up vote 2 down vote favorite
2
share [g+] share [fb]

Some authors are irritating such as Stein, more here, they keep throwing people all kind of new "investing" strategies and different ideas after their last ideas stopped working. I feel such authors are either damn lazy, too proud, too arrogant, trying-to-be-mock-smart or trying-to-fool-newbies. I have tried this with some people and offered some of their writings -- and it is hilarious how easy it is fool people! So I am looking for some easy way to do back-testing. I feel back-testing is important part for credibility.

For example, let's take Stratton's posting (source here):

                                 Tangent    Tangent    Tangent
Asset Class            Ticker      20         25         33
==============================================================
BONDS
Treasury                IEI        30 %       25 %       21 %
Inflation               TIP        30         25         21
Foreign                 BWX        20         10          3
--------------------------------------------------------------
LOW-BETA STOCKS
Energy                  IXC         3 %        6 %        8 %
Consumer Staples        KXI         3          6          8
Utilities               JXI         3          6          8
Health Care             IXJ         3          6          8
--------------------------------------------------------------
SMALL/VALUE STOCKS
U.S.                    VBR         4 %        8 %       12 %
Developed Markets       DLS         3          5          7
Emerging Markets        DGS         1          3          4
--------------------------------------------------------------
Total Expense Ratio              0.29 %     0.31 %     0.33 %
% Non-Dollar Assets                30         30         30

Is there any easy way to backtest this fast? I know I could try to put/guess right times and things in Morningstar with their tickers but is there some shortcut? By back-testing, I mean that I am interested in covariance matrices, SDs, returns, X-square ratio and their evolution over time with different combinations. I know how to do the math and calculations with random data but I do not have the data so I need some provider to offer the back-testing. Any back-testing sites or good suggestions for back-testing?

link|improve this question
88% accept rate
Speaking of Stein, have you read his 2007 article, "Chicken Little’s Brethren, on the Trading Floor", where he lampoons concerns about the "venerable and clever financial house" that is Bear Stearns? – James Roth Jun 26 '11 at 17:09
feedback

3 Answers

up vote 2 down vote accepted

Back-testing itself is flawed. "Past performance is no guarantee of future results" is an important lesson to understand. Market strategies of one kind or another work until they don't.

Edited in --

AssetPlay.net provides a tool that's halfway to what you are looking for. It only goes back to 1972, however. Just to try it, I compared 100% S&P to a 60/40 blend of S&P with 5 yr t-bills (a misnamed asset, 5 yr treasuries are 'notes' not 'bills') I found the mix actually had a better return with lower volatility.

Now, can I count on that to work moving forward? Rates fell during most of this entire period so bonds/notes both looked pretty good. This is my point regarding the backtest concept.

GeniusTrader appears more sophisticated, but command line work on PCs is beyond me. It may be worth a look for you, JP.

ETF Replay appears to be another backtest tool. It has its drawbacks, however, (ETFs only)

link|improve this answer
My grandfather made a healthy profit on stocks, and wouldn't know a covariance matrix from a ham sandwich. Technical analysis gets effective when you're running enough money to affect the market. – duffbeer703 Jun 26 '11 at 2:18
1  
@JoeTaxpayer: how should I respond to this kind of suggestions then? It is very hard for me to objectively "falsify" such statements without proper historical investigation. But perhaps I will be lazy and care less -- and just keep my world-portfolio allocation with rebalancing in the right direction. – JP. Jun 26 '11 at 23:26
@duffbeer703: did he rebalance to certain allocations or just acted impulsely? – JP. Jun 26 '11 at 23:28
@JP - In truth, mine should have been a comment. I didn't 'answer' your question. I'm a numbers guy and a skeptic. Even the site that you linked to had posters commenting that the author keeps changing his recommendations. If the N-year backtest proved out his idea, why all the tinkering? Note I edited my answer to contain two tools I looked at. – JoeTaxpayer Jun 27 '11 at 13:21
1  
The effect comes from selling some of the outperforming asset, S&P rises 20%, now it's more than 60% of assets. 5yr bond helps mitigate the years when S&P is down. The drag of lower average bond returns in god S&P years is less than their positive impact in negative S&P years. You should look at the site and see the impact of different mixes. BTW - Taxes are not included from what I see, it looks like they do not make any assumptions on taxes, as if in IRA type accounts. – JoeTaxpayer Jun 30 '11 at 13:21
show 6 more comments
feedback
up vote 0 down vote

I'd start with a Google search for "best backtesting tools."

Does your online brokerage offer anything?

You already understand that the data is the important part. The good stuff isn't free.

But yeah, if you have some money to spend you can get more than enough data to completely overwhelm you. :)

link|improve this answer
1  
@mbhunter: you must be joking with that first one :P Google returns more spam than useful results, perhaps as JoeTaxpayer hinted, backtesting is flawed and I should forget it. – JP. Jun 26 '11 at 23:37
@JP If you want to backtest, that's up to you. It can give you some insight at some level, I suppose. It's not something I have a strong opinion on one way or the other. As for Google, I was just telling you how I'd start looking. If you see that as fruitless, that's up to you as well. – mbhunter Jun 27 '11 at 16:18
Google is only as good as the words you search on. There are times its responses are dead on, and times I realized certain words are needed to filter conflicting things. This one may be tougher than average. – JoeTaxpayer Jun 27 '11 at 17:22
@mbhunter: well I would never put the word "best" in google result as my first search, such keywords are usually bloated for sure as with this one. It is better to search with more specific search term such as "discreate time rebalancing", to kill the junk. – JP. Jun 28 '11 at 20:15
Usually when I add "best" it gives me a roundup of several good ones in a single post. Hence why I use it. – mbhunter Jun 28 '11 at 20:49
show 2 more comments
feedback
up vote 0 down vote

yAnother potential tool for you would be a Monte Carlo Simulator. here's one http://financial-dictionary.thefreedictionary.com/Business+Fundamentals

I know that past performance is no guarantee..... but I think it's in many cases not exactly a flawed tool, and especially with respect to money managers a good way to find good ones. If a manager has shown an ability over time to consistently beat the market, yes he might be due for a bad day, but you'd generally expect that they should be able to continue that trend.

I'd apply the same logic to pundits. If their track record sucks, and they constantly seem to whipsaw you with their advice, why listen to them other than

  • the value they provide in terms of comic relief
  • a potential basis for taking a contrary position (if they prove over time to be wrong way more often than right, perhaps the best thing is to listen to what they say and then do the opposite of what they recommend.)
link|improve this answer
feedback

Your Answer

 
or
required, but never shown
discard

Not the answer you're looking for? Browse other questions tagged or ask your own question.