even aware of this! It's very useful.

I uploaded data for the G Fund since 2003 using https://www.tspdatacenter.com/ for 2003 through August 2011, and my own records of the TSP loan rate (equal to the G Fund return) since. I might be able to make it longer if I can access a site with older loan rates. Regardless, since 2003:

CAGR: 3.07%
Stdev: 0.34%
Sharpe Ratio: 6.33
Sortino Ratio: 48.98

Now I'll be able to use it in backtesting asset allocations.
That same site has annual returns going back to 1988 but I haven't found monthly returns going back that far. Might be there somewhere. If I find it, I'll post it.
https://www.tspdatacenter.com/annual_rates

Post Reply