My EA Gift to Traders | Page 32 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Screen shot indicator free gift 0 replies

DECEMA - Xmas gift! 6 replies

Vegas 1HR GIFT: Big opportunity coming! 5 replies

Attachments: My EA Gift to Traders
Exit Attachments

My EA Gift to Traders

  • #621
  • Feb 20, 2015 3:50pm Feb 20, 2015 3:50pm
Quoting Dave_C
{quote} I think we were both posting at the same time. If you look at Post #614 you can see that I have edited the text. The point is as you infer, tick data back testing is more accurate. It's just that optimization using tick data is almost impossible except over really short periods. With that in mind, I wrote the EA to use the "Open price" model. The post 2 above is an illustration of a confirmation of the strategy using tick data.
I think it still needs an edit for clarity. If you would like to take a look, I can remove this post later to avoid cluttering the thread.
  • #622
  • Feb 20, 2015 3:59pm Feb 20, 2015 3:59pm
New update at Post #1.

Fix of error preventing combination of strategies and filters.
  • #623
  • Feb 20, 2015 4:03pm Feb 20, 2015 4:03pm
Quoting drofwarc
{quote} I think it still needs an edit for clarity. If you would like to take a look, I can remove this post later to avoid cluttering the thread.
If back testing produces favourable results using the "open price model", you should endeavour to confirm the results using the most accurate data available -
"tick data" model.
  • #624
  • Feb 20, 2015 4:10pm Feb 20, 2015 4:10pm
Quoting Dave_C
{quote} If back testing produces favourable results using the "open price model", you should endeavour to confirm the results using the most accurate data available - "tick data" model.
Hey Dave, that statement is very clear, but this one is still murky:

"I suspect the reason that your results are completely different is because you are using the very tick mode. The EA is explicitly designed to use the model -
"Open prices only (fastest method to analyse the bar just completed, use only for EA's that explicitly control bar opening)" Whilst you are getting more accurate results, the EA does not look at any activities that occur within the current bar affecting SL calculations which every tick back testing does in fact do. However to do comprehensive optimization on tick data would take weeks, hence the design of the EA."

I really don't mean to bust your chops. I just figure you would want to convey your meaning on this important point in a crystal clear way for your readers.
  • #625
  • Feb 20, 2015 4:33pm Feb 20, 2015 4:33pm
Quoting drofwarc
{quote} Hey Dave, that statement is very clear, but this one is still murky: "I suspect the reason that your results are completely different is because you are using the very tick mode. The EA is explicitly designed to use the model - "Open prices only (fastest method to analyse the bar just completed, use only for EA's that explicitly control bar opening)" Whilst you are getting more accurate results, the EA does not look at any activities that occur within the current bar affecting SL calculations which every tick back testing does in fact do....
Thank you for your comment!
  • #626
  • Feb 20, 2015 5:19pm Feb 20, 2015 5:19pm
Quoting Dave_C
{quote} I suspect the reason that your results are completely different is because you are using the very tick mode. The EA is explicitly designed to use the model - "Open prices only (fastest method to analyse the bar just completed, use only for EA's that explicitly control bar opening)" Whilst you are getting more accurate results, the EA does not look at any activities that occur within the current bar affecting SL calculations which every tick back testing does in fact do. However to do comprehensive optimization on tick data would take weeks,...
Thanks a lot, i just need a simple strategy for a school work with small profit, by optimizing the EMA cross i got 2,43 profit factor, but wont work in the long time anyway, i will go on working on in for a bit, still trying to understand fully your EA

btw, if you need some more computing power to help optmize some inputs, let me now
  • #627
  • Feb 20, 2015 11:21pm Feb 20, 2015 11:21pm
How many times (in average) did you get consecutive losses while using the StopLossMultiplier?
I know that martingale is dangerous, but it worth the try, because there's nothing impossible in FX world, we can use any possible way to get some pips
OrderSend(Symbol(),OP_BUY,10,Ask,30,SL,TP,"",0,0,clrGreen);
  • #628
  • Feb 21, 2015 4:38am Feb 21, 2015 4:38am
Quoting mangblekok
How many times (in average) did you get consecutive losses while using the StopLossMultiplier? I know that martingale is dangerous, but it worth the try, because there's nothing impossible in FX world, we can use any possible way to get some pips
That is a very significant question and exposes the whole Martingdale system. Alex was the only person to take this head on and I emplore anybody to refute what he is saying. I haven't always agreed with him relating with his idealogy but he has added a new dimension to the Martingdale discussion.

My experience is 16 losses is enough!

We are no longer bound to the traditional system. Expand your minds! Post #527 refers.
  • #629
  • Feb 21, 2015 7:03am Feb 21, 2015 7:03am
Quoting Dave_C
{quote} That is exactly the .set which demonstrates what we have all been looking for. I ran this .set through actual tick data as you can see with the 99.9% accuracy in the report. I have enclosed Alex's set file here. Rename to .set. I'm going to start a Trade Explorer next week with this strategy and post on this thread because it looks very impressive. The EA is now at Post #1 {image} {image} {file}
Glad to hear that i can contribute to your thread positively (i hope i understood correctly your post , because English is not my native language and not sure if i get the correct meaning. ) .

What is considered to scaling it was only because of your suggestion , i really never have traded with scaling MM and not my favorite. I prefer to do my analysis and enter a trade with SL and TP and wait for the outcome. I hate managing trades afterwards because the only thing you get is MESS nothing more and i am absolute on this statement.

I wish i could pursue you to listen to me and try to adopt range or renko bars and see then the results. Because you can see that what i was explaining so far and the whole concept with the progression this set proves my words. I am not that person who suggests something that don't knows the outcome of that . And i know the outcome because i am doing this already. Think of it such as a movie that you already have seen and you know exactly what is gonna happen next. I am not saying that i am a GURU but i believe that i have enough experience to stand for my every saying.

Wish all the goods to you my friend because thanks to you i have reduced some time in front off the monitor what was and the original goal for me. So if we stay here i am already good i am not greed person , if we can move further that is better. my advice work in fine tuning the entries that is what this EA still needs , because there is more than enough Money management settings. Because i really believe that there is not always a way to get to a destination my suggestion to this you already know but maybe you will come up with something better , or someone else give some hint on this topic.

Alex
  • #630
  • Feb 21, 2015 2:56pm Feb 21, 2015 2:56pm
Quoting alex8585
{quote} Glad to hear that i can contribute to your thread positively (i hope i understood correctly your post , because English is not my native language and not sure if i get the correct meaning. ) . What is considered to scaling it was only because of your suggestion , i really never have traded with scaling MM and not my favorite. I prefer to do my analysis and enter a trade with SL and TP and wait for the outcome. I hate managing trades afterwards because the only thing you get is MESS nothing more and i am absolute on this statement. I wish...
Alex I realise you have your own particular trading ideas but this thread is not about any particular trading system. It is a set of tools to use with your trading to refine your own ideas and will never be specific. Your set file is an example of what can be done. The issue of Renko and HA candles has been addressed and solutions provided.

At this time, I will deal with all issues relating to the EA and given a high degree of stability will only then look at further ideas. Some ideas are relatively easy to implement such as choice of trading days and can be addressed.
  • #631
  • Feb 21, 2015 3:30pm Feb 21, 2015 3:30pm
Hi Dave,

I'd like to suggest another feature for you to consider adding as an option.

I believe one of the reasons backtests can be inaccurate is if the EA is set to ignore new trades when a trade is currently open. It seems to me that the best way to determine the robustness of a strategy is to take every trade and see how it performs. I understand that if you take a large enough number trades when ignoring new trades, the results should be close to the results when taking every trade. However, that's not always possible. For example, I just finished testing a strategy over 5 years of tick data and the EA opened only about 300 trades. In that period, there were likely closer to 1,000 or 1,500 trade opportunities, which would surely make for more reliable backtesting.

I'd also like to test shorter timeframes for walk-forward analysis and to optimize parameters for recent conditions. If I were to test this particular strategy for only the last 12 months, I'd have only about 60 trades to work with. Again, the sample would be much larger if the EA were taking every trade.

So I would certainly appreciate having this option, and I'm guessing others would benefit from it, too.

I have no idea what would be involved in implementing this feature. Over to you to consider if it's within the scope of the EA and doesn't involve too much additional work.

Kind regards,
drofwarc
  • #632
  • Feb 21, 2015 4:45pm Feb 21, 2015 4:45pm
Quoting drofwarc
Hi Dave, I'd like to suggest another feature for you to consider adding as an option. I believe one of the reasons backtests can be inaccurate is if the EA is set to ignore new trades when a trade is currently open. It seems to me that the best way to determine the robustness of a strategy is to take every trade and see how it performs. I understand that if you take a large enough number trades when ignoring new trades, the results should be close to the results when taking every trade. However, that's not always possible. For example, I just finished...
When you have a situation of an EA opening a trade every time an indicator gives such an indication, from an automation point of view each trade has to be managed just like yourself opening up simultaneous trades. At what point in your scenario, do you end each trade? Do you continue to open up new trades until all of your available margin is used? When you have used your available margin you then have to start ignoring trades anyway. Do you then start selling so you can open up extra trades. What happens when a major news event turns against you? There are no simple answers. In the scenario you describe there needs to be more definition as to how you are going to manage this free flowing trade scenario.

Whilst I understand your logic, I don't agree with the idealogy. The idea of a back test optimization is to consider all of the variables and to come up with the best solution using the variables available.

A simple example -
Is it better to have a 5/20 EMA cross with a small SL-10 and TP-10 resulting in say 300 trades
or have at 30/150 EMA cross with a SL-150 TP-150 resulting in 20 trades

The optimization considers all scenarios with the variables available and comes up with the most cost effective solution.

Regarding back testing in general -
I personally, in my own trading, do not back test beyond 3 months ago and always redo the back test on a monthly basis together with a review of the previous months performance against the previous 3 months back test period. In summary - a new 3 monthly back test at the end of every month.
  • #633
  • Feb 21, 2015 5:00pm Feb 21, 2015 5:00pm
Quoting Dave_C
{quote} When you have a situation of an EA opening a trade every time an indicator gives such an indication, from an automation point of view each trade has to be managed just like yourself opening up simultaneous trades. At what point in your scenario, do you end each trade? Do you continue to open up new trades until all of your available margin is used? When you have used your available margin you then have to start ignoring trades anyway. Do you then start selling so you can open up extra trades. What happens when a major news event turns against...
You've raised a number of issues I hadn't considered, so I'll have to think on this before responding fully. For now, I'll add that, for this particular strategy, I would open a single trade per trade trigger, similar to UsePriceEMAEvent. So that helps limit the number of open trades.

Have to run now and won't be able to respond again until tomorrow.
  • #634
  • Edited 9:33am Feb 22, 2015 9:20am | Edited 9:33am
Some more interesting results for long term trading. Backtest period from 2010 - 2015 H4 timeframe.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: Screenshot1.png Size: 143 KB
Click to Enlarge Name: Screenshot2.png Size: 148 KB
Click to Enlarge Name: Screenshot3.png Size: 143 KB
Click to Enlarge Name: Screenshot4.png Size: 144 KB
Click to Enlarge Name: Screenshot5.png Size: 143 KB
Click to Enlarge Name: Screenshot6.png Size: 147 KB
  • #635
  • Feb 22, 2015 10:13am Feb 22, 2015 10:13am
Quoting alex8585
Some more interesting results for long term trading. Backtest period from 2010 - 2015 H4 timeframe. {image} {image} {image} {image} {image} {image}
Hi Alex,
Some wonderful results there. Could you please post the settings that gave these results? I have spent enough time on this promising EA but it appears my settings are not very good to get better trades.

Please help.

Thank you all for the wonderful work you are doing.
  • #636
  • Edited 10:39am Feb 22, 2015 10:14am | Edited 10:39am
Quoting alex8585
Some more interesting results for long term trading. Backtest period from 2010 - 2015 H4 timeframe. {image} {image} {image} {image} {image} {image}
Thanks for the input Alex. It would be great if you could attach your .set files. If you drag the blue title bar of the Tester, you can make that window float to give greater definition. (even onto another monitor). The final image is just a capture of any area of the screen using just the printscreen key with the free program Greenshot. I find it much more versatile than the Windows Snipping Tool
Attached Image(s) (click to enlarge)
Click to Enlarge Name: 2015-02-23 02_24_22-914382_ GlobalPrime-Live.jpg Size: 432 KB
Click to Enlarge Name: 2015-02-23 02_29_55-914382_ GlobalPrime-Live.jpg Size: 324 KB
Click to Enlarge Name: 2015-02-23 02_32_50-914382_ GlobalPrime-Live.jpg Size: 174 KB
  • #637
  • Feb 22, 2015 7:47pm Feb 22, 2015 7:47pm
Quoting Dave_C
If you have been frustrated by the inability to get successful back tests by combining strategies and filters, I have just discovered a major bug in the EA. Before I release it, I would appreciate a few testers to give me some feedback. As an example, enclosed are 2 set files and an updated EA which highlights the value of being able to correctly mix strategies and filters. With the EA as it stands at Post #1, no successful results would result in strategy set 2 back test with optimization. With the revised EA, strategy set 2 displays a completely...
Dear Dave_C,from what you have posted on#611 that the second image,It similar the profit is 2800%+,very different from your imaging,Ha.

But with my tick test ,same period,the profit is 250%+.What can cause that?It is puzzled.
  • #638
  • Edited 11:09pm Feb 22, 2015 9:47pm | Edited 11:09pm
Quoting ykspy
{quote} Dear Dave_C,from what you have posted on#611 that the second image,It similar the profit is 2800%+,very different from your imaging,Ha. But with my tick test ,same period,the profit is 250%+.What can cause that?It is puzzled.
That back test was done using the opening price model which provides for fast optimization. All of my optimizations are done using that model, As I have mentioned previously you confirm the results with tick data. In that case the curve has turned parabolic. Simulated tick data is not necessarily accurate.

I have just completed the back test using actual tick data and the graph and report are produced below. Make sure that you nominate the chart time frame at the 4th last entry on the inputs.
Attached Image(s) (click to enlarge)
Click to Enlarge Name: 2015-02-23 13_37_42-Tester.jpg Size: 122 KB
Click to Enlarge Name: 2015-02-23 13_38_03-914382_ GlobalPrime-Live.jpg Size: 89 KB
  • #639
  • Feb 23, 2015 1:16am Feb 23, 2015 1:16am
I have just added a Trade Explorer at the top of the page relating to alex8585's Post #613 which has now been optimized for the 7 majors.

The 7 set files as a zip are enclosed. The strategy uses candlestick recognition confirmed by an RSI cross within 5 bars ago and being on the correct side of the 5 EMA shift 2 with stop loss and risk reward multipliers and break even settings.
Attached File(s)
File Type: zip Alex8585.zip   14 KB | 270 downloads
  • #640
  • Edited 5:49am Feb 23, 2015 5:27am | Edited 5:49am
Hey, just wanted to say thanks Dave C for the awesome EA. Sorry i can't really contribute more as I'm still pretty much a newbie to all this. One thing i can contribute though is set files i test out. So for my first one i have a EUR/USD 60 min that put out some nice numbers for Dec 2014 to present. I try not to back test to far since the markets are always changing and 1 method is never going to give you good results 100% of the time. Currently messing around with USD/JPY now I'll be sure to post any good findings.

Edit: Just wanted to add that i Demo with $500-$1000 since that is a more realistic number for me.

Attached File(s)
File Type: txt EURUSD MACD 60min Test Dec2014-Feb2015.txt   13 KB | 214 downloads
1 trader viewing now