Trading System Development Process: I'm interested in your experiences
Hi everyone,
first I'd like to say, that with this question I'm actually addressing to the more experienced traders around here - thank you for your understanding.
I'd like to share experiences concerning the development process of (profitable) trading systems. So far, I've got only one system that is allowed to trade with real money. I'm permanently working on my 2nd system, but it just takes soo much time... (Especially with MetaTrader, but that's another story....) Automated trading for me is one of the greatest endurance tests that one can find. I'm working my way through, getting from one disappointment to the next, for months. If there is a small success, it'll be smashed by a curve heading downwards at least a couple of sessions later. Experiencing this for not months but years - I start to ask myself if I'm maybe doing something wrong?
I often try to remember the process of how I developed my first profitable system, trying to find certain aspects of this process to adapt to the current one, however without success so far... Is it all about being lucky after all?
I'd be glad if you would share some of your experiences here.
;)
i think it is yr strategy, not devlopmental process
Hi everyone,
first I'd like to say, that with this question I'm actually addressing to the more experienced traders around here - thank you for your understanding.
I'd like to share experiences concerning the development process of (profitable) trading systems. So far, I've got only one system that is allowed to trade with real money. I'm permanently working on my 2nd system, but it just takes soo much time... (Especially with MetaTrader, but that's another story....) Automated trading for me is one of the greatest endurance tests that one can find. I'm working my way through, getting from one disappointment to the next, for months. If there is a small success, it'll be smashed by a curve heading downwards at least a couple of sessions later. Experiencing this for not months but years - I start to ask myself if I'm maybe doing something wrong?
I often try to remember the process of how I developed my first profitable system, trying to find certain aspects of this process to adapt to the current one, however without success so far... Is it all about being lucky after all?
I'd be glad if you would share some of your experiences here.
;)
Hi Friends,
My experience:
1. In 2008.01 I selected platform for automatic trading.
2. In 2008.02 I have started study of MQL4.
3. In 2008.03 I have started programming in MQL4.
4. In 2008.07 I have finished my first MQL4 program with 100% automatic multicurrency trading "A System: Original".
5. In 2008.08 I have decided to develop stable and reliable trading strategy.
6. In 2008.09 I have finished my first successful trading strategy "A System: Daily Metrics".
7. In 2008.10 I have started programming attempt to improve performance of this trading strategy.
8. In 2009.01 I have finished this attempt without positive results.
9. In 2009.02 I have started development of brand new trading system, based on Advanced Market Profiles.
10. In 2009.04 I have finished preliminary stage of this development project.
11. In 2009.05 I attempted to heighten cyber security level of my computer system.
12. Today I am learning to trade manually and preparing to finish development of my new trading system.
About plug-and-play trading software.
AIS1 Trading Robot works in Real Accounts Competition since 2009.02.16 without any human intervention.
MT4 login "1601573", password "ais11", server "212.158.161.8:443" or "75.125.134.162:443".
AIS1 Trading Robot uses trading strategy "A System: Daily Metrics".
Main performance metrics are:
1. life time = 0.3 year;
2. total trades = 42;
3. relative drawdown = 4.32%;
4. relative net profit = 16.76%.
Best regards,
Ais
I'd recommend the following:
0. Document a development lifecycle (entry and exit criteria out of each stage like the initial idea, the Excel mockup, the MQL mockup, the backtesting, the forward walk, the final EA etc.) and stick to it.
1. Keep your strategy simple so you can test using a simple tool such as Excel, avoiding lost development times; try to only code a strategy that you are fairly sure of.
2. Keep a decision log, so that you don't go around in circles in your strategy-definition process.
3. Build componentised MQL assets - avoiding monolithic code will help you speed up your development times as you "plumb" existing components and minimise new code.
4. Maintain a data dictionary showing range and usage of all variables, interface and behaviour for your functions etc.
5. Maintain a version history (and for each EA a set of release notes) showing improvements, bug-fixes, known issues and user instructions.
Hi! Thanks for your answers so far.
ronaldosim wrote >>
i think it is yr strategy, not devlopmental process
What do you mean with "your strategy"? By the time, I've implemented dozens of trading strategies, if that's what you mean.
n8937g wrote >>
.I think I may add another point....a very important point....There will never be a truely successful EA, there can only be a successful trader that uses an EA. You MUST be able to trade a strategy PROFITABLY manually and COMPLETELY understand an EA before you can be successful. A plug and play EA can never exist...stop looking for the impossible and do your homework...learn how to be a successful confident/ competent trader first....my .02
OK, I think I got where you're coming from. If I got this right, you're saying that you have to learn how to trade manually first and then automate this process with an EA. Honestly, I just don't have the time to learn how to trade manually. If you want to develop a profitable automated system, this consumes an unbelievable amount of time, that's just what I experienced. Second, I'm not really sure if this approach actually works at all. As far as I found out, manual traders always have a disctretionairy component in their systems, even if they call it mechanical (and even if they deny it...). And this small peace of humanity in a strategy may just be the "tiny" peace in the puzzle, that makes all go up, that makes the strategy profitable.
...
Hi ais, you already know that I really appreciate your work. A summary like this is what I missed the whole time... ;)
I'd recommend the following:
0. Document a development lifecycle (entry and exit criteria out of each stage like the initial idea, the Excel mockup, the MQL mockup, the backtesting, the forward walk, the final EA etc.) and stick to it.
1. Keep your strategy simple so you can test using a simple tool such as Excel, avoiding lost development times; try to only code a strategy that you are fairly sure of.
2. Keep a decision log, so that you don't go around in circles in your strategy-definition process.
3. Build componentised MQL assets - avoiding monolithic code will help you speed up your development times as you "plumb" existing components and minimise new code.
4. Maintain a data dictionary showing range and usage of all variables, interface and behaviour for your functions etc.
5. Maintain a version history (and for each EA a set of release notes) showing improvements, bug-fixes, known issues and user instructions.
You mentioned several really important points here. This is pretty much the way I handle it, except that I'm not using excel. By the time I accumulated a quite usable collection of frames and functions to be able to implement most ideas quickly.
Sometimes the main problem is when to drop a strategy and when to stick to it and invest more time in it.
I'd recommend the following:
0. Document a development lifecycle (entry and exit criteria out of each stage like the initial idea, the Excel mockup, the MQL mockup, the backtesting, the forward walk, the final EA etc.) and stick to it.
1. Keep your strategy simple so you can test using a simple tool such as Excel, avoiding lost development times; try to only code a strategy that you are fairly sure of.
2. Keep a decision log, so that you don't go around in circles in your strategy-definition process.
3. Build componentised MQL assets - avoiding monolithic code will help you speed up your development times as you "plumb" existing components and minimise new code.
4. Maintain a data dictionary showing range and usage of all variables, interface and behaviour for your functions etc.
5. Maintain a version history (and for each EA a set of release notes) showing improvements, bug-fixes, known issues and user instructions.
CB,
Really appreciate your shared information on the EA development process. When it comes to FX trading, I consider myself a trader first and a programmer a very distant second so that is why I appreciate your insight to the development process that a professional programmer would take.
I found your comment on using Excel interesting. I too use Excel extensively, but it was my inability to analyze market data "conditionally" that finally made me bite the bullet and learn MQL4. Actually, I was already headed in that direction, but it was the inability to perform complex analysis in Excel that pushed me to learn MQL sooner rather than later.
Now that I have an amateur's handle on MQL4, I code up small routines with a signal I'm considering and FileWrite market data subequent to the signal (say 40 bars worth) to a *.csv file so I can analyze it in Excel.
All of your comments are excellent. I've been coding for nearly 6 months and my MQL assets are starting to mount. I need to organize them as recommended so I can utilize them more efficiently.
Cheers
Sincerly I have doubt that a profitable automated system exist.
Don't you think that with all people working in banks on financial market, with all the power computer available now, and even with all these scientists in mathematics and physics who are working on trading systems, we couldn't find a system that works ?
Sincerly it's more profitable to sell books and pseudo working automated system.
I have been knowing financial market for 20 years and it's my opinion.
Welcome to the one that will boost morale of all those seeking the profitable automated system.
Wap
Sincerly I have doubt that a profitable automated system exist.
Don't you think that with all people working in banks on financial market, with all the power computer available now, and even with all these scientists in mathematics and physics who are working on trading systems, we couldn't find a system that works ?
It depends what you're expecting, and what "works" means. Outside of things like the Gann community, few people believe that systems work and last forever. Most of the people I know would estimate the lifespan of a successful system as a maximum of 10 years, and substantially less than that for some classes of system.
It also depends on what returns you're expecting. For example, almost all retail financial advice is based on a "system": go long on equities. This system has worked for a very long time (provided that you keep "trading" it for long enough), but its returns aren't very exciting.
Hi everyone,
first I'd like to say, that with this question I'm actually addressing to the more experienced traders around here - thank you for your understanding.
I'd like to share experiences concerning the development process of (profitable) trading systems. So far, I've got only one system that is allowed to trade with real money. I'm permanently working on my 2nd system, but it just takes soo much time... (Especially with MetaTrader, but that's another story....) Automated trading for me is one of the greatest endurance tests that one can find. I'm working my way through, getting from one disappointment to the next, for months. If there is a small success, it'll be smashed by a curve heading downwards at least a couple of sessions later. Experiencing this for not months but years - I start to ask myself if I'm maybe doing something wrong?
I often try to remember the process of how I developed my first profitable system, trying to find certain aspects of this process to adapt to the current one, however without success so far... Is it all about being lucky after all?
I'd be glad if you would share some of your experiences here.
;)
I've bean at it for about 2.5 years, developing a mostly DSP based EA (with some fractal idea's of my own on top). Currently trading with ~66% profitable trades / avg profit about 20% bigger then the losers. IE, not good enough, but on the right track maybe.
The way I see it:
1). There is no magic. Only math. (Repeat about 100 times...) So calculate, don't guesstimate. Guesstimating puts you in an endless loop, going nowhere.
2). Think out side the box. Most trader lose, so don't try to copy them.
3). A good trader isn't nesesarly a good EA programer. And a good EA programer may very well be a lousy trader.
Trading takes discipline, and that you like repeating doing the same thing over and over again. -You don't need that to write a good EA. Programing an EA is all about simplifying, IMHO more related to physics then any other science.
Just my 2 cents.
When I read this, I realized that "guesstimating" is probably the greatest mistake I'm doing at the moment and is what makes me take so long.
I guess I'im going to print this out and stick it on the wall above my monitors...
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Hi everyone,
first I'd like to say, that with this question I'm actually addressing to the more experienced traders around here - thank you for your understanding.
I'd like to share experiences concerning the development process of (profitable) trading systems. So far, I've got only one system that is allowed to trade with real money. I'm permanently working on my 2nd system, but it just takes soo much time... (Especially with MetaTrader, but that's another story....) Automated trading for me is one of the greatest endurance tests that one can find. I'm working my way through, getting from one disappointment to the next, for months. If there is a small success, it'll be smashed by a curve heading downwards at least a couple of sessions later. Experiencing this for not months but years - I start to ask myself if I'm maybe doing something wrong?
I often try to remember the process of how I developed my first profitable system, trying to find certain aspects of this process to adapt to the current one, however without success so far... Is it all about being lucky after all?
I'd be glad if you would share some of your experiences here.
;)