[kimchi]

[ QUOTE ]
Aside: I never understood the idea of "forward testing". It is just the slowest possible way to do backtesting: later.

eastbay


[/ QUOTE ]


I meant more about splitting data into chunks and doing separate test on each.

Yeah, I wouldn't use forward testing for strategy testing so much. I think forward testing is more useful to see how you get along with actually using your system day to day. For example, I spent some time backtesting some ideas on The US markets only to discover during forward testing that my broker didn't allow the movement of guaranteed stops outside of market hours. And US market hours didn't fit into my normal daily routine.

Or maybe searching, screening, filtering, datamining and entering/adjusting orders is proven to take too much time.

[ QUOTE ]
You appear to have done considerable homework. You seem to have an good grasp on many important trading concepts. Given my impression of your study in trading, I am surprised to see that you are apparently thinking about zeroing in on one equity instrument.

[/ QUOTE ]

I'm not thinking about trading a single market, and certainly not necessarily RGU. It's just that RGU exhibits some of the qualities that I feel might suit this type of trading. I think we can discuss markets and trading vehicles more later in a 'markets' sub-thread.

Hopefully, the forum will suggest a basket of (perhaps lowly/non correlated) markets to try and trade after we've made a list of criteria needed to be condidered before a market is included in the system.

[ QUOTE ]
Truly robust systems work well using a very narrow range of 'optimal' parameters tested against a great many instruments

[/ QUOTE ]

I intend to do some testing using the same parameters across different markets which will include some stress conditions (ie - 10/1987, bond/debt crises, 9-11, etc)

[ QUOTE ]
The book New Trading Systems and Methods 4rd edition by Perry Kaufman is an excellent resource for more information on this and many other essential system construction concepts.

[/ QUOTE ]

Thanks for the suggestion. I have a few of Kaufman's books on my bloated Amazon wishlist but I've only read some articles and such, and never a book. I have tried to incorporate his adaptive moving average into some systems without successs. I suppose I should just read more about his ideas in his books. I think his adaptive MA is one of those fantastically simple ideas which I've been unable to implement. I actually spent a long frustrating, fruitless time trying to develop my own AMA before I realised it had already been done before - Doh!