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

Re: To MarkBrown: Real-time datafeeds in Python



PureBytes Links

Trading Reference Links

Cameron,

>re ... "It would be interesting to know what you could not do in
>TradeStation or QuantStudio that prompted you to undertake such a major
>effort."
>
>i havent found anything i cant do in tradestation ...[ some things werent
>easy but they werent imposible ......yet ].

Good heavens, from my first month or two of using tradestation I
identified a bunch of stuff I couldn't do:

== Backtesting of portfolios -- the single biggest weakness of TS

== Optimizing against any result *I* define, not canned results like
   net profit

== Executing trades on ANY price data stream, not just data1

== Backtesting of spreads determined by rules on the fly, between
   several data streams (i.e. you can't predefine the "spread data")

== Plotting indicators from strategies

== Double precision data types and double precision math

== Complex number data types, and complex math operators (not using
   user function calls)

== Data structures (makes passing groups of variables back and forth
   between functions a lot easier, especially if you have an array of
   groups of variables)

== Structured programming, including organizing my code into logical
   directories

== No decent way to break out of the middle of a loop (even a 'goto'
   statement would be useful for this)

== No way to operate on characters within strings

== No backtesting of combinations things displayed in different
   charting windows (you can do this in real time with PushPop DLL,
   but backtesting DOES NOT work with PushPop due to TradeStation
   not calculating each indicator in a synchronized way)

== Reliable pass-by-reference operations in function parameters
   (you never know when it won't work; i.e. a passed-by-reference
   parameter that changes within a function doesn't change in the
   calling procedure -- seems to be more reliable if all NumericRefs
   are in the beginning of the parameter list)

... and so on.

>i personally wouldnt swap to another platform unless the language
>was either 'as easy' as easy language.... or ummm maybe if it had
>some sort of EL-converter then i might consider it!

More powerful languages will by necessity be not as 'easy'.
However, something similar to EasyLanguage with extensions to the
language would be really nice.  Python is pretty easy too.

-Alex