[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Multi-symbol backtesting and optimization



PureBytes Links

Trading Reference Links

Hello everybody...

This is my first post to this list...
I have just bought and received my copy of TradeStation 4.0, a few days ago.

Being French (sorry...I can't do much against it...), I hope you will
forgive me for my approximate English, my arrogance...and my bad habits of
eating frogs, snails....and other disgusting animals...

I Am not a Solutions Provider...As my pseudo suggests...I Am trading
part-time (and part-time doing some independent computer consulting) and I
am always barefoot ! I have also written 13 books about Numerical Analysis
and computers, in French mainly but also in Italian and English. The last
one is entitled "Mathematical Software Tools in C++".

Ok..that's it for the short presentation...feel free to ask more.

I have been quite disappointed up to now to realize that Tradestation was
backtesting and optimizing for only ONE symbol at a time.

I find that particularly negative to backtest some long-term-daily systems,
for which I could have only a handful of signals for the entire 5 years
historical period...Doing statistics on 3 or 4 trades...would not be of a
great interest, would it ?

Do any of you know of a method allowing to do some multi-stock backtesting
and optimization and to bypass this structural limitation of Tradestation ??
Any idea welcome...

I imagined a solution which could help : Tradestation is waiting ONE and
only one symbol at a time.Well ! We will give it only one...but a virtual
one....constructed through a separate routine, which will concatene the
historical values of several stocks together, after some normalization, and
the insertion of some specific patterns allowing in Easy Language in the
system to realize that a new action is coming...and that some specific
actions should be taken.

The main limit to this approach is the 13000 bars limit, which makes
nevertheless possible to concatene 13 actions on 5 years each (always
speaking of daily...). Better than nothing....the ability to backtest and
optimize on a dozen of stocks simultaneously...and I have read somewhere
that this limitation should be overcome in version 5, is that right ?

"Bricolage" as we would say in French....but it could work...what do you
think of it,  you experienced users of the product ?? Do you have any better
idea ??

Many thanks in advance for your answers...

-------------------------------------------------------------
Barefoot Trader, le trader aux pieds nus