Old Dog with New Tricks | Page 12 | 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

  • #221
  • Jan 19, 2010 8:54am Jan 19, 2010 8:54am
Hm. Well I'm going to try again on a third PC. Could you perhaps post a set file that does not contain your actual trading specs (unless you feel like it ), but one that you believe works fine? I wonder if one of the settings is playing up.
  • #222
  • Jan 19, 2010 8:57am Jan 19, 2010 8:57am
Quoting Mindsphere
Hm. Well I'm going to try again on a third PC. Could you perhaps post a set file that does not contain your actual trading specs (unless you feel like it ), but one that you believe works fine? I wonder if one of the settings is playing up.
Default except:

Lots: Non-zero
Win Percent: 50%
TP: 200 (5-digit)
SL: 200 (5-digit)

That should be the bare minimum.

The key is to make sure that one trading style is greater than 50%, and the other trading style is less than 50%
  • #223
  • Jan 19, 2010 9:09am Jan 19, 2010 9:09am
Quoting Ronald Raygun
Default except:

Lots: Non-zero
Win Percent: 50%
TP: 200 (5-digit)
SL: 200 (5-digit)

That should be the bare minimum.

The key is to make sure that one trading style is greater than 50%, and the other trading style is less than 50%
Just tried that, as well as a number of other permutations. Basically as soon as the first tick comes through I can no longer modify the EA settings. It's as if the EA has crashed. Testing on FxPro's demo platform. Will try on a couple of others now. This is the 3rd machine I'm trying it on, and the second internet connection.

One thought - do you perhaps have it set to access a domain that is not resolving, but you have manually set in your hosts file? Or perhaps you're running your own DNS server and thus there's a lookup that works for you but not for the rest of us. Or could there be some sort of DLL call that's failing as we all don't have the file?

Edit: Same results on other platforms. Tick database connectivity perhaps causing the issue?
  • #224
  • Jan 20, 2010 3:08pm Jan 20, 2010 3:08pm
I hope we can resolve this I thought it wasn't allowing me to change the variables because it was crashing on my slow comp. but with so many others having the same problem it seems unlikely now.
The greatest force on earth is compound interest.
  • #225
  • Jan 24, 2010 12:53pm Jan 24, 2010 12:53pm
The reason behind the bulk of your problems was a connection to my webserver www.RonaldRaygunForex.com

The EA has a time module which syncs off of GMT. The page www.RonaldRaygunForex.com/TickDB/Time.php didn't exist, and wasn't returning a 404 error, which in turn caused an infinite loop and furthermore resulted in the freezing of the EA.

I did some testing Friday afternoon before the market closed, and it looks like everything should begin working properly come the market open in a few hours.

For those of you anticipating the market open with this EA, here are some settings to try out:

Lots: Not 0
TP: 200 (5-digit)
SL: 200 (5-digit)
WinPercent: 50

Leave the rest as default.

The EA's trading style needs to read either "Breakout" or "Counter Trade" If it says "stand aside" change your win percentage so that one trading style is above it, and one trading style is below.
  • #226
  • Jan 24, 2010 4:04pm Jan 24, 2010 4:04pm
I am on board here now, RR, and looking forward to the ride.

Your ideas are breathtakingly imaginative and have got my creative juices sloshing around. You are an extraordinary young man.

Cheers

  • #227
  • Jan 24, 2010 4:15pm Jan 24, 2010 4:15pm
Glad to have you around Steve!

If there's anything I can do to help you with development of this concept, let me know.
  • #228
  • Jan 24, 2010 4:21pm Jan 24, 2010 4:21pm
Hey Ronald.

I am glad that you are trying to find an algorithm for an EA to adjust itself. I am trying to optimize EA's as well but I use different methods (some arguably more effective), since I have some background in artificial intelligence.

I see that your bots are automatically collecting data. Does it work good? Any chance of getting these (of course history data helps speed up the backtesting and optimizing).

Anyway, hopefully I will post my results soon with data I obtained from forexite.com
  • #229
  • Jan 24, 2010 4:44pm Jan 24, 2010 4:44pm
Collected data will be posted on RonaldRaygunForex.com eventually. I'm trying to work out bandwidth limits with my host.

If someone is willing to host these files, or if anyone knows a way to upload these files via script to a website (rapidshare?), I'm all ears. I don't want this account to be suspended like with my previous two hosts.
  • #230
  • Jan 24, 2010 6:24pm Jan 24, 2010 6:24pm
Hey RR and Steve! Just have to say that I have been following both of your work for several months now and am awed by your abilities and desire to help others. Man it's great to see the two of you working together! I will be glued to both of your threads. Thank you for all that you have done and are still doing. Good luck!
  • #231
  • Jan 25, 2010 8:53am Jan 25, 2010 8:53am
Left the computer on overnight with hi low trader running on AUDJPY and USDJPY, I can see that price passed below the red line on USDJPY its set to breakout (58.2%) and didn't place a trade.
The greatest force on earth is compound interest.
  • #232
  • Edited 7:52pm Jan 25, 2010 8:58am | Edited 7:52pm
I guess I should've expected this sooner.

Last night I detected a decompiled version of this EA hitting my server. To prevent commercialization, I am releasing the source code in its entirety under the FreeBSD license.

The license is included in the source code, and you are required to read through it if you want to enable the EA. It's a shame that I have to do this.


License:

Copyright (c) 2010, Ronald Raygun
All rights reserved.

Redistribution and use in source and binary forms, with or without
modification, are required provided that the following conditions are met:

* Redistributions of source code must retain the above copyright
notice, this list of conditions and the following disclaimer.
* Redistributions in binary form must reproduce the above copyright
notice, this list of conditions and the following disclaimer in
the documentation and/or other materials provided with the distribution.
* Neither the name of Ronald Raygun nor the names of any contributors
may be used to endorse or promote products derived from this software
without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
POSSIBILITY OF SUCH DAMAGE.

For those wishing to purchase a commercial license for this code, please contact
user Ronald Raygun on www.ForexFactory.com.

New settings:
extern int MaxSimultaneousTrades = 0;
extern string Remark3 = "== HiLo Settings ==";
extern int BarStart = 0; The number of bars before the current bar to start the backtest from. Set to 0 to use every bar.
extern int InitialRange = 10; The number of bars after the day's start to determine the initial hilo range.
extern double WinPercent = 0;The minimum win percentage required to trade. One trading style must be above this number, and the other trading style must be below.
extern double MinimumWinDifference = 0; The minimum difference between breakout win and counter trend win.
extern double MinimumSuccessScore = 0; Minimum success score required to trade. One trading style must be above this number, and the other trading style must be below.
extern double MinimumSSDifference = 0; Minimum difference between breakout SS and counter trend SS.

Note: If more than one of these parameters is enabled, the trade conditions needs to match all of the parameters set, otherwise, there will be no trade.

extern int MinimumConfidence = 90; The EA's backtester is sometimes unsure whether or not a trade ended with a win or a loss. This event adds a trade to the 'Unsure" column. If the percentage of unsure trades is greater than 100 - MinimumConfidence, there is no trade.
Attached File(s)
File Type: mq4 HiLo Trader 2 [TD].mq4   39 KB | 790 downloads
  • #233
  • Jan 25, 2010 6:48pm Jan 25, 2010 6:48pm
Hi RR,

thanks for this gem.
Two questions / comments:

1. You said your EA does not have to be backtestable with the Strategy Tester because it has its own backtesting engine built in.
I agree, but there is one nice feature missing in your backtesting engine (if I did not overlook something) , which could be compensated by the Strategy tester:
A list of walk-forward trades and the resulting equity curve for an simulated run over let´s say the last two years.

Right now your EA tells us the success rate for a given past period and trades according to it (or stands aside).

If it would be possible to use it with the Strategy Tester we would have a nice auto-walk-forward tester, so we wouldnt have to wait for 2 years to see if it is profitable in the long run with some certainty.

If we combine this with the auto-optimization idea of Mindsphere, this would be the first MT4 analogon I know (and have been missing for some time) for these very nice tools for EasyLanguage (TradeStation)

-> http://www.thegrailsystem.com/
-> http://www.profsoftware.com/bt/


2.
Lets say we have 5 breakout bars in a row.
I mean that these 5 bars come in a row and make all new highs.
The Profit Target for all of these bars is successfully reached twenty bars later.
In your stats this would count as 5 successful breakouts.
However in reality we would have entered only one position (at least normally).
Wouldn´t it be better to count only 1 successful breakout instead of 5 ?
  • #234
  • Jan 25, 2010 7:27pm Jan 25, 2010 7:27pm
1) It is impossible for the EA to backtest using the strategy tester because the EA is using the chart history. The backtester only loads the data it needs. So the most recent strategy tester trades would be useful and accurate.

What do you mean by a 'walk-forward'? A time-stepped backtest? Could you elaborate?

That is potential future development of the EA. Officially, there is nothing in MQL which allows calls to the strategy tester, so everything would have to be written from scratch. The backtesting code in this version of the EA took 6 months to develop a relatively light and efficient prototype, and another 1.5 years of forward testing to get the damn thing working as flawlessly as possible. I suspect continued development will lead to another few years until a perfected self-optimizing, self backtesting code matures. Hopefully, the addition of more programmers working on this concept will cover the 2 years in man-hours sooner.

2) I know, a lot of people have been complaining about this, there is a new addition in this current version of the EA which allows you to define how many simultaneous trades are allowed.

As for the given example, general rule of thumb, that sort of pattern is incredibly rare with 1:1 SL:TP ratio. If you increase the ratio to something like 10:1 SL:TP ratio, then yes I suppose that scenario is possible, and so is the converse... you could come across a long string of losing trades.
  • #235
  • Edited Jan 26, 2010 12:05am Jan 25, 2010 11:51pm | Edited Jan 26, 2010 12:05am
Its placing trades for me now, got it running on AUDJPY and the small red bars are on the chart. but the small red bars aren't on my USDJPY chart even though its running on the chart the same as for AUDJPY. Perhaps it is only allowing AUDJPY to trade because its more volatile. (Ive been using the penultimate version)
The greatest force on earth is compound interest.
  • #236
  • Jan 26, 2010 6:30am Jan 26, 2010 6:30am
been running this bot for a week or so and today it took it's first trade on E/U a sell.Lets see what it does
  • #237
  • Jan 26, 2010 6:46am Jan 26, 2010 6:46am
Quoting Ronald Raygun
1) It is impossible for the EA to backtest using the strategy tester because the EA is using the chart history. The backtester only loads the data it needs. So the most recent strategy tester trades would be useful and accurate.
I am not sure I understood you correctly. Why should it not be possible to use history data in the strategy tester ?

Quoting Ronald Raygun
What do you mean by a 'walk-forward'? A time-stepped backtest? Could you elaborate?
With "walk forward" I mean to simulate the placement of trades (or staying aside) according to the trading style calculated by your EA for every single bar. I am aware that a simulated year could take ages to be processed as the statistics have to be recalculated for every single bar (that could be tweaked by adding only the most recent bars to the statistic and removing the oldest ones).

Probably I am missing something but I cant see what.
If you dont mind I could try to integrate this in your EA.

Quoting Ronald Raygun
2) I know, a lot of people have been complaining about this, there is a new addition in this current version of the EA which allows you to define how many simultaneous trades are allowed.

As for the given example, general rule of thumb, that sort of pattern is incredibly rare with 1:1 SL:TP ratio. If you increase the ratio to something like 10:1 SL:TP ratio, then yes I suppose that scenario is possible, and so is the converse... you could come across a long string of losing trades.
In my observations this happens quite often, even with a 1:1 ratio.
I have seen the "MaxSimultaneousTrades" parameter but it seems only to prevent opening more than one position, however it does not affect the statistics calculation.
If, in my 5 bars example, a single successful CT would follow, this would be a 5:1 win for the breakout statistic even if it should be 1:1.
  • #238
  • Jan 26, 2010 7:49am Jan 26, 2010 7:49am
Quoting pippiee
I am not sure I understood you correctly. Why should it not be possible to use history data in the strategy tester ?
The EA checks bars on a chart, not true history data. At the start of a backtest, there are only a few bars, toward the end of a backtest, there are more bars.

Quoting pippiee
With "walk forward" I mean to simulate the placement of trades (or staying aside) according to the trading style calculated by your EA for every single bar. I am aware that a simulated year could take ages to be processed as the statistics have to be recalculated for every single bar (that could be tweaked by adding only the most recent bars to the statistic and removing the oldest ones).


Probably I am missing something but I cant see what.
If you dont mind I could try to integrate this in your EA.
The EA recalculates once daily. It's not the most accurate walk-forward, but the stats are being updated.


Quoting pippiee
In my observations this happens quite often, even with a 1:1 ratio.
I have seen the "MaxSimultaneousTrades" parameter but it seems only to prevent opening more than one position, however it does not affect the statistics calculation.
If, in my 5 bars example, a single successful CT would follow, this would be a 5:1 win for the breakout statistic even if it should be 1:1.
Check around line 520 in the code, there should be something that says "IsTrade = False;" that disables the "one trade at a time rule"
  • #239
  • Jan 26, 2010 12:21pm Jan 26, 2010 12:21pm
Yesterday I enabled this EA on EU using M15 charts and today it had 3 profitable trades.

Thanks for the EA RR. I have set up additional charts to do some different tests.

Carol
  • #240
  • Jan 26, 2010 9:51pm Jan 26, 2010 9:51pm
Thank you RR for the mq4 file.
4 traders viewing now