$$$Anyone interested in collaborating on Trading System Development$$$

[demarcog]

Graphical displays of indicators and the plethora of "trading systems" are certainly invaluable tools. The ultimate implied objective is net profitabiltiy, with minimal draw down, a far more difficult accomplishment.

Anyone who has spent time any significant amount tweaking and manipulating "systems" will soon realize that the more people working together the better.

We have an opportunity with the amazing expertise available here and the ability to "work together" via the net that can greately benefit all of us.

I have been backtesting every concievable "system" and indicator published here and elsewhere yielding a precious few that make money.

I would like to suggest setting up a group/file share designed exclusively for systematically testing and quantifing as many indicators and systems as possible. It's way too much work for any one person to do effectively and welcome any help.

Toward that end, if anyone has any suggestion on an overall architecture I would appreciate hearing them.

I currently have a well structured moduliarized "front end", which indicates buy/sell, graphs profit targets and trailing stops yielding accurate backtesting results.

I would like to use the efsexternal capability to call individual indicators/systems testing for trade triggers, ideally returning a -1, for short, 0 for neutral, and +1 for long.

If the indicator/system returns these values, then a back test could be easily run for a single indicator/system or multiple indicator/system.

Unfortunatley most of what is published requires more then a trivial rework to return a simple buy/sell and typically are standalone applications.

If anyone has done anything similiar or has any suggestions toward changing or modifiying this approach pleease let me know.

If there is interest I will setup a fileshare to contain the relevent EFS studies.

Sharing with others in this forum, however difficult it is to "divulge" our hard earned "secrets", has once again reinforced to me the principle of "karma" (talk about counter intuitive!). I've posted a few profitable systems, and not only were very few people interested enough to down load them, but the help I received by sharing a few "winners" was well worth the investment.

There can never be a holy grail in a dynamic market environment, having an arsenal of viable systems that work in different markets during different time periods, easily quantified by a quick backtest where experience and familiarity with the markets/systems will determine the What? When? and How? of system selection.

[jimsmithe]

Groups don't work.

Groups don't generate ideas.

Ideas come from individuals.

Keep it simple.

[sams]

jmsmithe,

you are a individual.

Is your system working?

I talked to a broker, he told me that 80% of the trader are loosing money. Are 80% all group members?


Anyway I think it´s a good idea from demarcog.

demarcog,
do you have a system only for back testing, or do you test also with the tick replay tool.
Because there is a big difference.

regards
sams

[demarcog]

Originally posted by sams
jmsmithe,

you are a individual.

Is your system working?

I talked to a broker, he told me that 80% of the trader are loosing money. Are 80% all group members?


Anyway I think it´s a good idea from demarcog.

demarcog,
do you have a system only for back testing, or do you test also with the tick replay tool.
Because there is a big difference.

regards
sams

I did alot of backtesting first. Recently I started doing tick replay to verify the results of the back test.

Depending on the strategy, sometimes I have to run the tick replay a few times to get the correct results, must be doing something our of sequence or something.

BTW, don't take too seriously the comments made by "jimsmithe" it sounds very much like another member who shall remain name less that was permanently banned from the bulletin board ...

[FX989231]

jimsmithe

I can not agree with your statement, it is a well known fact that 2 brains are better than one.

demarcog

If you set up something I will join and contribute in what ever way possible.

Regards

[arbifox]

I have been at this since December of 2002. I have an area in the fileshare where I have tried to post my progress time permitting. I still work for a living, so sometimes it is hard to get to post and keep the net updated. I read this post and it struck a chord. Maybe I can share some ideas here that can help.

It took me 1 year to fully master just the EFS side of things. How it works, the basic fundamentals. I think it took me another year to start to mess with actual systems and see how they worked. The down and dirty methodology of taking an idea and automating.

Then I had to take a year off or so for personal issues, and I have been back hard at this for the past year. I read the posts, I read tons of books and articles, and everyone claims that they have something which makes their system 100% flawless.

After all this time, this is what I can tell you I KNOW works.

My trading results improved only after I stopped trying to "tune" out every loser. Losing is part of all trading strategies, and the better you can track them and understand why they happen, the better the system you can define.

I divided my signals for getting in and out of trades into setups and triggers. This was huge. There are times in the market that you have a potential for a trade, but then you have to wait for a trigger. I started to keep these seperate, and then act only when they both followed specific patterns.

Time frames. I use fast ones, and then slow ones. I trade off the faster frames, and use the direction to trade from the slow ones. This concept, I can not tell you how vital it is, naturally, in my opinion.

Lastly, adjust your expectations to what the market has to offer. I trade primarily the Russell E-Mini now since it is the best suited for my means. I used to try to tune on the ES, but it was just too choppy and narrow for my taste. Once in a trade, there is a point at which I simply make the code say, "The profit is good, leave while you can." I don't wait forever, I take what I can get. I keep track of risk, and sometimes I use profit targets.

The net result is this. Many say it is not possible to have a winning system. I disagree. I think you can develop a system using MACD and Stochastics on normal charts if you like to, if you just stick to basics. Any signal can be profitable, as long as you manage the trade effectively.

I worked the earlier years on only signal generation, then I started to realize the signals are fine, it is my management which sucks. So the end result, is to have a blend of the both. Having that, and taking only what you need from the market, can give you a healthy living. Case in point, 1 point a day from the Russell is worth a 33% return per month on $3150 trading capital. So why swing for the fences?

If anyone reading this ramble is interested, I am more than happy to share ideas. Some of my work is restricted, but much of it I have already shared with the community.

All the best . . .

Thomas.

[DavidZ]

arbifox, I can definitely relate with you.

I like Emini Russell, on tiny Renko bars. ES was too choppy for me too.

I agree in separating functional area of code, it greatly improves code readability and helps me focus to improve that area. My "open a position' code is a very large switch/case structure where I try my ideas and move on, BUT can still come back and retest old efforts. An old 2 year idea that worked poorly at that time now works nicely, the rest of the code had gotten better ( various stoploss methods, profit targets ,etc).

I'm playing with a new charting/backtesting package and trying some simple stuff to learn the basics. After getting the strategy in and seeing some promise with a basic strategy I did my usual - 'lets experiment' , AB Z6. I was glad I was seated. Changing from a 5 min bar to 6 min, gave 28 % improvement ( 45 days). 6 was the best I found for my test. Quick test with eSignal (my big code, very similar but not an identical test) showed 6 min 18% better. So does timeframe matter -- YES. You may have a winning system but just not know it.

I generate profit targets to chart but really don't use them. The only exception is hitting 8 points on the Russell is an immediate exit.

[demarcog]

Originally posted by arbifox
I have been at this since December of 2002. I have an area in the fileshare where I have tried to post my progress time permitting. I still work for a living, so sometimes it is hard to get to post and keep the net updated. I read this post and it struck a chord. Maybe I can share some ideas here that can help.

Hey Arbifox,

Thanks for sharing your experiences. Can you provide a link to your fileshare I am very interested.

It took me 1 year to fully master just the EFS side of things. How it works, the basic fundamentals. I think it took me another year to start to mess with actual systems and see how they worked. The down and dirty methodology of taking an idea and automating.

Then I had to take a year off or so for personal issues, and I have been back hard at this for the past year. I read the posts, I read tons of books and articles, and everyone claims that they have something which makes their system 100% flawless.

After all this time, this is what I can tell you I KNOW works.

My trading results improved only after I stopped trying to "tune" out every loser. Losing is part of all trading strategies, and the better you can track them and understand why they happen, the better the system you can define.

I divided my signals for getting in and out of trades into setups and triggers. This was huge. There are times in the market that you have a potential for a trade, but then you have to wait for a trigger. I started to keep these seperate, and then act only when they both followed specific patterns.

How is a setup different from a trigger?

Time frames. I use fast ones, and then slow ones. I trade off the faster frames, and use the direction to trade from the slow ones. This concept, I can not tell you how vital it is, naturally, in my opinion.

I've tried using multiple intervals in my strategies but there was a forward looking aspect that greatly increased the profits reported in the backtested results as compared to realtime. How do you incorporate the larger timeframes into your strategies?

Lastly, adjust your expectations to what the market has to offer. I trade primarily the Russell E-Mini now since it is the best suited for my means. I used to try to tune on the ES, but it was just too choppy and narrow for my taste. Once in a trade, there is a point at which I simply make the code say, "The profit is good, leave while you can." I don't wait forever, I take what I can get. I keep track of risk, and sometimes I use profit targets.

The net result is this. Many say it is not possible to have a winning system. I disagree. I think you can develop a system using MACD and Stochastics on normal charts if you like to, if you just stick to basics. Any signal can be profitable, as long as you manage the trade effectively.

I also use MACD and Stochastics in almost all of my systems but
when backtesting using the standard buy sell signals found that they traded too often and also when the market is tranding strongly sometimes the signals are best ignored. How do you filter out the signals during a strongly trending market?

I worked the earlier years on only signal generation, then I started to realize the signals are fine, it is my management which sucks. So the end result, is to have a blend of the both. Having that, and taking only what you need from the market, can give you a healthy living. Case in point, 1 point a day from the Russell is worth a 33% return per month on $3150 trading capital. So why swing for the fences?

If anyone reading this ramble is interested, I am more than happy to share ideas. Some of my work is restricted, but much of it I have already shared with the community.

All the best . . .

Thomas.

As I said I'm interested in hearing whatever you are comfortable sharing...........Glen [email protected]

[demarcog]

Originally posted by DavidZ
arbifox, I can definitely relate with you.

I like Emini Russell, on tiny Renko bars. ES was too choppy for me too.

I agree in separating functional area of code, it greatly improves code readability and helps me focus to improve that area. My "open a position' code is a very large switch/case structure where I try my ideas and move on, BUT can still come back and retest old efforts. An old 2 year idea that worked poorly at that time now works nicely, the rest of the code had gotten better ( various stoploss methods, profit targets ,etc).

I'm playing with a new charting/backtesting package and trying some simple stuff to learn the basics. After getting the strategy in and seeing some promise with a basic strategy I did my usual - 'lets experiment' , AB Z6. I was glad I was seated. Changing from a 5 min bar to 6 min, gave 28 % improvement ( 45 days). 6 was the best I found for my test. Quick test with eSignal (my big code, very similar but not an identical test) showed 6 min 18% better. So does timeframe matter -- YES. You may have a winning system but just not know it.

I generate profit targets to chart but really don't use them. The only exception is hitting 8 points on the Russell is an immediate exit.

Davidz,

What do you mean by tiny Renko bars?

I've been trying to use Renko bars but going crazy with the drifting problem, how have you handled it?

[DavidZ]

tiny bars for Renko - since AB moves in 0.1 tick , small for me is 0.3 to 0.7 for box size.

Renko Drifting - great topic - I have to do a manual reload for each indicator in every pane; for 3 years now. From what I 've gathered in the forum, Alexis created the script. eSignal had Tick charting problems that we all waited months for that to be solved, thinking that should fix Renko also - since both are not locked to fixed time intervals.

some good threads - Renko & Tick issues:
http://forum.esignalcentral.com/show...=EDL#post89348

http://forum.esignalcentral.com/show...5&pagenumber=1

http://forum.esignalcentral.com/show...7757#post77757

http://forum.esignalcentral.com/show...&highlight=EDL

[demarcog]

Davidz,

Steve Hare was kind enough to write something (see renkodrift4.efs which checks for a difference between the currentbarcount recorded by esignal and an internal counter he maintains, and reloads when there is a difference.

I've been testing it for a few days and having some problems: occasionally esignal hangs up and I'm having another problem where a plotted value is still appearing incorrectly in the chart after the reloadEFS() and I still have to manually reload it. I'm not sure if it's a bug in my code and about ready to give up trying to figure it out.

He was nice enough to write this and I hesistate to impose on his kindness again so I'm trying to avoid asking for his help again and figure it out myself but there is alot more to EFS then meets the eye sometimes.

If you can email me at [email protected] when you get a chance there are some things I can go on at length about and don't think it appropriate for this forum....

thanks glen

[arbifox]

Take a look at the file share I have, it should answer most of your questions about time frames and such. I update the share when I have a chance. It can be found under Arbifox Trading.

http://share.esignal.com/groupcontents.jsp?groupid=9

I am not really a coder, and I think most of us can code our strategies. I think most are excellent coders, and poor at creating viable strategies. I focus on the blend of the both, and I have tried to post up there digests of what has worked for me and what has not.

Take a look around, it may be lots of reading, and when I have more time, I will try to write more articles that might be helpful to the community. I am just pressed for time, but I do hope to get some docs up there which can be used as a tutorial.

Multiple time frames . . . many are doing this and there are quite a few links on how to do this. The theory is that you use a fast time frame with many signals, and only take those signals which agree with the slower time frame.

All the best . . .

T

[demarcog]

Originally posted by arbifox
Take a look at the file share I have, it should answer most of your questions about time frames and such. I update the share when I have a chance. It can be found under Arbifox Trading.

http://share.esignal.com/groupcontents.jsp?groupid=9

I am not really a coder, and I think most of us can code our strategies. I think most are excellent coders, and poor at creating viable strategies. I focus on the blend of the both, and I have tried to post up there digests of what has worked for me and what has not.

Take a look around, it may be lots of reading, and when I have more time, I will try to write more articles that might be helpful to the community. I am just pressed for time, but I do hope to get some docs up there which can be used as a tutorial.

Multiple time frames . . . many are doing this and there are quite a few links on how to do this. The theory is that you use a fast time frame with many signals, and only take those signals which agree with the slower time frame.

All the best . . .

T

Can't wait to check your stuff out.....applied for access to you fileshare....

I agree with the multi time frame but ran into a little "forward looking" backtesting results where 90 percent of my trades were profitable using a simple stochastic signal on 3 or 4 time frames that tested like gang busters and realtime was a different story altogether....was really discouraged and gave up on multi time frame for a while, recently looking into it again...

thanks for all the help, it's greatly appreciated, when I find the holy grail...ish of trading systems (and trust me if it exists I will stumble on it quite accidentially eventually) you'll will be one of the people on my cc list along with Jason, Alexis, SteveHare, DavidZ and a few others, of course none of you will have the need or desire to trade something inferior to what you all probably use, but hey I need to at least keep the karma flow moving along....glen