Old Dog with New Tricks | Page 13 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Old Dog's Taming of the Beast? 34 replies

Old Dog Fib Trading - EA Development 18 replies

Old Dog Fib Trading - EURUSD 399 replies

5 Tricks of Trend Trading, by Rob Booker 25 replies

How To Work Around A Market Maker's Tricks 0 replies

Attachments: Old Dog with New Tricks
Exit Attachments

Old Dog with New Tricks

  • #241
  • Jan 27, 2010 1:42am Jan 27, 2010 1:42am
Hi RR,
I'm testing your new EA (HiLo Trader 2 [TD].mq4). However, when I run it, it does not show any status at the left corner of chart.
I have the default settings except LicenseCheck1, 2, 3, 4 and LicenseCheck5 changed to "I agree".

Please help. Thanks.
Attached Image
  • #242
  • Jan 27, 2010 4:23am Jan 27, 2010 4:23am
Quoting alankw88
Hi RR,
I'm testing your new EA (HiLo Trader 2 [TD].mq4). However, when I run it, it does not show any status at the left corner of chart.
I have the default settings except LicenseCheck1, 2, 3, 4 and LicenseCheck5 changed to "I agree".

Please help. Thanks.
You didnīt read the license (in the source code)
  • #243
  • Edited 8:42am Jan 27, 2010 7:52am | Edited 8:42am
Quoting alankw88
Hi RR,
I'm testing your new EA (HiLo Trader 2 [TD].mq4). However, when I run it, it does not show any status at the left corner of chart.
I have the default settings except LicenseCheck1, 2, 3, 4 and LicenseCheck5 changed to "I agree".

Please help. Thanks.
You'll figure it out when you read the license carefully.
  • #244
  • Edited 3:57pm Jan 27, 2010 3:43pm | Edited 3:57pm
hi Ron,

what the hard way have you chosen ?
But thanks for all you are doing and giving part of your knowledge.

Some points are not clear to me:
- bot make stats based on daily data: when day is starting : gmt, server time , else ? What about 24 or 48 or 72 h slide ?
- is that bot staying away from dead hours in small timeframes ?
- is that bot able to use multi core cpu or gpu with opencl or else ?

On my own, i try to trade and think that way: the simpler the better.
So, the title above:
- the biggest time frames include the smallest, right ? So, it's easy to find trends, ranges and decide if price will continue or bounce. That is the big picture (see gif below).
- then in the smaller timeframe, follow the big picture.

you can see on gif:
- the 2 yellow lines are 38 & 50 fibo-fan
- the purple line is 200 ema
- the others are slope & trendlines
- blue & orange horizontal lines are months sup/res
- range is built with i-regr

Of course, the weekly below can be split in daily then hourly and smaller if you want, but the big one is your guide anyway.
Attached Image
  • #245
  • Jan 27, 2010 4:35pm Jan 27, 2010 4:35pm
here fibo- fan confirm that trend change on that small timeframe
Attached Image (click to enlarge)
Click to Enlarge Name: gu-m5.gif Size: 25 KB
  • #246
  • Jan 27, 2010 7:15pm Jan 27, 2010 7:15pm
Quoting dinofx
hi Ron,

what the hard way have you chosen ?...

The underlying system is hardly complicated at all. The complicated part is teaching the EA when to trade with the underlying system. The damn thing won't listen to my lessons, so I have to write its brain myself.

For further clarification

 

  1. The bot generates stats based on bar history and the number of bars that it is told to check. The EA scrolls back and scans through the bars on your chart. The only timeframe that is relevant is the timeframe that the EA is attached to.
  2. No
  3. No

Pardon my ignorance, but how is mentioning your system relevant to what I'm trying to do in this thread?

This thread--in case you haven't read it all the way through--is specifically geared toward designing an EA which can take an event, and decide on its own what its reaction should be during that event. I'm not looking to add indicators or filters. I am only interested in refining the method.

  • #247
  • Jan 27, 2010 9:42pm Jan 27, 2010 9:42pm
Quoting Ronald Raygun


extern double MinimumWinDifference = 0; The minimum difference between breakout win and counter trend win.

extern double MinimumSSDifference = 0; Minimum difference between breakout SS and counter trend SS.
Hi RR,
Could you please explain more what is MinimumWinDifference and MinimumSSDifference? How to define what is the suitable value for both? Any examples? Thanks so much.
  • #248
  • Edited 2:34am Jan 28, 2010 2:15am | Edited 2:34am
[quote=Ronald Raygun;3408002]Pardon my ignorance, but how is mentioning your system relevant to what I'm trying to do in this thread?

In any case i push "my system" against yours, it's not a "system". The general idea behind my comment here is that high time frames teach you where the prices are going, following a dynamic trend channel. Don't mind about what time frame is used as every smaller one is included into a bigger one, only spirit is important. That is where it's "relevant" when you teach the brain's bot. Nobody can see the landscape looking at his feet.

 Just catch the spirit here, the gifs above only illustrate what i talk about: where is the main trend, no need multiple filters or indicators, it's only eyes catching. Hope you understand my bad English and i always read "all the way through" when you post something i can see, don't want to be rude anyway, just bring an other sight and contribute with my experience.

Regards
  • #249
  • Jan 28, 2010 6:22am Jan 28, 2010 6:22am
RR

The ea in post 1 is it the same logic as the last one you posted
or do you recommend the latest posted
  • #250
  • Jan 28, 2010 7:41am Jan 28, 2010 7:41am
Quoting alankw88
Hi RR,
Could you please explain more what is MinimumWinDifference and MinimumSSDifference? How to define what is the suitable value for both? Any examples? Thanks so much.
Both the Breakout and CounterTrend Trading Styles have a WinPercent and a SuccessScore.

The purpose behind those two parameters is to tell the EA to trade when one trading style's win percent or success score is greater than another style's win percent or success score by x points.

For example, let's say that I set MinimumWinDifference to 20. That means that in order for the EA to trade either breakout or counter trend, one must have a winpercent that is 20% higher than the other. This I think will be more useful for people whose chart data isn't as complete as the rest of us, and which results in a low confidence level.

The same applies to the SuccessScore.

Quoting dryclean
RR

The ea in post 1 is it the same logic as the last one you posted
or do you recommend the latest posted
I'd suggest HiLo trader2. The main advantage over Hilo trader is that it can have multiple trades open--something a lot of people think is necessary.

[quote=dinofx;3408932]
Quoting Ronald Raygun
Pardon my ignorance, but how is mentioning your system relevant to what I'm trying to do in this thread?

[color=RoyalBlue]In any case i push "my system" against yours, it's not a "system". The general idea behind my comment here is that high time frames teach you where the prices are going, following a dynamic trend channel. Don't mind about what time frame is used as every smaller one is included into a bigger one, only spirit is important. That is where it's "relevant" when you teach the brain's bot. Nobody...
I think you're trying to tell me that it is better to go with the trend, no?

This EA was intended as a swing trading/scalping EA. The complete landscape does not apply as much as the pit you would fall into if you didn't look at your feet.

This EA is designed to look for an edge during a certain event--in this case a new local high, or a new local low. During either of those events, there are two possible solutions: Trade the event as a breakout (New High: Long/New Low: Short) or trade the event as a counter trade (New High: Short/New Low: Long). The EA inspects the past to try and extract an edge. Does one trading style succeed x% more than another trading style? If so, use the other trading style. Simple as that.

In the context of trying to trade with the trend, that is certainly necessary if your trade will last a long time, and by long time, I mean greater than a week.
  • #251
  • Jan 28, 2010 10:10am Jan 28, 2010 10:10am
thanks Ronald to take time to answer,

i'm sure you have understood what i meant, see any pair with W or Mn time frame like the weekly gif i've posted previouly. Month after month, price goes inside a nice channel. That don't mean i'm telling to trade with the trend on smaller time frame !!!

Just to say that when this bot analize the risk and choose one scenario, it have to know if price is near or not the main channel border and if this channel is rising or not. This setting have to be known by the bot to exclude high risk and maybe send a buystop or sellstop.
  • #252
  • Jan 30, 2010 7:40am Jan 30, 2010 7:40am
ok, lets get some good discussion kickstarted here....i havent tried this EA yet...i assume market needs to be open and so i will have to try it on monday...

i have been trading for the past 7 yrs....a year back, i met a friend from holland who was developing a similar ea, not for mt4 though...and i was very very complicated....the idea though was same....we joined together to make the ea better....the ea had 2400 parameters and we kept optimizing each parameter, backtested it using our own data and tried to find the best setting for each pair....

we found that for GU alone (which took all of 2 yrs with 40 systems running 24 hrs a day) but by then i was too exhausted and couldnt believe that trading could be so difficult and so complicated....and i stopped....

i can see that RR is also thinking on the same lines and its good to see....so how do we progress, is something that is still open to interesting discussion....
  • #253
  • Jan 30, 2010 9:20am Jan 30, 2010 9:20am
The first batch of ticks have been delivered.

www.RonaldRaygunForex.com/TickDB/Archive

Currently, it's unrestricted, but if demand becomes too high, I will either restrict the download speed, the number of simultaneous connections allowed, or both. As an option, will someone take the responsibility of uploading the files to rapidshare, megaupload, or similar websites? That way, we can at least host these files in a cost-effective way.
  • #254
  • Edited 11:22am Jan 30, 2010 10:37am | Edited 11:22am
Thanks Ronald for this database.

I'm downloading the files at the moment. When I finish I'll upload them to rapidshare.


EDIT:

Rapidshare link to the database:
http://rapidshare.com/files/34346943...2010.01.23.rar
  • #255
  • Feb 16, 2010 3:29am Feb 16, 2010 3:29am
Hi Ronald

Sorry if this is not the right place, I was looking at an results of an EA you wrote, opto 123, i need to ask a few questions on that, whats the best way to discuss that...thanks....sam
  • #256
  • Edited 6:48pm Feb 21, 2010 12:50pm | Edited 6:48pm
Ronald et. al.

I've just come across your work here at Forex Factory in the past two days. I've read several of your threads from start to finish and by no means have I read all of them yet.

I've got a few burning questions: 1. I was fascinated and thrilled to see that in your OZ work you used an SL of 40, but it wasn't just an SL, but also a close and re-entry to catch the trend for the ride. If you got stopped out 4 straight times, your loss 160 pips. The genius being in you didn't open the number of loss contracts plus 1, but only one contract then moving your SL to break even as soon as it was reached. Did you consider opening the number of lost contracts plus one, two or three times? What was your decision based on?

2. In decisions of when to enter positions, have you considered times to trade, both entry and perhaps time to close all trades, like market close on Friday afternoon? Reading your flip a coin entry strategy, suggests the trade management is as much if not more important than entry. Have you done any statistical analysis to determine if one directional trading, ie only long or only short, is more profitable than both long and short trading?

3. What are your thoughts on trade direction and identifying and catching the trend?

3. What is your daily or weekly or monthly pip target?

Silly Patch
In VA
  • #257
  • Jun 4, 2010 1:34am Jun 4, 2010 1:34am
Hi Ron
Your expert advisor makes a simulation trades using history data, yes? Is it possible to export this results to Exel or other programm and sort it by 'open hour'?

I will try to explain it on example:
EA opened 90 trades between 9.00 and 10.00 GMT (in simulation on history data) 60 closed by TP 30 by SL, but between 12.00 and 13.00 GMT 130 trades, 50 closed by TP, 80 by SL (it's only sample values).
If we have got a results from backtest, we will compare it on chart and we will know which hours are best for counter trading (or breakout).

Can I ask you for do it? I can't programming.
What do you think about this idea?
  • #258
  • Jun 4, 2010 5:57am Jun 4, 2010 5:57am
Sorting by hour is a possibility. I just haven't released the EA with this feature yet. Still working out the IP rights.
  • #259
  • Jun 5, 2010 3:15pm Jun 5, 2010 3:15pm
Could you add trading hours parameters to EA code?

Regards
  • #260
  • Jun 15, 2010 11:27am Jun 15, 2010 11:27am
Quoting Ronald Raygun
Quoting Nighthawk
I get a zero divide error upon loading the ea on eur/usd 4h chart.
What do the stats on the left say? Most likely, one of those values is 0

Hi Ron,

i get the same divide error issue...

it is due to the stats counters being zero because it would not find any winning and/or loosing trade during optimization.

This is also why we get high cpu load: the error pops all the time until we get those counters !=0, and that causes a looooooooot of "divide error" printouts in the journal (and it also creates huuuuuuuuge log files...)

To fix this, i changed the following code (2 places in the code):

Inserted Code
//Calculate Stats
double BOWinRate = 0;
double CTWinRate = 0;
double ErrorRate = 0;

if (BreakoutTotal!=0) BOWinRate = (BreakoutWin / BreakoutTotal) * 100;
if (CounterTotal!=0) CTWinRate = (CounterWin / CounterTotal) * 100;
if ((BreakoutTotal + CounterTotal) != 0) ErrorRate = ((BreakoutTie + CounterTie) / (BreakoutTotal + CounterTotal)) * 100;
now it runs silently...

Sq;
1 trader viewing now