Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

Enter The Dragon - GBP/JPY 2 replies

Info on the Borihei-Dragon Scalping System? 15 replies

Jiva's 5,5 Trade Journal 5 replies

The Dragon 441 replies

***Dragon II*** 15 replies

Attachments: Jiva Dragon
Exit Attachments

Jiva Dragon

Quoting kauaibobby
i tried back testing and all i get in journal is cannot open file, hmmm.
i am sure my settigs arent the ways yours are set up.

Post your screen shot from journal.
EURGBP looks to be setting up long...
Attached Image (click to enlarge)
Click to Enlarge Name: eugb.gif Size: 29 KB
Optimized settings for EURCHF for Alpari... 1h

Before optimization...............5198.87Gross profit
After optimization................28805.63Gross profit
Attached Image (click to enlarge)
Click to Enlarge Name: eurchmtfoptalpari.gif Size: 8 KB
Attached File(s)
File Type: zip eurchmtfoptalpari.zip   1 KB | 177 downloads
EURUSD looks good for possible long
Attached Image (click to enlarge)
Click to Enlarge Name: eu.gif Size: 29 KB
Quoting kapybara
Post your screen shot from journal.
2010.10.25 22:12:03 2010.09.30 23:59
This is what is in the journal, over and over:

The time say ; 10-25-2010 I was testing back to 1-1-10

Cannot open file 'Crogram FilesTadawulfx Trader 4expertsindicatorsRobby DSS Bressert Colored with alert v1.1.ex4' on the GBPUSD,H1

Cannot open file 'Crogram FilesTadawulfx Trader 4expertsindicatorsT3MA.ex4' on the GBPUSD,H1
Quoting kauaibobby
2010.10.25 22:12:03 2010.09.30 23:59
This is what is in the journal, over and over:

The time say ; 10-25-2010 I was testing back to 1-1-10

Cannot open file 'Crogram FilesTadawulfx Trader 4expertsindicatorsRobby DSS Bressert Colored with alert v1.1.ex4' on the GBPUSD,H1

Cannot open file 'Crogram FilesTadawulfx Trader 4expertsindicatorsT3MA.ex4' on the GBPUSD,H1
Does your settings look like mine?
You have to check in Tools - Options - Experts tab
Attached Image
Quoting kapybara
Does your settings look like mine?
You have to check in Tools - Options - Experts tab
it is like yours except the two disable experts are not checked
Quoting kauaibobby
it is like yours except the two disable experts are not checked
Maybe try downloading from the indicators from the first page again.
Maybe something went wrong during download.
The ea needs a trending market. Otherwise it won't open any positions. that's good we don't want to get caught in a sideways market. I like that saying, "Better to be wishing you'd gotten in, rather than wishing hadn't gotten in".
so even though the pairs are dipping in and out of the hi lo channels it's all still sideways. The multi time frame moving averages will identify when there is actually a trend worth risking a trade. So maybe nothing for a few days. We can wait. And be ready for when it's right. We have lots to do to get ready.
Hi Jiva,

I do think what you are building here is great however I do have problems with the backtesting you are doing.

With only 50-60 trades over a year you can optimize to find the perfect combination of variables that gives you perfect upward sloping profit lines like the ones you have been displaying HOWEVER these results are not statistically significant because the 1yr window of data you have chosen to optimize on is not representative of all types of market conditions.

I realise it is no easy job to find high quality data to do longer tests over but to be a REALLY great EA the same .set file should be able to be run on any given year and although not be a star performer sometimes it shouldn't destroy the account either.

To illustrate this point please see the attached screenshot showing the backtest from Jan 2009 to Oct 2010 using your eumtf1hopt.set file which you optimized Oct09-Oct10.

You will notice that I had to run the backtest using an initial deposit of 100k rather than 10k as the 10k was wiped out before we got half way into 2009! It will come as no surprise that the period you optimized for was trade 66 onwards.

That doesn't mean to say the EA is not good - the problem is if you continue to best fit your variables over such a small timeframe then you will find a model that always does amazingly well if the market is VERY similar to what we have seen over the coming months but possibly wipes out your account if it does not.

The only solution is to optimize over the longest period you can and although the charts you produce won't look so beautifully smooth you should be able to have more confidence that you won't lose all your money either.

Good luck, I do think you are going down the correct route and the MACD may well filter out some of the mess - but try hard to capture at least the last 3 yrs in your optimizations and then hope that when you run that .set file over longer periods you don't get any surprises - then you will have a truly robust methodology.

Cheers

Jono
Attached Image (click to enlarge)
Click to Enlarge Name: eu.jpg Size: 52 KB
Quoting jm2110
Hi Jiva,...
Thank you. I totally agree. That's next.
Reminder: I hope no one is actually using real money at this point!!
We are definitely not ready for real money yet, as inviting as it may seem.
I have all confidence that we will be ready soon.
Whatever it takes, no stone unturned, we'll get there...
optimized settings for EURUSD Alpari 1h 12 months

Before optimization......6100.80Gross profit
After optimization........22236.60Gross profit
Attached Image (click to enlarge)
Click to Enlarge Name: euoptalpari.gif Size: 8 KB
Attached File(s)
File Type: zip eumtfoptalpari.zip   1 KB | 212 downloads
Quoting jm2110
Hi Jiva,...
This is a very central point about expert advisers. When the market conditions change they can lose their "wisdom" and eat an entire account balance in about 3 gulps.
so how to deal with this.
JM211o has very kindly brought this to the forefront while I am having a great old time optimizing perfect "curve fitting" settings. Something Taiyakixz has been hounding me about for months.
You are mostly "curve fitting" he chides.
Anyway, we all will agree there must be some "curve fitting", so I'm getting really good at!!!
But how do we deal with the inevitable, change in market conditions?
JM offers that backtesting and curve fittng for at least 3 years will offer some protection. The graph won't be so pretty, there will be dips but overall an account balance will be protected and most likely show some profit if the ea has any statistical edge to it at all.
At first glance this looked like the solution. But then my "evil" mind taunted, "There is no guarantee that simply going out 3 years would offer protection from changing market conditions. Maybe the market would change after 3 years and the optimized setting for 3 year period would become ineffective". In other words a curve fit for 1 year or a curve fit for 3 years or even 5 years would all have the same inherent short coming of becoming ineffective at some point. The problem is not knowing when that point of ineffectiveness occurs until the losses start!!!
So 1 year, 3 years, 5 years, 10 years!! All the same problem.
What to do?
What I am thinking, and I share it with the forum, is that market conditions most likely do not change overnight (of course we get those wild spike and dips but I mean serious change over an extended period of time), they begin to change gradually as the participants adjust to the new economic environments. So by performing optimizations regularly, on a daily basis would be superb... then whatever changes are occurring should be gradually incorporated into the optimized curve settings.
Not go out a million years, but zoom in on a day to day basis if possible, if not daily, then surely a weekly optimization should account for any changes in market conditions as they happen staving off losses.
If that were the procedure it might be possible to optimized for 6 months instead of 12 months and get an even higher profit curve. But 12 months seems reasonable.
6 months might be introducing a bit more of a risk factor than we might want. With higher profits comes higher risk. It's already risky enough!!!
But we can research it a bit at some point.
Any thoughts on this. It's really central to using an ea.
As an after thought. This makes sense, to perform the optimization regularly. The strategy is sound and will hold up in any market condition. This I am fairly certain of. And of course as we continue on this will seen either true or false. So if the strategy is sound then it is merely certain parameters that will change. The intensity of the of trend for instance, or the volatility.
One reason I believe the strategy is sound is that it has been applied to 10 different pairs that all trade slightly differently and with the right settings backtests profitably in all 10, and generously so.
If the pairs trade differently, and they do, and the strategy holds for all 10 pairs, that is it keeps its statistical edge, then it should hold in different market conditions also. And it would simply be a case of making slight adjustments as we go.
Maybe it is that traders want to set an ea up and forget about it. But maybe that is not the way to use an ea. Just like buy and hold stock investing is no longer practical. So set and forget may not be the best use of an ea. Maybe set and regularly monitor and adjust as needed is the best route. So set and forget for a few days maybe... But not set and forget for a year or 3 years!! Not if we want to make some serious money.
Quoting jiva34
As an after thought. This makes sense, to perform the optimization regularly. The strategy is sound and will hold up in any market condition. This I am fairly certain of. And of course as we continue on this will seen either true or false. So if the strategy is sound then it is merely certain parameters that will change. The intensity of the of trend for instance, or the volatility.
One reason I believe the strategy is sound is that it has been applied to 10 different pairs that all trade slightly differently and with the right settings backtests...
Absolutely agree.

I think in Forex doesn't exist system, which is announced and sold for bunch of money, which can be set and forgett.
Watching the trades and studying them in the back tests what seems to be happening is the ea does not necessarily catch the first break out in a long term trend. Or what might be seen in wave theory as the 1st wave, which isn't usually the biggest move.
The ea very often catches an entry on a pullback. And rides what can be seen as the 2nd wave or 3rd wave which is usually the stronger moves in a trend.
Because all the moving averages need to be lined up for an entry, it won't necessarily be able to enter on the initial breakout since the moving average on the high time frame hasn't had a chance to turn yet.
So it catches trades in the middle of the trend. Very good..... We want that.
Only weakness is sometimes it might buy a top or sell a bottom. I want to work with macd which we now have in v14, thank you, Taiyakixz, not posted yet but is in testing to see if we can stop that.

Since the market is a little sideways and retracing now we might not get any trades for a while.
Quoting jiva34
This is a very central point about expert advisers. When the market conditions change they can lose their "wisdom" and eat an entire account balance in about 3 gulps.
so how to deal with this.
JM211o has very kindly brought this to the forefront while I am having a great old time optimizing perfect "curve fitting" settings. Something Taiyakixz has been hounding me about for months.
You are mostly "curve fitting" he chides.
Anyway, we all will agree there must be some "curve fitting", so I'm getting really good at!!!
But how do we...
Have you considered automating a "built-in" dynamic set file to any EA?
What I suggest is as RR has shown in his "Old Dog new tricks" Thread? Traders agree (1) optimization is necessary to give best performance results, and (2) markets for each pair change over time, thus outgrowing the EA's. If an EA can be coded it would seem logical an internal module to any EA could contain capabilities to self optimize using various dynamic measurements. One approach short of neural network applications would be to adjust the architecture of EA to variable mtf look back tables (PA) from which calculations could maintain in-turn variable set files. This could be done weekly, or bi-weekly or monthly using a standard 5 trade day week. Updates could be run after market hours (weekends) or every fourth week, etc.

We have been fighting a losing battle (in most EA's) with static programming
following lagging indicators.....! If we moved from static setting parameters needing manual optimization to flexible set adjusting methods the EA's built on
approaches moving in concert with market price movements may be a worthwhile direction involving toward "real hands-off" or longer term non-intervention?

Just thoughts....
Atlas1
Quoting Atlas1
Have you considered automating a "built-in" dynamic set file to any EA?
What I suggest is as RR has shown in his "Old Dog new tricks" Thread?...
Yes this does seem to be the bug-a-boo.
Can you post any other relevant links... for discussion of this issue?? I will visit Old Dog...
Thank you...
Hello Jiva34 and everybody at the Thread this is my first post, I am a Newbie in Forex, I have been trading for less more than 6 months, and have been reading Forexfactory's Threads since 3+ months

I am very interested in your system and have started testing EA in EUR/USD H1 since yesterday, Demo account with Instaforex, Good results since then, I also tryed to backtest EA with "EurUsd" but no success maybe is because I run MT4 plataform inside Linux, anyway Thanks for your priceless Effort and your good will with all of us .

If there is a Way I can contribute this System and Thread I will , Please be so kind of pointing what I can be usefull to help. (I am graphic designer, Web and Linux programmer) maybe a "logo" for this incredible EA could help people here and elsewhere to trigger mentally here, just an idea)

_______________
Sequoia
Quoting kapybara
Absolutely agree.

I think in Forex doesn't exist system, which is announced and sold for bunch of money, which can be set and forgett.
I think set and forget IS possible, just very difficult.
0 traders viewing now