The weaknesses of software.

Why do I feel a pitch coming for some latest and greatest software coming from this guy?

 

Let me first say that I am not a salesman and you will not get a pitch about software from me. I endorse nothing.

I will also tell you that I use First Alert. I know it well. It has many strengths and many weaknesses. It's ability to screen the market for opportunities that meet your criteria is its greatest attribute. The load it puts on your computer and strange glitches still lingering from the DOS version reduce its reliability. I do know that their server and data sucks! see Bryan Roberts post

I do not use TS. I cannot use it b/c I don't want to use them as my b/d. Didn't the company quit supporting the stand alone version? It doesn't really matter either way. I don't think it is any better than FA.

There are weaknesses in FA and many other analysis programs in the ability to quickly go from signal to trade. Most of us (pro firm traders) use a separate application for our execution. This is very slow. Also, the execution platforms we use are not very good for managing multiple positions. For my personal style, I would like to have something that helps me manage 20-30 positions simultaneously. Does anyone know where I can find that?

 

Specific?

How about:
1. neural network routing which will choose the best route based on the time of day, traffic, the issue your trading, size available, execution speed (historically w/ that MM) compared with the other MMs.

2. Integration of analysis and execution---no Plugin or links. One platform (TS on steroids).

3. Multiple sources of data that are compiled and constantly check for speed and reliability.

4. Less than one percent downtime partly b/c of #3. (not quite the magic 9s) No bugs.

4. Automation and risk management functions that allow you to click a few buttons and go play 9 holes. This is due to #4.

Is this specific enough?

Sure some people probably aren't having "fits." Some people are perfectly satisfied with status quo. Some people seem to be entertained by playing PONG.

 

xicaju:

Some may be satisfied with what you call the "status quo". But, how can we grow unless others share knowledge? Before I had a hamburger, I thought bologna was pretty good. Before I had a steak, I thought a hamburger was good.

Not exactly having fits here, but thank you for being more specific with your explanations.

Oh... I kinda always got more of a kick out of Ms PacMan!!

Regards,

BSAM

 

xicaju: Nice list. (Sorry for poking you a bit. )

Were you implying that you've seen such integrated software, but are unable to access it now? Or was what you saw still a patchwork of stuff that might still be no better than TS with a lot of fancy DLLs? Thanks.

Anyone else know of any "TS killers" on the horizon? I guess we could flesh one out up here, but perhaps there are better places to direct such efforts. It's frustrating trying to get the current software vendors to make even the most minor improvements.

 

dkamp,

I'm a "fully involved" TS2000 user, but I do keep my eye out for the up and coming alternatives. (I don't have any of the many problems/bugs that other TS users are bothered with, and I think it's primarily a function of system resources and the OS.)

The three that I've been watching are Neoticker (www.tickquest.com), AmiBroker (www.amibroker.com), and Wealth Lab Developer. All of these seem to be backed by very agressive and responsive programmers who are making fixes and enhancements at impressive rates. I also understand that Behold! would fit in this category, but I don't have that much information on it.

All of these would of course have a learning curve with a good time investment. So, for my needs currently, TS2000 with freely available and easy to use .dlls has worked very nicely. There are idiosyncrasies, but I can easily work around those. I *do not* consider TS6 an alternative, in it's current form, to do any kind of serious backtesting and analysis because of the data limitations in the feed.

 

Thanks DaveN. I'll keep a closer eye on the mentioned products and report back here if I ever make the big switch from TS.

 

You know, there is one simple way to get the kind of software that will do absolutely anything you could possibly want. Write it yourself. If you are already spending time trying to learn easy language or any other scripting language, why not learn C instead. You won't have to put up with the restrictions of EL or anything else. Here is what you would need, and the aprox. amount of time it might take you:

1. Learn basic C. You could take a course at a community college, or simply buy a good book. I would recommend Deitel and Deitel's "C, how to program". Read the first 14 chapters that deal with C programming. Ignore the rest. This should take you not longer than 1 day per chapter ( I'd say probably 8 hours a day minimum.)

2. Buy Visual C++ (the compiler, that's the program you use to build the programs you write.)

3. Read the first 10 or so pages of the book that comes with the compiler, it will tell you how to compile a console application. In other words you'll learn how to make programs that run in an ms-dos window. The output will of course be in text, but you don't need graphics to backtest your trading ideas.

4. Take a look at the API documentation for one of the data vendors. A good example might be toolkit.taltrade.com. This explains how you communicate with the servers. You won't understand most of it without learning C (or some other language.) But after reading the C book, it should be fairly simple and shouldn't take you longer than a week to understand. There are examples of C programs that pull data off of the servers on the web site.

You don't need any math background for this, you don't need any background except basic computer literacy and of course some problem solving skills, but as long as you have enough of that to develop a trading strategy, you should have enough of it to learn basic programming. Some level of self-discipline is a must, you might come up against a brick wall and have to read some material 2 or 3 times to get it (such as the stuff on pointers, all though you won't need most of it to test your trading systems.)

So in less than a month you can do back testing without any of the limitations imposed on you by other people's software.

Of course this applies to back testing software, possibly to scanners if you don't mind interacting with something that's text based, no charting - you'll need to learn the Windows API to use graphics.

voodoo