Have the skills, but lost and all alone.

Champion · Aug 30, 2005

Quote from syswizard:

Champion - don't discourage the boy....several hedge funds are doing quite well with fully automated approaches. (Of course, we have the venerable John Henry who began this mech-approach years ago and he is down HUGE lately).

No question however, we may be years (decades ?) away from the time when machines will consistently beat experienced human traders with razor-sharp intuition. IBM created "Big Blue" to beat chess champions, right ? Where's their "Big Blue" for beating TRADING champions ? Bottomline: THEY CAN'T DO IT....much harder to beat the market than to win at chess.

I liked your post syswizard.

Truth to tell the Holy Grail is a problem solving exercise that only the few, the one or two, solve and win. Reason: it does not require maths or programming. Its a research and testing approach wherein you ask the right questions to establish an accurate predicative numerical model. It is complex but not that complex.

Big error: starting with answers. You see it all the time here. Only one or two ever find the series of questions they need to answer.

syswizard · Aug 30, 2005

Truth to tell the Holy Grail is a problem solving exercise that only the few, the one or two, solve and win.

Amen Brother. I've been at it 3 years and counting...still awaiting that glorious piece of code that will immortalize me !
It's a humbling experience to try to "beat" the market CONSISTENTLY with "dumb" software.
I've got a client list of several who have been "leveled" over the past 3 years by using system approaches that were successful in 1999->2001.

Bitstream · Aug 30, 2005

Quote from syswizard:

Amen Brother. I've been at it 3 years and counting...still awaiting that glorious piece of code that will immortalize me !
It's a humbling experience to try to "beat" the market CONSISTENTLY with "dumb" software.
I've got a client list of several who have been "leveled" over the past 3 years by using system approaches that were successful in 1999->2001.

Yep, by the time the software becomes popular and gets to the masses it's already out-of-date.

jason_l · Sep 15, 2005

Don't get bogged down on any fancy GUI stuff if you're wanting to automate a mechanical strategy. As many will point out, finding and validating your edge will take every ounce of effort you have, you definately don't want to waste time on coding frills. The code could care less if there's any GUI window to it or not, so start with the minimum. once you have something profitable going, feel free to purty it up all you want

.

I would say if you go down this route, invest the time to build a good flexible framework for backtesting. I can code/backtest/monte carlo test a complete idea in a matter of minutes. Somthing as simple as a crossover system would take probably 30 seconds total for me to code and test. It just dumps the test results into files, which I then can open in excel if I desire to see things like charts, equity curves, etc.. Another 30 seconds, and I could have it trading live if I wanted.

maxpi · Sep 18, 2005

Quote from jason_l:

Don't get bogged down on any fancy GUI stuff if you're wanting to automate a mechanical strategy. As many will point out, finding and validating your edge will take every ounce of effort you have, you definately don't want to waste time on coding frills. The code could care less if there's any GUI window to it or not, so start with the minimum. once you have something profitable going, feel free to purty it up all you want .

I would say if you go down this route, invest the time to build a good flexible framework for backtesting. I can code/backtest/monte carlo test a complete idea in a matter of minutes. Somthing as simple as a crossover system would take probably 30 seconds total for me to code and test. It just dumps the test results into files, which I then can open in excel if I desire to see things like charts, equity curves, etc.. Another 30 seconds, and I could have it trading live if I wanted.

Exactly. Were I a real programmer I would do like that. As it is I can code up a system and test it in Tradestation's RadarScreen in minutes. It is flexible enough, I could test 20 systems on 1000 symbols if I wanted to. I copy the results to Excel for further analysis if it is necessary. If I want more speed I'm told I could link to Matlab for computations but for now it's fine.

Cheese · Sep 18, 2005

Quote from JMowery1987:

I have been programming C++/C# on and off for the past few years, and recently got into trading.

I just really need pointed in the right direction right now.

I hear about tradestation, I hear things about Wealth-Lab, but if I use wealth-lab I have to figure out a broker, and then how to connect to that broker and use data from the broker and from another source. It just sounds so confusing.

What is the easiest way for me to set-up within a few weeks, and get to programming my strategy, testing it, and trading it without having to buy extra services from data feeds and such.

I don't mind learning more, but I'd just like to see if this is for me.

I know it won't be easy, but I am just having a lot of trouble understanding how everything connects together.

I currently have Ameritrade as my broker so I guess I have to drop them. But then what broker should I go to?

Just needing a kick in the right direction to get started, then I can start understanding what is going on and how everything works.

Thanks in advance.

Short answer: you're never going to make a fortune by programming to trade markets.

stargrazer · Sep 18, 2005

Quote from Cheese:

Short answer: you're never going to make a fortune by programming to trade markets.

Can you expand on that somewhat? Does that mean:
a) programming won't make you money
b) trading won't make you money
c) automating your successful trading formula won't make you money
d) automating your untested trading formula won't make you money
e) all of the above
f) none of of the above
g) some of the above, and if so, which?

Cheese · Sep 18, 2005

Quote from stargrazer:

Can you expand on that somewhat? Does that mean:
a) programming won't make you money
b) trading won't make you money
c) automating your successful trading formula won't make you money
d) automating your untested trading formula won't make you money
e) all of the above
f) none of of the above
g) some of the above, and if so, which?

Fair observation.

Its g: a,c & d.

nitro · Sep 18, 2005

Quote from Cdntrader:

...As an avid cook who is noted for her creative dishes, I am still unable to bake a cake that is light, delicate and edible. The reason is that I am too creative to just follow the recipe. I want to add a touch of this and a touch of that and use all of my creativity in baking the same way that I do in the rest of my cooking...

My ex-wife never made any "recepie" the same way twice. What is funny is, the first time she tried it, it was great (probably because she was reading from a book and her "creativity" was reigned in to making it more salty or more saucy etc.) Then it usually deteriorated with each "creative" iteration

nitro

nitro · Sep 18, 2005

To the original poster:

It is possible, but you are embarking on the journey of a lifetime and I guarantee you you are looking for a needle in a mountaneous haystack. If you concentrate on Arb style systems you may get there faster. If you just go in for a general system that trades say SIF consistantly, be prepared for years of study and trial and error.

Throw out TS/WL and all the canned trading software and start thinking in C++/C#. I just saved you five years. Make checks payable to:

Finally, I will tell you this, when you go live because you think you got it, something funny happens that will send you back to the beginning. Then you will begin to know the "place" for the first time...

nitro