scooby-doo BOJ JPY pairs trading robot | Page 338 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Trading correlation pairs by using the other pairs 843 replies

Export All pairs or selected pairs to .csv with script 3 replies

Pairs of Currency Pairs 4 replies

Trending Pairs / Ranging Pairs 0 replies

Robot trading 120 replies

Attachments: scooby-doo BOJ JPY pairs trading robot
Exit Attachments

scooby-doo BOJ JPY pairs trading robot

  • #6,741
  • Feb 25, 2010 4:33am Feb 25, 2010 4:33am
Quoting SteveHopwood
Well, I did code it without prompting, so it would be a bit strange if not.

Default retracement appears to kick trades in a bit too soon and started losing money immediately, so now I am demoing defaults except for UseH1ATRrt = true. This has started out much better - several open trades, no losses, a couple with 50% closure and stop moved to be, a decent amount of draw up; early days.

An unsure start but seems to be doing better now. Very early days. In 2 days had 11 closed trades. You can see the results for yourself.
Kind regards
Attached File(s)
File Type: pdf Scooby-doo Trend Retracement statement.pdf   57 KB | 497 downloads
  • #6,742
  • Feb 25, 2010 4:38am Feb 25, 2010 4:38am
Quoting scooby-doo
Checking it every 12 hours wont work either as it can get updated quickly.

What I would do is change the news filter so that if it cant open the file then assume there is news. Better to be safe than sorry.

Scoobs.
Yep, sounds good.

How is it checking news right now? If the setup to open a trade is in place, it then checks the news as a final filter for each individual trade candidate?
  • #6,743
  • Feb 25, 2010 4:39am Feb 25, 2010 4:39am
Quoting astral77
An unsure start but seems to be doing better now. Very early days. In 2 days had 11 closed trades. You can see the results for yourself.
Kind regards
Bumping the retracement requirement up a level has worked well since yesterday.

One thing I have noticed is that the trend can change to 'flat' whilst a trade is open. Since the setup that caused the trade is no longer valid, then it needs closing. This is an enhancement I shall add in time for markets opening on Sunday night.

  • #6,744
  • Feb 25, 2010 4:54am Feb 25, 2010 4:54am
Quoting SteveHopwood
Bumping the retracement requirement up a level has worked well since yesterday.

One thing I have noticed is that the trend can change to 'flat' whilst a trade is open. Since the setup that caused the trade is no longer valid, then it needs closing. This is an enhancement I shall add in time for markets opening on Sunday night.

I had it on the default setting. changed the H1ATRrt to TRUE as suggested. I see at present it has few xxxjpy open trades. The trend for these pairs has been down for the last 48 hours and it seems BOJ is allowing the YEN to find its new higher level. It would be interesting to see what SCOOBS thinks regarding xxxJPY.

Kind Regards
  • #6,745
  • Feb 25, 2010 4:56am Feb 25, 2010 4:56am
Quoting scooby-doo
Checking it every 12 hours wont work either as it can get updated quickly.

What I would do is change the news filter so that if it cant open the file then assume there is news. Better to be safe than sorry.

Scoobs.
I witnessed a problem yesterday with the FF News Calandar, the fix may be of use (I assuming here that the fix hasn't been implemented within any of the excellent EA's on this thread).

1. Within about 30 minutes of news, my EA (not one from this site) was reporting the impending News correctly.
2. At some point closer to the News time, the FF servers must have become heavily used (as previously reported on this thread).
3. When step 2 happens, a return code of -99999 is returned from the FF indicator (or equivalent)
4. If the returned value overwrites the previously valid values received for the News, this effectivly renders the News Calendar useless.

I've now changed my code as follows:

double temp1 = iCustom(Symbol(),0,"FFCal",IncludeHigh,IncludeMedium,IncludeLow,IncludeSpeaks,true,0,true,-1,-1,ReportAllForUSD,1,1);
double temp2 = iCustom(Symbol(),0,"FFCal",IncludeHigh,IncludeMedium,IncludeLow,IncludeSpeaks,true,0,true,-1,-1,ReportAllForUSD,1,0);

if (temp1 != -99999)
{
MinutesUntilNextEvent = temp1;
news1 = DoubleToStr(MinutesUntilNextEvent,0);
}

if (temp2 != -99999)
{
MinutesAfterPrevEvent = temp2;
news2 = DoubleToStr(MinutesAfterPrevEvent, 0);
}

if ((MinutesUntilNextEvent <= MinutesBeforeNews) || (MinutesAfterPrevEvent <= MinutesAfterNews)) NewsTime = true;

So, I change the value of my key variable MinutesAfterPrevEvent ONLY if there is a non -99999 value being returned from the News Calendar indicator. This preserves the previously correct information coming from the News Calendar.

I'm not sure if your EA's are coded like mine was, but thought it may be useful to pass this on.

Cheers,.
Nick.
  • #6,746
  • Feb 25, 2010 4:59am Feb 25, 2010 4:59am
Another basket win last night the 2nd closed in 5 min
FXDD Live
Scary
  • #6,747
  • Feb 25, 2010 5:16am Feb 25, 2010 5:16am
BOJ Original 2.4 is available from post 1, with different inputs for OB and OS time limits.

  • #6,748
  • Edited 7:37am Feb 25, 2010 6:26am | Edited 7:37am
Quoting zoogle
As for your cad/chf position, I don't see how that could have been a valid sell since the h01 ema has been above the h04 ema since jan. 26th meaning we only buy.
LOL, that's a good point and looks correct, but it's simply the last trade that the bot (your version of Scooby's 20 ema - 200 ema etc. EA) took. No discretion here and not a favourite trade or pair of mine: the system simply took the short while price was > 200 ema on 1 and 4 hours. No data problems here and the EA is running: my ema's agree with yours. No idea why, but it took the trade and I just used it as the most recent example

The irony is that all of this 20/200ema and RSI filtering on MTFs generally in this and other systems may mean diddly squat compared simply with entering far enough away from the daily open (% of daily ADR), fading it, so that the current price move is objectively OB/OS (exhausted) and we can trade the mean reversion back somewhat toward the daily open whether it's in the direction of the 20/200 ema's on any TF or not. As long as we can somehow get TP > SL realistically, without having a narrower SL repeatedly hit (as happens in just about all EAs), this will be a positive expectancy system.

I agree with your change to the TP and SL values: I just don't know if the SLs are repeatedly going to be hit (before trades have a chance to become winners), because this is not an "accurate" system. Am watching on demo to see how a narrower SL does. One thing I know for sure: any system with a SL 3 * the TP is an account killer. No need for testing. When the TP/SL ratio cany vary by hundreds of % (in this case 300% on the default with TP set at 0.25 ADR and SL 0.75 ADR), as it can vary in any sys., it always swamps accuracy and it makes little practical difference whether the system is 50% or 90% accurate (latter in any case is impossible to maintain for any system with winner dollar size > loser dollar size per trade; you will find that any "accurate" system has avg. losers $ size > avg winners $ size over any reasonable sample size or length of time, and so = net losing/negative expectancy system). Key system variable is avg. dollar win vs. avg dollar loss accounting for both accuracy (not very important) and TP/SL ratio (absol. crucial).

If we can't get TP > SL reliably, then it means the TA basis of the system has no validity/no edge (there's no valid info. content/signal value being added by the TA variables in the system). We might as well be tossing a coin. The irony with the CADCHF trade above is that entering at circa 44% of ADR (=88% of ADR, or de facto near 100% of ADR, on a bi-modal price distribution day around the day's open) was a winner). It was irrelevant where price was in relation to the 20/200emas, RSI's whatever, on MTFs. What matters is entering at a price where the TP/SL ratio can realistically be set > 1.

The only thing trade frequency and "accuracy" here are doing is enriching the brokers and their clearing banks. They love bots that do this and encourage trading that does this.
  • #6,749
  • Feb 25, 2010 6:41am Feb 25, 2010 6:41am
The problem with the news indicator stems in the fact that it gets loaded and reloaded constantly. Each time this happens it has to download the news xml file from the FF site, when high impact news occurs the file can't be retrieved and 99999 is returned which by our EA's gets picked up as no upcoming news.

The fix is to turn the indicator into an include that will be loaded with the EA and stay loaded, then let it determine once every four hours if it needs to grab the xml file again. This is how it behaves when it is used as an indicator, which works quite nicely I might add. (Here we are again with the iCustom issues)

If you look closely, these strings of losses occur around major news events and happen in just about all of the EA's here.
  • #6,750
  • Feb 25, 2010 6:42am Feb 25, 2010 6:42am
Here goes.
Attached File(s)
File Type: pdf The Holy Grail v1.9.pdf   89 KB | 609 downloads
  • #6,751
  • Feb 25, 2010 6:45am Feb 25, 2010 6:45am
Quoting scooby-doo
Here goes.
Woowee! That looks promising
  • #6,752
  • Feb 25, 2010 6:50am Feb 25, 2010 6:50am
Quoting zoogle
The problem with the news indicator stems in the fact that it gets loaded and reloaded constantly. Each time this happens it has to download the news xml file from the FF site, when high impact news occurs the file can't be retrieved and 99999 is returned which by our EA's gets picked up as no upcoming news.

The fix is to turn the indicator into an include that will be loaded with the EA and stay loaded, then let it determine once every four hours if it needs to grab the xml file again. This is how it behaves when it is used as an indicator,...
Wouldn´t consider a 99999 retrieved value as being equal to news=true be an easier way out, as Scooby proposed earlier? I´m no programmer, so cannot really tell...
  • #6,753
  • Feb 25, 2010 6:55am Feb 25, 2010 6:55am
Quoting Middleastern
Wouldn´t consider a 99999 retrieved value as being equal to news=true be an easier way out, as Scooby proposed earlier? I´m no programmer, so cannot really tell...
You also get 99999 if there are no more news for the week.
  • #6,754
  • Feb 25, 2010 7:26am Feb 25, 2010 7:26am
Quoting SteveHopwood
BOJ Original 2.4 is available from post 1, with different inputs for OB and OS time limits.

Cheers Steveo
  • #6,755
  • Feb 25, 2010 7:28am Feb 25, 2010 7:28am
Quoting scooby-doo
Here goes.
Right on Scoobs!
  • #6,756
  • Feb 25, 2010 7:33am Feb 25, 2010 7:33am
Quoting zoogle
You also get 99999 if there are no more news for the week.
I'm currently testing a 'fix' for the News Calendar issues that was documented here - http://www.forexfactory.com/showpost...postcount=6747

I'll let you know how I go on.

Cheers,
Nick.
  • #6,757
  • Feb 25, 2010 7:34am Feb 25, 2010 7:34am
Try this......

//+------------------------------------------------------------------+
//| expert News function |
//+------------------------------------------------------------------+

bool NewsExist()
{

string xmlFileName = "http://cloud.forexfactory.com/ffcal_week_this.xml";
bool News;

if(FileOpen(xmlFileName,FILE_BIN|FILE_READ|FILE_WRITE) < 0)
{
News = true;
}
else
{
for(int d = 0; d < ArraySize(Pairs); d++)
{
int MinutesSincePrevEvent =
iCustom(NULL,0,"Economic News",Pairs[d],true,false,false,true,true,1,0);

int MinutesUntilNextEvent =
iCustom(NULL,0,"Economic News",Pairs[d],true,false,false,true,true,1,1);

if(MinutesUntilNextEvent < 60 ||
MinutesSincePrevEvent < 15)
{
News = true;
}
}
}

return(News);

}
  • #6,758
  • Feb 25, 2010 7:59am Feb 25, 2010 7:59am
apparently, the function MathMax is able to find the maximum value between 2 values only
In order to find the maximum value inside 8 values, I have write like this :

intermedaire1 = MathMax(JpyKgRS,UsdKgRS)
intermedaire2 = MathMax(GbpKgRS,EurKgRS)
intermedaire3 = MathMax(ChfKgRS,AudKgRS)
intermedaire4 = MathMax(NzdKgRS,CadKgRS)
intermedaire5 = MathMax(intermedaire1,intermedaire2)
intermedaire6 = MathMax(intermedaire3,intermedaire4)
MaxKG = MathMax(intermedaire5,intermedaire6)

Is there a less stupid way to write this ???
  • #6,759
  • Feb 25, 2010 8:05am Feb 25, 2010 8:05am
Quoting maxou888
apparently, the function MathMax is able to find the maximum value between 2 values only
In order to find the maximum value inside 8 values, I have write like this :

intermedaire1 = MathMax(JpyKgRS,UsdKgRS)
intermedaire2 = MathMax(GbpKgRS,EurKgRS)
intermedaire3 = MathMax(ChfKgRS,AudKgRS)
intermedaire4 = MathMax(NzdKgRS,CadKgRS)
intermedaire5 = MathMax(intermedaire1,intermedaire2)
intermedaire6 = MathMax(intermedaire3,intermedaire4)
MaxKG = MathMax(intermedaire5,intermedaire6)

Is there a less stupid way to write this ???
Use arrays
  • #6,760
  • Feb 25, 2010 8:10am Feb 25, 2010 8:10am
Quoting maxou888
apparently, the function MathMax is able to find the maximum value between 2 values only
In order to find the maximum value inside 8 values, I have write like this :

intermedaire1 = MathMax(JpyKgRS,UsdKgRS)
intermedaire2 = MathMax(GbpKgRS,EurKgRS)
intermedaire3 = MathMax(ChfKgRS,AudKgRS)
intermedaire4 = MathMax(NzdKgRS,CadKgRS)
intermedaire5 = MathMax(intermedaire1,intermedaire2)
intermedaire6 = MathMax(intermedaire3,intermedaire4)
MaxKG = MathMax(intermedaire5,intermedaire6)

Is there a less stupid way to write this ???
i suppose you have declared an array to store all these "mathmax" values
2 traders viewing now