|
PureBytes Links
Trading Reference Links
|
> > First, why is it so hard to create a good trading software. I know that
> > there have been a lot of them tried. Some on this list seem to know a lot
> > about programing, why haven't some of you done the job?
I have been working on a trading software product since December 1988. Full
time since Spring 1991. So, as someone who as actually done it (well, almost
done it, the product is still prerelease), I think I can speak with some
authority on this topic.
The recent posts from M. Simms and Cash were right on the mark. M. Simms said:
<<<
A basic trading platform using Java or even VBA within Excel is not that
tough..... what kills you is the extra "features" that basically become
"requirements"...once you see what you REALLY NEED to make it a winning
platform
- backtesting
- optimization with reporting
- performance analysis
- graphics and text drawing
- built-in programming language
- etc, etc.
>>>
This was my experience. The basic platform went easily and quickly. Then the
complexity blows up geometrically as you fill in the pieces. Cash went into
more detail. His analysis makes me feel better about how long this project
has taken me.
I have been involved with trading system programming for a long time, and
therefore have some inside knowledge. I have heard of many failed attempts,
and I am not only referring to the public failures everyone on this mailing
list knows about. I am also talking about private projects which go through
several restarts and change of programmers before anything is actually
accomplished, or fail when the programmer has already spend many times as
much time as expected and the programmer decides to cut his losses by saying
he is going to walk unless he gets more money.
Why? Well, programmers always underestimate, but for many reasons a trading
software application in particular seems to appear to be easier than it is.
Using myself as an example, it has taken me years longer than I would have
expected to get to where I am now. It is what the previous posts have said. A
basic environment isn't hard, but a full featured environment is very hard.
Furthermore, a commercial quality full featured environment is very very hard.
This is how I have addresses some of the problems Cash mentioned. Rather than
a lot of money, I have put a lot of time into it, and remained self funded
from custom programming. Also, I have not been in a hurry to make the product
shrink-wrap. Rather, I have spent years working with customers making fixes
and customizations as necessary, sending updates as needed. As far as quality
testing, I have managed to find customers interested enough in the unique
features the software offers, as well as my programming services, so that it
has really been customers who have done the testing, not me. I have instead
concentrated on the software features to attract customers who are willing to
be a guinea pig for a new feature.
This is the really tough one (quoting Cash):
> And it has to be all things to all people. You need things for
> equity traders, futures traders, options traders, etc. You need to
> be able to handle people that want intraday and swing trading,
> and long-term investing. You have people that want fundamentals
> and you have people that want technicals.
I have added feature after feature requested by one customer or another. No
matter how many features I add, the next person needs something else.
However, for me, I can see the light at the end of the tunnel. The current
software is only a subset of TradeStation (PowerST specializes in heavyweight
system testing, not real time charting), but it is reaching the point where I
expect there will be a significant marketplace for the current feature set.
It is a difficult application directed at a very specialized marketplace. It
isn't like a mass market application with literally millions of prospective
customers, which justifies 100 programmers working on the project. I wonder
how many programmers Microsoft throws at an application. I really don't know,
but I would guess 100 programmers and testers easily. Probably more.
Bob Bolotin
President, RDB Computing, Inc.
Developer of "PowerST: The Power System Tester"
www.powertesting.com
powerst@xxxxxxxxxxxxxxxx
847-982-1910
|