10 Month Modified SMA Method (M-SMA)

[david.alan.williams]

I thought I'd start a thread specifically for the Modified 10 Month Simple Moving Average Method (M-SMA).

I improved the safety factor and, in doing so, improved the performance. Naturally, when placing a new safeguard into the system, you would think performance would decrease but the safeguards eliminated many negative returns while only eliminating a few positive returns. Below, I will explain the M-SMA System.

After Mod Before Mod

2002 3.91% -2.49%
2003 45.59% 45.59%
2004 19.79% 16.99%
2005 7.11% 10.45%
2006 13.21% 13.21%
2007 6.13% 1.48%
2008 3.75% 3.75%
2009 21.87% 27.71%
2010 17.67% 15.97%

9 Year - 139.02% 132.68%
Avg. per year - 15.45% 14.74%


Features:

* - Uses only S Fund and F Fund
* - Allows 2 IFT's per month. (if needed)
* - First of the month Checkpoint, Mid-Month Checkpoint
* - If a loss of -2.99% or greater is detected, system places you in the F Fund for 2 weeks (the new safeguard mentioned above)
* - If current share price is below the 10 mo. M-SMA, system places you in the F Fund for 2 weeks
* - Backtest thru 2002 with F and S Funds.
* - Method Backtest thru 1995 using F and C Funds.
* - Backtesting has NO negative years.
* - Averaged 15.45% per year 2002-2010
* - Simplicity, just enter the S and F Fund share prices, and the spreadsheet will automatically tell you to BUY or SELL S Fund.

About the Original System:

The 10 Month SMA (Simple Moving Average) system was discussed in Mebane Faber's "The Ivy Portfolio" and also, an article in Fedsmith dated 7/28/10. In this system, there is only 1 IFT per month. The end-of-month share prices for the last 10 months are added together and divided by 10 to calculate the SMA. If the end of the current months share price is below the 10 month SMA, then a SELL (Stocks) signal is created. If the end of the current months share price is above the 10 month SMA, then a BUY (Stocks) signal is created.

The Modifications:

* - Increased from 1 IFT to 2 IFT's per month - When backtesting, I found that having a mid-month IFT greatly increased reaction time to the stock market, decreased losses, and increased gains.

* - Used last 10 end-of-month share prices and last 10 mid-month share prices, then divided by 20 to get the M-SMA. I found that this greatly increased the accuracy and profits during the backtesting.

* - Placed additional safeguard in the system - If you lose -2.99% or more during a checkpoint, the system places you in the F Fund for 2 weeks even if the current share price is above the SMA. (Override) This actually eliminated all negative years above and added an addition 3/4% per year on the average.

Why S and F Funds?

I backtested all funds and many different percentages between the funds. The S and F Funds outperformed, in combination, to beat all other scenarios.

Attached is the automated spreadsheet for 2011.

D

 

[akpackfan]

I own the book "The Ivy Portfolio" and was intrigued by the 10mo. SMA method talked about in it so I started doing my own research on it. I got the end of month closing prices all the way back to 1928 from "Think or Swim" software and plotted a 10 month SMA line to see when it called for a sell or buy and how much the loss or gain was by following the signals. The market crash in the '30's wreaked havoc on even this system because if you strictly followed it and waited till the beginning of the month to move your money you still lost about 42% (still less than the actual crash amount). Other than that the worst year was in 1990 around the May to August time frame when you lost about 10% if you waited to mover till the beginning of the month.

It did get whipsawed several times over the last 80 or so years but the loss was typically only around 3 to 4%. Other than the crash in the 30's it kept you out of all the serious bear markets and kept you in during all the bull markets. If you used a mid month signal like you describe I would imagine the results would have been even better. It is currently on a buy since Sept. 1 '10. Before that you would have sold and went to safety on May 1 of '10. I am seriously thinking about using this strategy but haven't totally convinced myself yet. Maybe you can convince me......

 

[ExtremeWX]

Do you have plans on creating a new user id on this site for the program? I think it would be interesting to see how it works out.

Maybe - 10MonthSMA as the user name. ;-)

I see you are 100% G on the tracker and your system is 100% S. Am I reading it correctly?
Regards,
Bryan

 

[david.alan.williams]

Yes. I just finished my research and was following I_T Trading System during his free trial. I posted a few days ago that I would start using this M-SMA system at the first of the month.

I believe that, if there is enough interest, they might put it on the tracker. They wouldn't let me create a 2nd account but I will be on this system starting 1 Feb. if you want to track me. Also, you can use this spreadsheet to track it if you would like.

Thanks!

D

 

[david.alan.williams]

I own the book "The Ivy Portfolio" and was intrigued by the 10mo. SMA method talked about in it so I started doing my own research on it. I got the end of month closing prices all the way back to 1928 from "Think or Swim" software and plotted a 10 month SMA line to see when it called for a sell or buy and how much the loss or gain was by following the signals. The market crash in the '30's wreaked havoc on even this system because if you strictly followed it and waited till the beginning of the month to move your money you still lost about 42% (still less than the actual crash amount). Other than that the worst year was in 1990 around the May to August time frame when you lost about 10% if you waited to mover till the beginning of the month.

It did get whipsawed several times over the last 80 or so years but the loss was typically only around 3 to 4%. Other than the crash in the 30's it kept you out of all the serious bear markets and kept you in during all the bull markets. If you used a mid month signal like you describe I would imagine the results would have been even better. It is currently on a buy since Sept. 1 '10. Before that you would have sold and went to safety on May 1 of '10. I am seriously thinking about using this strategy but haven't totally convinced myself yet. Maybe you can convince me......

Lord, I don't know if I can convince myself, but I'm going to try it. LOL

Attached is the backtested data since 2002 (S Fund Inception) with the -3% fail-safe in place.

D

 

[JTH]

David, I forgot to mention there is a great amount of statistical data that favors up days in the beginning of the month and mid month. Have you tried testing the end of the 1st & 3rd week?

 

[david.alan.williams]

David, I forgot to mention there is a great amount of statistical data that favors up days in the beginning of the month and mid month. Have you tried testing the end of the 1st & 3rd week?

Thanks JTH, I'll test that.

 

[akpackfan]

David, what "day" should we use for the "mid Month" share price? Also I assume, for example, the "Jan 2011" entries would be the closing price of the last trading day of the month?

 

[david.alan.williams]

David, what "day" should we use for the "mid Month" share price? Also I assume, for example, the "Jan 2011" entries would be the closing price of the last trading day of the month?

You are correct on the "Month - Year" entries as being the last day of the month that the stock market closes.

As far as mid month, I try to calculate the number of days in the month that the market will be open and then pick a day in the middle. Some months might be the 15th, but some may be the 14th, 16th, or even 17th.

D

 

[akpackfan]

Thanks. For your 2011 spreadsheet to choose the right fund won't we need the prices for a lot of last years mid and end of month for a lot of this years months? Did you consider putting that data into the 2011 spreadsheet so the calculations work? Not complaining, just asking incase I am misunderstanding.

 

[david.alan.williams]

Thanks. For your 2011 spreadsheet to choose the right fund won't we need the prices for a lot of last years mid and end of month for a lot of this years months? Did you consider putting that data into the 2011 spreadsheet so the calculations work? Not complaining, just asking incase I am misunderstanding.

I'm sorry I confused you. I hid those rows to help prevent accidental change. If you notice, the rows count 1,2,3,4,5, then skip to 26,27,28.

Rows 6 thru 25 are hidden. You can unhide them if you like by Selecting the whole spreadsheet then right click and select "Unhide".

That was sneaky wasn't it?

 

[david.alan.williams]

Version 3 Attached

Improved Yearly Average from 15.45% to 21.35% by adding I and G Fund Options. Added more Protection. See "v3 Summary of Changes" and "Flowchart Decision Matrix" Tabs for more info.

All year show double digit gains except one. See "Backtest Summary" and "Backtested Data" Tabs.

NEARLY hit that 2% per month goal but I'm not so sure how achievable that is with years like 2008 where the highest fund was 5.45% but who knows.

D

 

[grandma]

That is a very interesting spreadsheet! Thank you for sharing it.
It is as if the market knowledge is compounding (like interest) daily since
Tom got this going!
I'm so lucky!

 

[david.alan.williams]

Thanks Grandma!

As you said a couple of days ago in another thread, it's about sharing idea's with each other and then making your on decisions as to what is best for you. I don't believe I would have changed anything if it hadn't been for JTH's MARCON thread, Birch's thread comments, Akpackfan and Happy Trails moral support, and many other contributors.

Now today I'll be going on JTH's recommendation to test the system using the last day of the first and third weeks of a given month. (Maybe he just knew that would keep me busy for a while, thus I can't post) :blink: Just KIDDING!

If I'd known how much enjoyment I would get from this site and how great the folks are here, I'd joined long ago.

A California members meeting??? We have to get passports and VISA's right?

D

 

[Happy_Trails]

Hey David,

I have an idea if you have time to check on it. Since you are going to look at the 1st and 3rd week, how about also looking at a modified SMA with a weekly check (weekly prices for the last 10 months / 40 using the rules you have developed for your existing system)? I would be curious to know if a weekly check could potentially reduce downside risk without hurting return. Only if you have the time and interest to check on it.

 

[Beavis]

Hey David,

I have an idea if you have time to check on it. Since you are going to look at the 1st and 3rd week, how about also looking at a modified SMA with a weekly check (weekly prices for the last 10 months / 40 using the rules you have developed for your existing system)? I would be curious to know if a weekly check could potentially reduce downside risk without hurting return. Only if you have the time and interest to check on it.

HT,

You are suggesting a 40 period SMA in the above post and David is actually using a 20 period SMA with his mid-month thrown in. I ran a quick spreadsheet using the weekly closing price with a 10 week SMA over Jan - Dec 2010. Using David's rules, it generated a buy S signal on 31 Dec 09 so you would have been in the S fund during the first week of Jan 10. From there it had a cumulative return of 49.86% by the end of 2010. It would also have generated another 6.47% in 2011 as of the close yesterday. Might be worth backtesting the other years.

One caveat with this system. If you get a fund change signal on Friday, you don't actually make the change until COB Monday. Not a huge deal when you're changing monthly, but becomes (possibly) a bigger deal as you go to weekly signals.

Beavis

 

[david.alan.williams]

Hello HT,

Great idea, I'll try that.

Hi Beavis,

This might be hard to explain but I'll give it a try. I think you made the same mistake that I DID when I first started out. It was hard to wrap my head around at first.

Lets take a given week in your sheet. 26 Feb 10

In the column "Allocation" is your prediction for the NEXT week thus those percentages can only be applied to compute your NEXT week earnings, not the current week.

Look at the attached modified spreadsheet. I renamed "Allocation" to "Current Allocation" and added a new Column of "Future Allocation". The Future Allocation is your buy/sell signal for next week and will automatically be copied into next weeks "Current Allocation" columns.

(I have a "Current Allocation" that is copied from the previous weeks prediction "Future Allocation")

Glance back and forth between the Current and Future Allocation columns and you will notice that everything is delayed by 1 week.

Gee, am I making any sense????

Recalculate it would be 20.95% for 2010.

D

 

[Beavis]

Hello HT,

Great idea, I'll try that.

Hi Beavis,

This might be hard to explain but I'll give it a try. I think you made the same mistake that I DID when I first started out. It was hard to wrap my head around at first.

Lets take a given week in your sheet. 26 Feb 10

In the column "Allocation" is your prediction for the NEXT week thus those percentages can only be applied to compute your NEXT week earnings, not the current week.

Look at the attached modified spreadsheet. I renamed "Allocation" to "Current Allocation" and added a new Column of "Future Allocation". The Future Allocation is your buy/sell signal for next week and will automatically be copied into next weeks "Current Allocation" columns.

(I have a "Current Allocation" that is copied from the previous weeks prediction "Future Allocation")

Glance back and forth between the Current and Future Allocation columns and you will notice that everything is delayed by 1 week.

Gee, am I making any sense????

Recalculate it would be 20.95% for 2010.

D

Yeah I figured that one out. Instead of predicting the best fund for next week, it was telling me the best fund for last week. I also had the cumulative total wrong using simply addition instead of multiplying the percents; been a while since I worked with this stuff.

Thanks for starting this process. I'm taking a look at the weekly data back to the same period you used. I'll let you know what I find out.

 

[akpackfan]

Thanks to all of you who are willing to crunch the numbers and share your wisdom. You are appreciated.

 

[akpackfan]

OK, so how do you all think the best way to implement this strategy would be in a brokerage account and/or an IRA?