not another robustness thread...

Discussions about the testing and simulation of mechanical trading systems using historical data and other methods. Trading Blox Customers should post Trading Blox specific questions in the Customer Support forum.
Post Reply
illuminati
Full Member
Full Member
Posts: 24
Joined: Sun Apr 15, 2012 12:52 am

not another robustness thread...

Post by illuminati » Thu Apr 19, 2012 3:49 pm

http://www.dorseywrightmm.com/downloads ... gement.pdf

The above link is a white paper that document the process dorsey wright takes in evaluating and testing systems. Although their paper is for relative strength, I wanted to get people opinion on it and if whether such process can be adapted in testing system components.

Ie, in the paper, they randomly select x number of top decile ranked stocks and hold them. In parallel, I will randomly select top x number of opportunities (given my entry method) from a basket of trades presented each day. This will probably require a massive portfolio. When a system exits a trade, it will again randomly select one of the opportunities and replace it. We define the global parameters like how many positions to hold at any single time. Each run represents a single simulation within a large pool of concurrently running simulations (1000+). This is to not assume that we are favourably picking entries.

Again, this requires a big number of instruments to present to the system so that enough opportunities pop up. It may be inappropriate for futures as the probability of any single day 10+ position popping up is low. But maybe for trend following in stocks?

hmm..

LeviF
Roundtable Knight
Roundtable Knight
Posts: 1428
Joined: Mon Dec 22, 2003 12:24 pm
Location: Des Moines, IA
Contact:

Post by LeviF » Thu Apr 19, 2012 4:32 pm

There is a random portfolio block around here somewhere that you can play with.

Post Reply