Phoenix is here!

This brings back memories when I was writing code in EasyLanguage for TradeStation. This was not on a professional level I assure, it was for me mostly. Back then forums like these were not as widespread so I was pretty much on my own. I was constantly making changes to MA's or RSI periods, not optimizing; I was looking for something that worked all the time no matter what. Incidently, as I recall I was working with intra-day data on the SP500, similar to what we use when trading FX. The main thing I found out was that the simpler I kept things, the better the result (backtesting) was. I was mainly using indicator conditions because I wasn't too 'good' at coding price action. But, if I can remember, I believe I stumbled on to something that only produced about 2 or 3 trades a week but the results were very good, results being win:loss ratio. This was on the SP500 ($500 a point), the dollar value return was just as good. But this was when the margin to trade the SP500 was something like $25,000. Uh, I was poor then and I'm poor now. That was more of a hobby and I had and still have that full time job away from home thing. EA's were not 'available' 10 years ago or if they were the real-time data and ability to trade that way in the first place was $$$$$. The point 'I'm trying to get to' is that I kinda burned myself out looking for that one thing that would fit, say, on a corn contract. Nope, what works for one market doesn't necessarily work on another. Different markets have different behaviors. Whether your trading something that has roots and hooves or makes a ching-ching sound in your pocket, they are different. Now, put a corn chart and a EUR/USD chart next to each other without any legends and they look pretty damn similar don't they? But, still they behave differently. I can't answer why the same EA shouldn't produce 'similar' results if the charts looks 'similar', they just don't. It was and still is very frustrating to write code (EA's) for hours on end and come up with, not nothing, because you always learn from doing. But, you do learn what not to do. Writing systems did teach me that just because an indicator is overbought(sold) that the market will turn the other way. Markets can stay overbought(sold) for a long time but indicators don't care. They are all formulated from price action. So, might that be the key??? Price action??? I think it's a lot more difficult to program the price action of a market (behavior) vs. 'IF RSI > 70 THEN SELL'. I applaud Hendrick for working on 'the project' and I doubt very seriously he came away with nothing. He learned something I gaurantee you that. Open up that bottle of red Hendrick and relax for a while. The markets will still be there when you wake up from your slumber. Hope to hear back from you soon.

Good Trading To All

Hendrick,

Development work is usually hard and contains quick and slow, progressing and retrograde sections. I am happy that you do not intend to give it up rather to develop something new and great.
You can do it!

Palmer, TraderSeven, Yardie and A1Cruiser:<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" /><o></o>

Thanks a lot for all your kind words. I will certainly not give up. I know that all things in live are subject to some kind of law: what goes up must come down and after every winter we can expect a summer. I'm absolutely certain that Forex is subject to its own laws. And that's my goal: study all aspects of Forex and find its "laws". Maybe I will succeed and maybe not. But a least I will try!

Although I think <?xml:namespace prefix = st1 ns = "urn:schemas-microsoft-com:office:smarttags" /><st1:City w:st="on"><st1lace w:st="on">Phoenix</st1lace></st1:City> will not become "the killer-EA of the century" I will still let it run on my demo-accounts so I can study the trades it made. I will keep you informed about its performance. Here is the statement for week 1.<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" /><o></o>

<o> </o>

The results are not bad at all.. But the graph is somewhat odd looking hehe up down up.

It's extremely hard to figure out something that is profiteble over a longer period of time. Thousands of people on all kind of fora are hammering the same question as you do.
Many knowalls know that it sucks because DD is to big or SL is to big and other crap. Kewl to read is books. But offering a solution.... It ain't gonna be easy. Not by a long shot.

Maybe you should go back to Firebird; locate it's weakness and strickly concentrate on that.....?
Or mix strong points of several EAs.

Anyway your Phoenix now officially is a real Phoenix.
It did well and it's burned to the ground now.
The myth states it rises from his ashes stonger than ever.
Make it happen!

Hendrick, i'll add my thanks to all the above posters. You have done a good thing by beginning this project. It stimulated some good conversations and many of us have gained much from the activity.

Best wishes!
Thom

Hi Hendrick,


it's really good that you put so much effort into Phoenix. I agree with you that an EA which always needs new settings isn't good(although self-forward-tuning EA are a nice thing), however I disagree with you that Phoenix has to perform the same with ALL pairs. Every cross has a certain character I think, so different permance is just normal. ;D

Some update on the the development of <?xml:namespace prefix = st1 ns = "urn:schemas-microsoft-com:office:smarttags" /><st1lace w:st="on"><st1:City w:st="on">Phoenix</st1:City></st1lace> version 4 (yes.. I killed version 3!). This version is working with a proper money management module (you can find an explanation here http://codebase.mql4.com/273). I like it because it keeps the MD as low as possible. It increases the number of lots (and your net profit) when the EA is performing well and decreases them when the EA is not performing as it should be.<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" /><o></o>

Version 4 is very different (maybe I should give it another name than <st1:City w:st="on"><st1lace w:st="on">Phoenix</st1lace></st1:City>). I have attached 4 reports that show you the importance of money management (all report 90% modelling quality):<o></o>

<o></o>

test 1 MM=true risk 0.02 decrease factor 3<o></o>

test 2 MM=true risk 0.02 decrease=false (MD will increase!)<o></o>

test 4 MM=false (more profit but MD will increase)<o></o>

test 4 MM=true risk 0.1 decrease 3 (only for the daredevils.. but look at the net profit!).<o></o>

<o></o>

Good to see you back Hendrick.
Although I feel MM/RM is a very important part of a strategy, I think it should not be used when testing an EA.
It just obsures results.
Any EA that shows profit without MM will improve when you add MM.
Just picking some MM isn't wise either. In most cases it will improve performance. But we are looking for peak performance.
There are many RM/MM models and settings. Only one works best with your EA/trading style.

Just a thought.

About Money Management:I have noticed that if we open position the same was as Main Trend goes to, we can increase our lot by some %, if we open it against main trend (Moving Average) we better keep it as default value. In addition we can increase our default lot size every extra $$$$ earned. I am playing with one of the multiple EA versions. Unfortunately it shows bad results on historical data. Like they say it empties all account...

Hendrick, downloaded the files and will dig in after my shift! As i post i have 10 copies of MT4 running different EA versions of FB and one other. Some wild times with the market in a trend the last couple of days!

Cheers,
Thom

Anyone read this article?:

http://articles.mql4.com/163

It made me very humble.

Hendrick is back! Hendrick is back! Hendrick is back! All hail Hendrick!!! The man, the one and only....Hennnnnnnnnnnnnnnnnnnnndrick!!!! Where's Phoenix?

Ha! What's up there dude??? Interesting read. I wonder how many of use are going to cut and paste that article. Hope you have been doing well. Have you woken from your slumber yet? Either way, hope your trading has been good to you, peace.

Hey Palmer!!

Slumber? Slumber??? Hey Palmer, you must be joking man! I was struggling and fighting <?xml:namespace prefix = st1 ns = "urn:schemas-microsoft-com:office:smarttags" /><st1:City w:st="on"><st1lace w:st="on">Phoenix</st1lace></st1:City>. Tough one this <st1:City w:st="on"><st1lace w:st="on">Phoenix</st1lace></st1:City>! But I will survive (is the key still under the mat?)<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" /><o></o>

<o> </o>

Yup, but I hate to say it....I threw the shrimp on the barbie' last weekend But, the seafood market is just around the corner!!!!

I read this article, it is very good. I even was in touch with the author of this article, he explained me why we are getting different results with different brokers. But it does not stop me from looking for optimal solution. I believe that as long as market is very changeable it worth to look for optimal parameters on regular basis or to make a set of optimal parameters based on current market status. I think in order to get the best result it is better to use Tenforex as a data provider (they provide real time high quality rates, sometimes several ticks a second, I run trial account with them, for those who is interested I can give one month rates) and MT as a decision maker. The only one problem is that I do not have a MetaTrader Server which cost about 40K (already 20K, I found one guy to share the price with, but it is still expensive) and would receive rates from Tenforex AS THEY IT IS. The more often you receive ticks the better decision you can take, even candle analysis works very well.<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" /><o></o>

My idea is to make a single control center for MTs of different brokers, using DLLs and primitive experts they would receive BUY/SELL commands from it. This would allow me to optimize expert only for one data provider. Why different brokers? Because they say: “never put all the eggs in one basket”….<o></o>

<o></o>

Hi All,

My greatest admiration and respect for your abilities, hard work, patience and generosity - especially Hendrick.

I have looked at and tested Firebird, and also Phoenix with very mixed results.

But, like you guys, I believe "the answer is out there."

I understand the various concepts you are trialling, but as you all know they will work sometimes but not consistently. I have always thought a large stop loss is very dangerous. And, if the system is left unattended a relatively large T/P is unrealistic too!

IMHO, perhaps smaller T/P & S/L because if you win on a reversal as you are trying to do, "a bird in the hand is worth two in the bush," but also if the reversal does not happen, you have minimised your loss?

Sorry if you've already "been there - done that"

Congratulations and thanks from an "admirer from afar . . ."

Kind Regards,

Grahame.
PS. My whole aim is to encourage you!

currently i'm using FB. and i was wondering, no one implement money management in it, modifying the s/l and t/p as it's in profit. after searching using the word money management, i found this. why didn't i read this thread?!!! i don't even know there's another EA (Phoenix) from FB.

good luck with the phoenix 4, hopefully, you'll post it here. i can't wait to test it

Hi Hendrick

Thanks a lot for all your efforts and sharing.
Could you tell me what the latest version of phoenix is so far and what the link for download is.

On the first page we only have version 1; you mentioned at some stage that all versions would be posted on the first page which is a great idea. Thanks again.

Howard