10mo. SMA System - Your Opinions?

[david.alan.williams]

Hello all,

I'd like your basic opinions and how you would rate this SMA system. (Very Good, Good, Fair, Poor)

This was Happy Trails initial idea taken from his reading of Mebane Faber's "The Ivy Portfolio" and also, an article in Fedsmith dated 7/28/10.

I worked with him on back-testing this formula using the TSP share prices from 2006-2010. What I noticed was that there was never a NEGATIVE year and only 1 year fell below the G-Fund. Here are the results:

2010 15.79%
2009 27.71%
2008 3.75%
2007 1.48%
2006 13.21%
5year total 62.12%
Average Per Year 15.53%

Here are a few of our top traders and systems. NOTE: I may have calculated some of these wrong!!!

Sentiment Survey 4yr, Per Year Average 10.55%
Aviator Guy 4yr, Per Year Average 6.94%
Birchtree 5yr, Per Year Average 5.64%
Trader Fred 4yr, Per Year Average 5.50%
Mac Daddy 3yr, Per Year Average 4.47%
Ed the Fed 5yr, Per Year Average 3.60%
Revshark 4yr, Per Year Average .45%
Ebbchart 4yr, Per Year Average .42%


Using the 10mo. SMA, I built-in a 1st of the month IFT, and a mid-month IFT. I tried all sort of variations between the funds and different percentages and the system that gained the most was to only use F-Fund and S-Fund. (above results). You are either 100% S or 100% F based on what the system tells you to do at the 2 IFT points. (1st and 15th)

I'm still working, testing, and retesting the system. I plan to go back 1 more year (2005) and see if it still holds up.

I'd like your basic opinions and how you would rate this SMA system. (Very Good, Good, Fair, Poor)

Thank you for your time and thanks to Happy Trails!
D

[JTH]

The approach is right, the further you step back and trade with the big picture, the better you can perform with a 2IFT system. What you don't have is enough backtesting, not nearly enough. Of 5 years tested only 1.5 was in a bear market. I would suggest going back to 1997. Test the Bull markets separately, from the bear markets, mix up your time-frames so you can get more results. Importantly, don't cherry-pick your data, it's all a random walk, so you want to test your system under various points in time. With the S&P 500 you can go back to the late 1930s.

Inevitably the biggest flaw in many systems is your inability to walk it forward and think in real time. It's easy to look at the past, but much harder when the turkey is in the oven. Make sure you have solid rules in place for every conceivable situation. Triple check your numbers, one screw-up can skew the whole thing.

I have a Monthly Talk thread that covers the 12SMA. http://www.tsptalk.com/mb/showthread.php?t=8514

[david.alan.williams]

Thanks! THE TSP Website only goes back to 2 June 2003 unless there is an archive somewhere. I will at least go back to 2 June 2003. It takes a LOT of time to key all this in correctly. I do want to go back as far as I can.

[Intrepid_Timer]

Well, I believe the average return of the S&P over the last 50 years is 6.2%, and since the S&P is the benchmark index, I'd rate any system getting close to that as just fair, since it isn't really beating a buy and hold strategy.

So I'd rate anything getting less than 5% as poor, 6-10% as fair, 10-15% as good, 16-20% as very good and anything over 21% as excellent.

On this scale, I'd rate it as good.

[Happy_Trails]

The approach is right, the further you step back and trade with the big picture, the better you can perform with a 2IFT system. What you don't have is enough backtesting, not nearly enough. Of 5 years tested only 1.5 was in a bear market. I would suggest going back to 1997. Test the Bull markets separately, from the bear markets, mix up your time-frames so you can get more results. Importantly, don't cherry-pick your data, it's all a random walk, so you want to test your system under various points in time. With the S&P 500 you can go back to the late 1930s.

Inevitably the biggest flaw in many systems is your inability to walk it forward and think in real time. It's easy to look at the past, but much harder when the turkey is in the oven. Make sure you have solid rules in place for every conceivable situation. Triple check your numbers, one screw-up can skew the whole thing.

I have a Monthly Talk thread that covers the 12SMA. http://www.tsptalk.com/mb/showthread.php?t=8514

That is interesting news from Bulkowski regarding the 12 month SMA vs the 10 month SMA. Thanks for the link.

Also, David, thank you for taking the time to backtest this system. :)

[david.alan.williams]

I just finished 2005 and it came in at 10.45%

I also finished 2004 (Mid March thru December) and it came in at 14.16%.

I can't go back any further until I find share prices earlier than 2 June 2003.

There was one error in my calculations but for the, almost 7 year period, it earned 86.74% an average of 12.39% per year. Still falls into Intrepid Trader's "Good" catagory.

Now to make it better.....LOL

D

[JTH]

For myself, it's not enough to maintain my 6 year 12% average, I'm more concerned with consistently beating the markets under all circumstances. A 10 SMA system has the potential to get slaughtered, should the market decide to trade with some sideways volitility. I should add that this year I am not using a system because I am unable to achieve the perfection I am looking for. Everytime I think of a new idea, I end up finding the flaws in it, and thowing it in the waste basket.

My inabilty to accept anything less than perfection has been and continues to be my biggest downfall...

[KevinD]

http://tsp.peacefulgains.com/Thrift-Savings-Plan/


Here at Peaceful Gains we've solved these problems, and we want to share the solution with everyone. We give you, in a single spreadsheet, the daily prices of all the TSP funds starting in January 1995. The spreadsheet is released under the Creative Commons Attribution-Share Alike 3.0 United States license. This basically means that anyone is free to use the spreadsheet as long as, when they distribute the resultant work, they release it under a compatible license and acknowledge us by linking back to our site.

http://tsp.peacefulgains.com/Historical-daily-prices-of-Thrift-Savings-Plan-funds/

[david.alan.williams]

Thanks Kevin!!!!

That help me finish going back as far as the S and I Funds existed.

2010 15.79%
2009 27.71%
2008 3.75%
2007 1.48%
2006 13.21%
2005 10.45%
2004 16.99%
2003 45.59%
2002 (10 Months) -2.49%

9 year total 132.68%

Average Per Year 18.95%

So far, very impressive. I wish I had earned an average of 18% per year!

I wish the C and S data went back further but since they don't, I decided to use the same system but use G, F, and C instead. All looks great back to 2002. C Retreating to F had no negative years and earned 89.60%. That's great considering the C Fund itself didn't earn but 48.34% over the same 9 years! ( Tomorrow, I will use G, F, and C to go all the way back to 1994.

D

[Birchtree]

Those numbers would look even better taking into account the social redemption of dollar cost averaging - that's the beauty of TSP. The 2008 price of $8-$10 on the C fund is now gone forever - but if someone bought those prices they are good for a lifetime. Always keep your money working in the stock funds even during bear markets - that can be a profitable blessing if you have guts and capital.

[Happy_Trails]

Thanks Kevin!!!!

That help me finish going back as far as the S and I Funds existed.

2010 15.79%
2009 27.71%
2008 3.75%
2007 1.48%
2006 13.21%
2005 10.45%
2004 16.99%
2003 45.59%
2002 (10 Months) -2.49%

9 year total 132.68%

Average Per Year 18.95%

So far, very impressive. I wish I had earned an average of 18% per year!

I wish the C and S data went back further but since they don't, I decided to use the same system but use G, F, and C instead. All looks great back to 2002. C Retreating to F had no negative years and earned 89.60%. That's great considering the C Fund itself didn't earn but 48.34% over the same 9 years! ( Tomorrow, I will use G, F, and C to go all the way back to 1994.

D

Hi David, thanks for the update. The 90s were generally really great years for the C fund and so, my guess would be that the system will look good, maybe even too good using just C fund retreating to F or G over that time period. My guess would be that C fund alone moving to G or F would have done a lot better than a combination of C,S,I back in the 90s.

[Happy_Trails]

Those numbers would look even better taking into account the social redemption of dollar cost averaging - that's the beauty of TSP. The 2008 price of $8-$10 on the C fund is now gone forever - but if someone bought those prices they are good for a lifetime. Always keep your money working in the stock funds even during bear markets - that can be a profitable blessing if you have guts and capital.

Birchtree,

Do you stay fully invested during bear markets? With a huge account like yours, it seems like it would be really tough to watch it drop 25-50% in a year. Our paltry 16000K annual contribution DCA'd over the year might help the account to recover somewhat quicker but getting back in with most of your pre-bear market capital intact seems like it would be more profitable. It seems like you should be able to buy a lot more shares at say $9 coming out of the bear market with most of your pre bear capital still intact.

[Birchtree]

Someday perhaps you'll have the opportunity to make the same decisions - it's always easier in retrospect. We all play the game our own way - things really open up once you get past 50 years old. I rode the bear and survived just fine from the depths of 2008 - using dollar cost averaging was the redeemer but it also took courage to throw money down a deep dark well accumulating shares that were depreciating daily but I knew from common sense there would be a rebound and I'd catch every nickle on the way back up. Now that I'm retired would I take the same risk again - probably not because I no longer can DCA - so the dynamics have changed. Statistics tell us that few people bought the March 09 lows but I didn't have that decision to make - I was already there to take the ride and it proved very profitable in both my accounts.

[Happy_Trails]

Someday perhaps you'll have the opportunity to make the same decisions - it's always easier in retrospect. We all play the game our own way - things really open up once you get past 50 years old. I rode the bear and survived just fine from the depths of 2008 - using dollar cost averaging was the redeemer but it also took courage to throw money down a deep dark well accumulating shares that were depreciating daily but I knew from common sense there would be a rebound and I'd catch every nickle on the way back up. Now that I'm retired would I take the same risk again - probably not because I no longer can DCA - so the dynamics have changed.

Thanks, BT. Well, I rode that train myself back in 2008 watching it drop lower and lower (I was strictly buying and holding/dollar cost averaging with my contributions each payperiod). It was incredibly difficult to keep with it but I did. In fact, I bought more shares by increasing my stock allocation. My account value has recovered. I cannot help but wonder if I would have been following the SMA if I would have missed most of the pain. I have a ways to go still - about 14 years until retirement. Either way (buy and hold or trend following/timing) takes patience and perseverence. Looking at the SMA system with rose colored lenses could be a bad thing as well. JTH makes a great point that in a sideways market, the system will struggle. I respect your views BT, I have read your account talk off and on for several months and appreciate what you have to say. :)

[david.alan.williams]

Thanks for all the opinions thus far! I would like for you to review the attached Final GFC 1997-2010 back-tested data and the Final GFCSI 2002-2010 back-tested data for your opinions and sanity check. I will upload the 2002-2010 data in the next post due to file size limits.

I agree that the 12mo. SMA DOES NOT WORK. I tried it along with 11, 10, 9, 8 month checks.

The 10 mo. SMA was the most stable and reliable. In my spreadsheet calcluations, I really didn't use a true 10mo. SMA. I used a modified version of the 10mo. SMA. The initial reason was just the ease of the calculation in excel. Since the plan was to have a mid-month IFT, I needed the mid-month share prices. It was then easier in excel to add the last 10 mo. ending share prices to the last 10mo. mid-month share prices, then divide by 20. (Thus why I calling it a modified 10mo. SMA.

I did go back this morning and change both spreadsheets to a true 10mo. SMA and to my surprise, all of the percentages DROPPED. The lean years got worse. I changed the spreadsheets back to the modified version. All I can say is that the modified version is better; more sensitive.

For the really aggressive people, this plan is probably too conservative.

However, there are many folks who would like to earn more than the G Fund but without seeing their earnings drop off the face of the earth during a bear market and also something that is low maintenance. So far (test back to 1996), this system will achieve those goals.

GFC results were:

15 years - Best was "C 100% retreats to F" - 184.45% or 12.30% average per year (basically, we gain all of the 15 yr C Fund earnings 132.67% and also gained well over half of the 15 yr F Fund earnings)
*** Only 1 year was negative -4.13%
*** 3 other years didn't beat the G Fund
*** 4 years gained more than 22%
*** 72% of the time in C Fund, 28% of the time in F Fund
*** 30 Total IFT's

GFCSI results were:

8 years 10 months - Best was "S 100% retreats to F" - 132.69% or 14.74% average per year (basically, we gain all of the 9 yr S Fund earnings 99.65% and also gained well over 3/4 of the 9 yr F Fund earnings)
*** Only 1 year was negative -4.13%
*** 1 other year didn't beat the G Fund
*** 2 years gained more than 22%
*** 72% of the time in S Fund, 28% of the time in F Fund
*** 18 Total IFT's

D

[david.alan.williams]

Here is the 2nd file. 2002-2009 GFCSI back-tested data.

[david.alan.williams]

Here is the 2011 SMA worksheet if anyone wants to use this system. I will start using it at the beginning of the month.

Wish me luck!

D

[tsptalk]

Hi David. I see that you are trying to register a new message board account. We can't do that, but I assume it is because you want to track this system.

We could possibly add it to the tracker, but if you are going to follow it, won't that just be be duplicate account to track?

Thanks

[Happy_Trails]

Here is the 2011 SMA worksheet if anyone wants to use this system. I will start using it at the beginning of the month.

Wish me luck!

D

Hello David and good luck my friend. Would it be possible for you to create another 2011 spreadsheet that includes all of the funds (not just F&S)? I am not quite ready to commit to only F&S for the year but would love a version with all of the funds and with your revised 10 month SMA calculations.

[david.alan.williams]

Happy,

I'd be glad to.

TSPTalk,

No problem. Just delete the request.

[david.alan.williams]

Happy,

Looking at the 2002-2010 Worksheet, which tab are you interested in?

33/33/34 retreat to G....or which one. I have to build it based on your choice.

v/r
D

[Happy_Trails]

Happy,

Looking at the 2002-2010 Worksheet, which tab are you interested in?

33/33/34 retreat to G....or which one. I have to build it based on your choice.

v/r
D

Hi David,

33/33/34 retreat to G or F. After looking at the data, I don't think I can afford to overlook F. It makes sense as bonds tend to not move in lock step with stocks.

Thanks

[david.alan.williams]

Here you go. CSI retreats to F.

In the Jan 2011 share prices row, put in a low number (like 2) in the C S and I cells and you will see how it will do.

Make sure you UNDO those otherwise the percentages will be wrong at the bottom.

Let me know if it's OK.

[Happy_Trails]

Here you go. CSI retreats to F.

In the Jan 2011 share prices row, put in a low number (like 2) in the C S and I cells and you will see how it will do.

Make sure you UNDO those otherwise the percentages will be wrong at the bottom.

Let me know if it's OK.

Thanks David - this is great, again thanks! :)