Jump to content

woodie robot


cbot

Recommended Posts

  • Replies 111
  • Created
  • Last Reply

Top Posters In This Topic

Re: woodie robot

 

I will put my testing here. It seems somewhere it is one loop too much in the Woodie_trend indicator. So my backtest on GBPJPY M15 data from 2009-07-01 to current 2009-08-21 on very powerful new computer, run for 24 hours, still not finished yet. later when it finished or I just cut it short, then I will post the settings here.

Link to comment
Share on other sites

Re: woodie robot

 

I will put my testing here. It seems somewhere it is one loop too much in the Woodie_trend indicator. So my backtest on GBPJPY M15 data from 2009-07-01 to current 2009-08-21 on very powerful new computer, run for 24 hours, still not finished yet. later when it finished or I just cut it short, then I will post the settings here.

 

we're waiting unpatiently dude :)>-

LAL: Laugh At Life

it's too short to be sad..

Link to comment
Share on other sites

Re: woodie robot

 

There must be one loop too much in the woodie_Trend indicator. my backtest from 2009-07-01 to 2009-08-21 is finally finished after 50 hours run on a very new dual cpu, 4G ram computer.

 

http://www.formulatrader.com/download/bild/woodieea_Tester.gif

 

Click here to download the who EA, my settings and backtest result

 

I have backtested it on the GBPJPY M15 2009-07-01 to 2009-08-21 on the real account data from atcbrokers. I do not know how much equity the EA needs, so I started the test from 100K USD. the result during the time is return in 3700%. the data quality is not at the best.

 

Tried a while on the EURUSD, does not give good result as the GBPJPY. Woodies cci EA is a trend follower, and I think it need the moment that GBPJPY have nowadays.

Link to comment
Share on other sites

Re: woodie robot

 

Click here to download the who EA, my settings and backtest result

 

 

In the file Woodies_the_robot_new_5and4digits.mq4 (to cater for 4 and 5-digit brokers), there is a code segment to calculate the point value:

 

if (Point == 0.00001) point = 0.0001;

else {

if (Point == 0.0001) point = 0.001;

else {

if (Point == 0.001) point = 0.01;

else point = Point;

}

}

 

should it be like below ?

 

if (Point == 0.00001) point = 0.0001;

else {

if (Point == 0.001) point = 0.01;

else point = Point;

}

 

(as Point == 0.0001 is for the old 4-digit price, and should not be changed to 0.001)

Link to comment
Share on other sites

Re: woodie robot

 

This version will speed up the EA. The problem was because the EA repeatedly calls the Woodie_Trend/Triple_CCI indis from the same bar whereas exactly same values will be returned, as these calls only ask for the values from past bars. Now, they are only called once at the bar open. The backtest on the GJ pair on the same period that took 50 hours now takes 5 minutes. The default settings seem to be profitable for GJ & UJ pairs.

Link to comment
Share on other sites

Re: woodie robot

 

Hi guys,

 

Thanks first of all for cbot for the great topic, and thanks as well for those who participated in sharing their opinions and modifying the EA.

 

I've read about this EA long time ago, but I still don't remember exactly where.

 

anyways, just few questions please to clear up everything and make it smooth for the newbies as well.

 

1- All EAs here are or can be used on 4 or 5 digits brokers, is that right??

 

2- Do we have include the indicators if we want to make a forward test??

 

Please let me know as soon as possible.

 

I hope we can share more opinions about this wonderful contribution.

 

All the best of luck.

 

Scarface

a New Year 2011 has come, and the challenge has just started 8-)
Link to comment
Share on other sites

Re: woodie robot

 

1- All EAs here are or can be used on 4 or 5 digits brokers, is that right??

2- Do we have include the indicators if we want to make a forward test??

1. Yes, to my posted WodiesRobotNew.mq4 & finimej's Woodies_the_robot_new_5and4digits.mq4

2. Yes, both Woodie_Trend & TripleCCI are required, for backtest and forward test.

Link to comment
Share on other sites

Re: woodie robot

 

http://www.formulatrader.com/download/bild/woodieea_Tester.gif

Good work with the backtest, though if there's an option to use fixed-lot position sizes you should choose that for backtests so that you can clearly compare drawdown periods at different dates; otherwise the early drawdowns appear small whereas the later drawdowns appear huge. In the equity chart you've shown, it appears that it doesn't go anywhere until much later, but it only looks like that because of the compounding effect (exponential growth). This problem can be prevented by using use fixed-lot position sizes in backtests.

 

I have backtested it on the GBPJPY M15 2009-07-01 to 2009-08-21 on the real account data from atcbrokers. I do not know how much equity the EA needs, so I started the test from 100K USD. the result during the time is return in 3700%. the data quality is not at the best.

Is it really data from ATC trade servers? I just tried to download data through their MetaTrader terminal and I get a warning saying the data is from MetaQuotes, so I doubt the data is actually from their servers. It would be awesome if we could get their actual trade server data though.

 

Is there a web site (other than this one) that discusses trading with this EA? It doesn't seem to be very well documented.

Link to comment
Share on other sites

Re: woodie robot

 

it's possible to speed up the EA execution and backtesting by include the Indicator code directly into the EA...

Yes, that is true, I've experienced big difference between using iCustom() and writing the code in the EA. If possible, iCustom() should never be used because it's so slow. The only case I think where it's necessary to use iCustom() is if you don't have access to the indicator code. But since we have the code, you should write the code into the EA.

 

It can take some work in translating indicator code into an EA, because you'd have to create arrays that hold the data (in place of the indicator buffers), and in some cases you may also need to create 2-dimensional arrays.

 

I'd try it but I'm deep into the Fractals 4 EA at the moment. The Fractals 4 EA used iCustom() for 3 different indicators, and some of the indicator code was already in the EA!

 

A lot of EAs out there haven't been programmed efficiently at all. I've spent so many hours fixing up EA code so that they run faster (both in live trading and backtesting). EA developers need to be aware that the slower an EA runs, the more incoming ticks that are discarded during live trading. This can affect the profitability of an EA in live trading, and lost ticks may possibly be one reason why profitability in live trading may be different from backtested profitability, because in backtests, all ticks are processed.

Link to comment
Share on other sites

Re: woodie robot

 

1. Yes, to my posted WodiesRobotNew.mq4 & finimej's Woodies_the_robot_new_5and4digits.mq4

2. Yes, both Woodie_Trend & TripleCCI are required, for backtest and forward test.

 

 

Thanks for your reply.

 

Maybe I wasn't clear enough. O.K, I mean by the 2nd question is that do we need to drag the indicators to the chart while backtesting???

 

O.K, how about forward testing?? is that the same?? I mean do we have to add the indicators to the chart or the EA is going to call the indicators???

 

One more thing, I setup the EA to both pairs GBPJPY and GBPUSD for 15TF and till now no one single trade, is that normal??

 

 

Please let me know if you have any idea about that.

 

Best wishes,

a New Year 2011 has come, and the challenge has just started 8-)
Link to comment
Share on other sites

Re: woodie robot

 

The woodie EA code is slow, because the woodie_trend indicator.

In the Woodie Trend inidicator the main function Start{} has a for loop that calculated for all the past bars on the charts. This is only the one loop needs.

 

Then the programmer in the subfunction, each subfunction, he unnecessary make a loop to check the past bars on the charts. This loops inside loop kills the backtest time.

 

//+------------------------------------------------------------------+
int DetectTrend(int shift, int l)
  {
  double BufferDown=0;
  double BufferUp=0;
 // while(BufferUp==0&&BufferDown==0&&shift<l-1)  //here we shall comment out the loop
   //  {
    // shift++;  //use the shift to replace i.
      if(iCustom(Symbol(),0,"TripleCCI",TrendCCI_Period,EntryCCI_Period,TurboCCI_Period,0,shift)!=0) BufferUp=1;
      if(iCustom(Symbol(),0,"TripleCCI",TrendCCI_Period,EntryCCI_Period,TurboCCI_Period,1,shift)!=0) BufferDown=-1;
     //BufferDown=
  //   }
  if(BufferUp==1) return(1);
  if(BufferDown==-1) return(-1);   
  }
  

 

I have no time until the friday. very busy week. So you have to help on yourself.

for mograst,

Here is the mirrored download at 4shared

Link to comment
Share on other sites

Re: woodie robot

 

The woodie EA code is slow, because the woodie_trend indicator.

 

Actually, it's not hanging in backtest due to that; test in visual mode, it just stops... From my tests, it seems to be a bug with the iCustom() MQL function and repeated recursion.

 

Due to this problem, I had to "split" my backtests and skip the dates with such errors (there were ~4-5 days across the 3 pairs I tested). I made a small script to "merge" all the test results into a single file. I tested it from 01.01.2009 up to 24.09.2009, timeframe M15, default settings on GBPJPY, GBPUSD and USDCHF. It's absolutely worthless on USDCHF so I didn't even merge the files from that. For GBPJPY and GBPUSD, it's profitable, but with rather large drawdowns (18% with 10k account size and 0.1 fixed lot size), as you will be able to see in the file linked below. Its strategy is dubious at best, it claims to trade trends yet very often it just misses a huge trend, not trading it at all and starting to trade as soon as the market starts moving sideways, after the trend is over. In some cases, it even trades against the trend repeatedly, resulting in multiple consecutive losses. Overall, although not a bad EA, I strongly doubt I would trade it on my live account. The resulted XLS file with the merged test GBPJPY+GBPUSD for 2009 is at http://[email protected]/file/127606541/a28e0618/.

 

Given the fact that it has a piss poor performance (managed to turn 10k into 7k) during 2008 on GBPUSD, which seems to be its favorite pair and given the existence of several settings to reverse the entry signals, I'm thinking that the EA is the result of the author's fiddling around with a few ideas, ending up in curve fitting the EA for 2009 and trying to sell it, which also explains why the results posted on his site are for very short periods of time (if you test it for a few months before his start dates, you will see it's not particularly profitable). I'm even slightly surprised that it worked okish after the release.

Link to comment
Share on other sites

Re: woodie robot

 

How closely does it follow the renowned Woodie strategy?

What is the author's web site? I tried http://www.yan_in.com.ua but it doesn't exist.

 

Also, for those who complain that backtesting is very slow, try using the Control Points model instead of Every Tick. It's not as accurate but at least you get some result. It looks like the statements that were posted were from testing using the Control Points model too, based on the red colored text.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.




×
×
  • Create New...