Mechanical Trading versus Discretionary Trading...Opinions?

Originally posted by hmap1
Do you do most of your programming of automated systms in VB or C++? Most of our systems start out in VB and end up in C++ once all of the bugs and parameters are bugged out. One piece of a system in C++ can usually run 30,000 lines of code. This makes great for debugging....:)
More...

C, interfaces in VB.
 
Originally posted by metooxx


Yes, full time. We spend more man hours on system development than man hours for traders. We tend to test with real money after a basic idea has been developed. With the type of trading we do real time execution is the unknown and can not be simulated or back tested.

We trade equity options, futures and equities.
More...


In my opinion this is beyond black box trading because the people that are trading with a black box system have nothing in the background. And as I can see you have a huge capacity that is responsible for what happens "in the box" :)

May I ask what kind of returns you make? By the way are you managing part capital of a hedge fund or is this capital from a trustee?

alain
 
Well, from my perspective, we are not premitted to discuss our systems because basically, they work and we don't want to jeapordize the revenue of our company. The returns are basically amazing. Some systems are simple, some are much more complex. The best systems work on a basic level with no curve fitting. Enhancements, when done well, only limit your risk. Risk is the most important element. Profits are only a small piece of the pie. As far a successful systems, I disagree with an earlier post on a board where someone said that a 50/50 system mean't that the system was a coin flip. The reality is that a system that is 40% winners and 60% loosers can be incredible, when your average win is say $500 and your average loss is $20. Combine that with many trades a day.
 
Originally posted by alain



In my opinion this is beyond black box trading because the people that are trading with a black box system have nothing in the background. And as I can see you have a huge capacity that is responsible for what happens "in the box" :)

May I ask what kind of returns you make? By the way are you managing part capital of a hedge fund or is this capital from a trustee?

alain
More...

Whether it is called a "black box" or not is not really relevant; the key point is we blindly take all opening trades that a system generates. The negative to this type of trading is you are always adding "idiot filters", whenever you take a huge hit from a trade that no sane person would have done, but it happened to meet the pattern match you were trading. At that point it is up to the trader to clean up the mess; and that is what they get paid for.

The type of returns that the theorist would say are impossible because they think you can compound the return into infinity and corner the market; without looking at operating costs or the fact that a particular game has a finate number of trades per month and it is not a function of capital.

We trade our own capital; private trading company.
 
Originally posted by hmap1
Well, from my perspective, we are not premitted to discuss our systems because basically, they work and we don't want to jeapordize the revenue of our company. The returns are basically amazing. Some systems are simple, some are much more complex. The best systems work on a basic level with no curve fitting. Enhancements, when done well, only limit your risk. Risk is the most important element. Profits are only a small piece of the pie. As far a successful systems, I disagree with an earlier post on a board where someone said that a 50/50 system mean't that the system was a coin flip. The reality is that a system that is 40% winners and 60% loosers can be incredible, when your average win is say $500 and your average loss is $20. Combine that with many trades a day.
More...

I agree; we win 40%, lose 60%; however are losers are 85% scratches, 15% losers and we have very low transaction costs.
 
I was the one that brought up the subject with the 50% to 50%. You did understand me little wrong. I meant if the profits are equal to the losses. I know that the best trend following systems have 35 to 45% probability for choosing winning trades. With the 50to50 I tried to explain something else. Anyway it's maybe because of my language because English is my third language and I rarely speak it.. only writing.

alain
switzerland
 
Originally posted by alain
I was the one that brought up the subject with the 50% to 50%. You did understand me little wrong. I meant if the profits are equal to the losses. I know that the best trend following systems have 35 to 45% probability for choosing winning trades. With the 50to50 I tried to explain something else. Anyway it's maybe because of my language because English is my third language and I rarely speak it.. only writing.

alain
switzerland
More...

I agree.
 
Originally posted by alain
" I was the one that brought up the subject with the 50% to 50%. You did understand me little wrong. I meant if the profits are equal to the losses. I know that the best trend following systems have 35 to 45% probability for choosing winning trades. With the 50to50 I tried to explain something else. Anyway it's maybe because of my language because English is my third language and I rarely speak it.. only writing"

alain
switzerland
..........................................

I agree too.
I am a TS tester because I have to correct ,in realtime, wrong decision making action.
By my experiences (only in Italy) TS are great to capture strong trends, and errors may occur with fast in/out trades.
For these reasons I love TS !

About commercial TS : there are about 3 or 4 very good TS but you need to optimize them. You need a 3 persons staff to work properly in optimize process.


have a great gain!
 
Discretionary trading? When a decision is made to enter a trade, the trader takes multiple factors into consideration and makes the trade. The traders mind has flittered through a few rules he has made up based on experience. The trader acting as a neural net, mechanically runs through his rules, and enters the trade. My guess is that most day traders have fewer than 20 rules they check before entering a trade. Is the broad market up or down (NAZ 100), has there been a breakout, was there positive or negative news.
Each of these items can be programmed and made into a Mechanical system. Any trade filters can also be programmed. Conditions relative to the current market can be programmed. I postulate that any Successful Trader can be modeled and programmed, provided the data they view can be input into a computer.
The exception would be floor traders that watch the twitch of a Major Player, or bid according the the volume of the drone in the Pit. These things could not be programmed.

See www.Korns.com:
These folks have modeled some traders:

"Agents with the domain expertise of published experts:
Ben Graham
Warren Buffet
Peter Lynch
Agents which attempt to mimic mutual funds:
Magellan
Fidelity
Kaufman
A market demon which tries to ruin the other agents.
A game theory agent which tries to beat the market demon. "
 
I agree with you 100%, TechGuy.

The problem with discretionary trading is that sometimes the "discretion" is not consistent. If the discretion (="instinct") is consistent, then you can probably code it. The advantage of a mechanical system is consistency - and I think this is one key to profitable trading. I have my biggest drawdowns whenever I try a new sytem.

Great backtesting results of mechanical systems is not a guarantee of profits but at least it will give you an idea of how the script behaves. You can find a system which suits your trading style by backtesting. I dont like big drawdowns, I dont like a string of consecutive losses, I like a big win/loss ratio and need only a small profit per win (no "let profits run" but just "sparrows taking bread crumbs" style) - so these are the backtesting results I look for in a script. WealthLab is great for showing these.

pretzel
 
You must log in or register to reply here.
Back
Top