Page 1 of 1

Testing synthetic databases in TBB

Posted: Thu Sep 16, 2010 10:59 am
by douglas_alcantara
Trading Blox Builder software allows it to be tested several synthetic databases generated from a single asset, one at a time, and allows me to be able to check the average profitability and maximum drawndown in all tests, as the author Tushar Chande does in his book Beyond Technical Analysis - pag. 353 ?

Thanks

Douglas

Posted: Fri Sep 17, 2010 12:07 pm
by sluggo
The user's guide for Trading Blox is freely available and can be downloaded from the Support / Documentation page on the main website. The programming manual is also free and downloadable. Quick links: (USER'S GUIDE) and (PROGRAMMING MANUAL)

I just opened each of them in Adobe Acrobat Reader and searched for the word "synthetic". It is not found in either of the manuals. I suppose this means that you would have to write your own program to generate several synthetic databases from a single asset.

Posted: Fri Sep 17, 2010 1:28 pm
by Tim Arnold
Trading Blox does not create synthetic data. But if you have different data sources for the same markets, you could run separate tests with each type of data.

Synthetic test data...

Posted: Sat Oct 30, 2010 1:10 pm
by zentrader
@Douglas,

perhaps my tool "Zen Monte Carlo Simulator" serves for you, because it generates random synthetic data derived from a basis historical data file (functionality: data simulation)

More information here:
http://www.zentrader.de/html/monte_carl ... ator1.html

bye,
Volker

Posted: Sat Nov 20, 2010 1:00 pm
by rgd
I had the good fortune of testing my systems on a 100 year synthetic portfolio. I believe it was a 20 market portfolio. There was a direct relationship between holding period and how quickly the equity was depleted. My longest term systems enured the entire 100 year test with a max drawdown of 55% and paltry low single digit returns. Which proved to me that robust, long term systems can extract returns from a random data series. Actually, the data series was not completely random, as it reordered the price series with 2 bar sampling from the real price series.