Automated Trader vs Manual Trader

[MajorUrsa]

Quote from salvar:

You must believe in your system with enough conviction to trade it and at the same time be on the lookout for a system failure. Some people can't handle that.

Good observation, thanks. Not bad for a 1st post, I hope you keep them coming.

Did u start out as a programmer or as a trader? Are you using all proprietary stuff or some commercial platform? And can u recommend a good developer book for someone with a programmer background who knows about trading (like me)?

Thanks,
Ursa..

 

[eudamonia]

Quote from MajorUrsa:

Good observation, thanks. Not bad for a 1st post, I hope you keep them coming.

Did u start out as a programmer or as a trader? Are you using all proprietary stuff or some commercial platform? And can u recommend a good developer book for someone with a programmer background who knows about trading (like me)?

Thanks,
Ursa..

Thanks Major. I actually started off knowing some Java and C++ although I don't use that much these days except for some .dll stuff. I've been trying to revamp my brain to learn C#. The good news is that once you've learned one program you can learn another much easier.

In the beginning I wasn't much of a programmer or trader!

I primarily utilize TradeStation which is pretty easy to use. There are a lot of work arounds for TS's various problems (which are documented on their forums and elsewhere) and I use those as well (lots of .dlls). Also there are some other great platforms out there I'm considering (like NeoTicker or SmartQuant).

Building my own platform is something that could be done but I think that it is a lot of extra work. It really depends on your needs. For me TS is acceptable (with workarounds) and NeoTicker or SmartQuant would be even better (I will probably switch over in the next few months). SmartQuant has the advantage of open source code.

I do use proprietary testing software and I develop my own. This is really what separates the system traders from the wanna bees.This requires both extensive programming know-how and trading know-how. Mostly, I own and design software that helps me develop and test.

Specifically, this might mean Genetic programming, genetic algorithms, neural nets, parallel processing, walk-forward testing, statistical analysis software, and others. Remember these are just tools - you must apply them properly.

All of my systems go through a rigorous design and weeding out phase designed to reduce curve fitting, maximize profit, maximize trades (to an extent), and minimize drawdown. Additionally, I look at things like the regression of the equity curve from the mean (in laymen's terms a straight line).

I hate to say it but there are few if any developer books that I'm aware of that address all of one's needs. If you find one let me know. There are some good books for beginners (and specific books about advanced topics) but nothing for someone who's knows even a modest amount about programming and trading and who wants a whole system design approach. If you want specific recommendations let me know. Also I tend to find that the internet has more info that most of these books.

So what's an aspiring systems designer to do? I would highly recommend trial and error. Build systems, throw away 90% of them and keep 10%. Then build new and better systems. Practice programming. Approach systems design the way you would for a complicated program application. Thinking is more important than coding (but you already know this if you can program).

I hope that helps.

Edward

 

[nononsense]

Quote from salvar:


Currently I have 14 strategies that I consider profitable and I use in my automated trading. Every so often, depending on the strategy (anywhere from every 1-6 months) I have to re-tweak these systems. That alone is a lot of work. Occasionally, these systems break. That means system development is ongoing. Creating new systems is highly time consuming and even more time consuming is testing them for robustness.

I wouldn't dear to call those 14 strategies "profitable" given the information that "occasionally these systems break" and need constant "re-tweaking".

I agree with you that developing automated systems is a lot of work. I have only ONE SINGLE system but I worked about 10 years on it.

 

[5Pillars]

Quote from eudamonia:

Thanks Major. I actually started off knowing some Java and C++ although I don't use that much these days except for some .dll stuff. I've been trying to revamp my brain to learn C#. The good news is that once you've learned one program you can learn another much easier.

In the beginning I wasn't much of a programmer or trader!

I primarily utilize TradeStation which is pretty easy to use. There are a lot of work arounds for TS's various problems (which are documented on their forums and elsewhere) and I use those as well (lots of .dlls). Also there are some other great platforms out there I'm considering (like NeoTicker or SmartQuant).

Building my own platform is something that could be done but I think that it is a lot of extra work. It really depends on your needs. For me TS is acceptable (with workarounds) and NeoTicker or SmartQuant would be even better (I will probably switch over in the next few months). SmartQuant has the advantage of open source code.

I do use proprietary testing software and I develop my own. This is really what separates the system traders from the wanna bees.This requires both extensive programming know-how and trading know-how. Mostly, I own and design software that helps me develop and test.

Specifically, this might mean Genetic programming, genetic algorithms, neural nets, parallel processing, walk-forward testing, statistical analysis software, and others. Remember these are just tools - you must apply them properly.

All of my systems go through a rigorous design and weeding out phase designed to reduce curve fitting, maximize profit, maximize trades (to an extent), and minimize drawdown. Additionally, I look at things like the regression of the equity curve from the mean (in laymen's terms a straight line).

I hate to say it but there are few if any developer books that I'm aware of that address all of one's needs. If you find one let me know. There are some good books for beginners (and specific books about advanced topics) but nothing for someone who's knows even a modest amount about programming and trading and who wants a whole system design approach. If you want specific recommendations let me know. Also I tend to find that the internet has more info that most of these books.

So what's an aspiring systems designer to do? I would highly recommend trial and error. Build systems, throw away 90% of them and keep 10%. Then build new and better systems. Practice programming. Approach systems design the way you would for a complicated program application. Thinking is more important than coding (but you already know this if you can program).

I hope that helps.

Edward

Have you ever looked at www.smartquant.com ?

 

[romik]

Quote from WmWaster:

Thanks scalper21 for your great inputs.
Thanks to cashmoney69 too.

One of the nicest thing I found in automated trading is it can trade 24 hours a day non-stop. I can earn money even if I'm sleeping.

In addition, it's nearly effortless after you have developed your system. It is run automatically. You don't need to spend efforts as manual traders have to consistently (although automated trader may need to maintain their systems once in a while).

As a retail trader being a "one-man-shop", relying solely on automated trading is too risky.

 

[WmWaster]

Actually what're the requirements of being an automated trader?

I'm not good at programing.

Does it mean I couldn't create a porfitable and long-term winning system?

Anyway, I have to try and see if automated system is viable to me.

 

[WmWaster]

Also for those who care, there are a few people selling systems that are legit. But they are half as good as anything I can come up with and probably 1/4 as good as someone who really knows something about automated trading.

Where're they?

How do they perform?
Returns Rate? Drawdown? Losing streak?

I would like to know because I could see how a purchasable (but legitimate) system is better than mine.

On collective2 (which seems to be a trustworthy website), it allows different system sellers to post their trades in real-time and offer table comparisons.

There's a system called Timac which achieve 1155% in 46 weeks (ie nearly 1 year).
However it just costs $150/month. How could that be possible? What're the purposes of the system author?

There's another system called future-pro which achieve 299.2% in 37 weeks (about 8-9 months).
Surprisingly, it is free to use. However the author said it may charge soon.

So here's the fastest way to get rich:
1) grab the few most profitable systems in the public
2) get some money and run them all together
3) wait for huge rewards

For people who are desperate to succeed, it might be okay to try one's of the free but successful systems.

 

[nononsense]

Quote from WmWaster:

[]

There's a system called Timac which achieve 1155% in 46 weeks (ie nearly 1 year).
However it just costs $150/month. How could that be possible? What're the purposes of the system author?

There's another system called future-pro which achieve 299.2% in 37 weeks (about 8-9 months).
Surprisingly, it is free to use. However the author said it may charge soon.

WmWaster,
You seem to be a fast learner coming up with the above question.

Now think twice. If you would have a system making 1155% in 46 or (only) 299% in 37 weeks, would you make this available for $150 a month?

By now you must have found the answer.

 

[scalper21]

Before anyone gets the notion that automated trading is easy or cheap read the following article.

http://elitetrader.com/vb/showthread.php?s=&threadid=72250
http://www.newsday.com/business/ny-bzsimons0705,0,3996439.story?page=1&coll=ny-homepage-mezz

Renaissance has invested more than $600 million in developing Simons' elaborate computerized system, an approach that has become somewhat the rage on Wall Street. As Institutional Investor recently noted, "At a time when big Wall Street firms are paring back budgets and covering fewer stocks with smaller rosters of less-experienced analysts, research firms that analyze shares using algorithms and computers rather than armies of humans are rapidly pushing to the fore."

Quote from WmWaster:

Where're they?

How do they perform?
Returns Rate? Drawdown? Losing streak?

I would like to know because I could see how a purchasable (but legitimate) system is better than mine.

On collective2 (which seems to be a trustworthy website), it allows different system sellers to post their trades in real-time and offer table comparisons.

There's a system called Timac which achieve 1155% in 46 weeks (ie nearly 1 year).
However it just costs $150/month. How could that be possible? What're the purposes of the system author?

There's another system called future-pro which achieve 299.2% in 37 weeks (about 8-9 months).
Surprisingly, it is free to use. However the author said it may charge soon.

So here's the fastest way to get rich:
1) grab the few most profitable systems in the public
2) get some money and run them all together
3) wait for huge rewards

For people who are desperate to succeed, it might be okay to try one's of the free but successful systems.

 

[makcim]

Hi, all.

I have spent months trying to create an automated trading system (mainly interested in index and currency futures, since they are quite volatile).

I have:

1. months of intraday market data
2. a simulator, which I wrote in perl

I have tested all kinds of strategies including but not limited to: market making with protective stops, entry stops with take profit limit orders, trailing stops of all sorts, order entry based on the standard deviation of X number of last prices, etc..

None of the strategies above gave me a positive result. I'm not sure how to proceed.

Does anyone have any suggestions?