Value Averaging Back Testing

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
Topic Author
Done2040
Posts: 4
Joined: Tue Sep 10, 2019 1:45 am

Value Averaging Back Testing

Post by Done2040 » Tue Sep 10, 2019 2:09 am

I recently read that Value Averaging (VA) as a better alternative to Dollar Cost Averaging (DCA). But having done back tests on the S&P 500, VA always seams to perform worse or the same as DCA.

The dataset I have is S&P 500 from 1950 - 2019 excluding dividends, I've tested with the following assumptions:
Monthly contribution: £2000
Value path annual %age: 7%
Investment period: 15 years
Borrowing to stay on the value path is not permitted.

The VA system sells/buys at the start of every month based on the previous months close.
DCA system buys £2000 of stock every month regardless.
Both buy at the month high and sell at the month low (this is what would happen when I invest :oops: )

I've assumed fractional purchases are ok.

The tests were run for all years from 1950 to 2004, in every result the DCA final portfolio value was more or the same as VA.


I just wanted to check if anyone else has run similar tests (or tried it) and got the same results or if I have missed something important.

Silk McCue
Posts: 2852
Joined: Thu Feb 25, 2016 7:11 pm

Re: Value Averaging Back Testing

Post by Silk McCue » Tue Sep 10, 2019 6:45 am

Welcome to Bogleheads!

We have a Wiki that discusses Value Averaging. I had never heard of it until I saw this post.

https://www.bogleheads.org/wiki/Value_averaging

Cheers

User avatar
Svensk Anga
Posts: 568
Joined: Sun Dec 23, 2012 5:16 pm

Re: Value Averaging Back Testing

Post by Svensk Anga » Tue Sep 10, 2019 8:34 am

I suspect that the issue is that your 7% path is too modest a growth rate for this period. The low growth targeted means that a lot of the time value averaging will be keeping cash on the sidelines. DCA wins by being fully invested all the time and enjoys greater than 7% growth. I suggest you run a sensitivity analysis with different, higher growth rate targets. You may need 10%+ for VA to dominate.

Topic Author
Done2040
Posts: 4
Joined: Tue Sep 10, 2019 1:45 am

Re: Value Averaging Back Testing

Post by Done2040 » Wed Sep 11, 2019 5:27 am

Thanks for the answer, currently my portfolio performance looks amazing but they are just paper profits waiting to disappear when the market turns. So I had hoped VA would give me a structure to take profits without negative long term side effects.

I've continued testing with a few other percentages. And the higher the percentage is the better VA performs vs DCA. But that's just a result of it now nearly matching the monthly stock purchases.
The only way it out performs DCA, is to give the cash a good interest rate (over 5%). And even then it's not constant.

For now I'll give VA a miss and keep to DCA. And have a look at optimising through rebalancing. Ideally I would like to have a structure that takes advantage of long bull runs.

But should anyone happen to see or find something that shows Value Averaging is better then Dollar Cost Averaging, please let me know as I still think I may have an error in my code or process.

MotoTrojan
Posts: 5824
Joined: Wed Feb 01, 2017 8:39 pm

Re: Value Averaging Back Testing

Post by MotoTrojan » Wed Sep 11, 2019 7:29 am

I think hard-coding it to buy/sell at the worst monthly times is doing a disservice to the value averaging as it is the only account subjected to both buys & sells, where-as DCA can have a bad buy but then over years it becomes a small driver; I'd probably just use the opening/closing monthly price and see what things look like.

In general though, the market went up over that period, so are you really surprised that the solution that has the largest time-weighted equity exposure is doing better?

User avatar
JoMoney
Posts: 7367
Joined: Tue Jul 23, 2013 5:31 am

Re: Value Averaging Back Testing

Post by JoMoney » Wed Sep 11, 2019 8:23 am

Will have to go searching, but I believe Dr. Bernstein had a lengthy blog that looked at Value Averaging across various past time periods, and came to the conclusion that it was often beneficial over short to intermediate time periods, but over longer time frames lost its edge.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

Topic Author
Done2040
Posts: 4
Joined: Tue Sep 10, 2019 1:45 am

Re: Value Averaging Back Testing

Post by Done2040 » Wed Sep 11, 2019 8:29 am

Hello Moto,
Thanks for the thoughts. I'm not sure which period you are referring to, I tested each 15 year period starting 1950. e.g. 1950-1965, 1951-1966, 1952-1967 ..... 2002-2017, 2003-2018.

I'm still hoping to find an issue with my code as VA sounds like a regulated way to take some profit as the market goes up.


Now using the month open as you suggested for buying and selling with a 10% path, VA performs better in 30 periods and DCA in 25.
However most of the years VA performs better it does so by a very small amount (a few hundred dollars difference in the final portfolio value)

The periods VA beat DCA by a larger amount ($20,000+) were the 5 that ended on and after 1987, as it had amassed cash reserves during the build up and could buy at the end of 1987.

On average these tests show DCA still resulted in higher returns.

Topic Author
Done2040
Posts: 4
Joined: Tue Sep 10, 2019 1:45 am

Re: Value Averaging Back Testing

Post by Done2040 » Wed Sep 11, 2019 8:34 am

JoMoney wrote:
Wed Sep 11, 2019 8:23 am
Will have to go searching, but I believe Dr. Bernstein had a lengthy blog that looked at Value Averaging across various past time periods, and came to the conclusion that it was often beneficial over short to intermediate time periods, but over longer time frames lost its edge.
That fits with what I am seeing. And Dr. Bernstein sounds like he knows more on the subject then myself.

Post Reply