Dreamliners’ MPAS EAs, Indicators, etc | Page 5 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Dreamliners’ MPAS (Mechanical Price Action Strategy) System 486 replies

MT4 doesn't save trendlines, indicators, etc 1 reply

Quantum leaps in tweaking MT4 (indicators, EAs, etc.) 0 replies

Merc's special Fibonacci trading,in combo w/QQE Gann etc.etc. 7 replies

indicators, etc... 12 replies

Attachments: Dreamliners’ MPAS EAs, Indicators, etc
Exit Attachments

Dreamliners’ MPAS EAs, Indicators, etc

  • #81
  • Jan 1, 2011 6:49am Jan 1, 2011 6:49am
Quoting netFX
with your solution back-testing on off-line chart is working.
I have noticed one thing though which is the modeling quality is like ~20%. I don't know with this modeling quality how reliable back-testing results would be.
I think the quality is around 20% because you are using Metaquote's data...

In order to have a better quality you need to use real tick data...for example those provided by Dukascopy, in any case you will need Bid data and Ask data to take in account the spread, and in any case the spread will never be the same you have with your broker...so the backtest will never be realistic enough...
Market is like a puzzle, YOU have to fit the pieces!!
  • #82
  • Jan 1, 2011 7:48am Jan 1, 2011 7:48am
Quoting Chris_B
I think the quality is around 20% because you are using Metaquote's data...

In order to have a better quality you need to use real tick data...for example those provided by Dukascopy, in any case you will need Bid data and Ask data to take in account the spread, and in any case the spread will never be the same you have with your broker...so the backtest will never be realistic enough...
Although during back-testing real spread is not taken into account but this system has SL-10 PIPs and TP-20 PIPs and therefore the win/loss % would be same in demo/live account as back-testing. I am not interested in profit pips but win/loss % at this time and back-testing would reflect the correct win/lost %.
AshN
  • #83
  • Jan 1, 2011 7:57am Jan 1, 2011 7:57am
Quoting smjones
Hi Chris, I modified your EA, so I could run it on a separate chart other than range bars. Time frame does not matter I am running it on 5 minute....
Here are the 2 indicators. "offline ohlc global" goes on the Range Bar chart and "DL Lines and RangeBars" goes on any other chart chosen as long as it is the same symbol.

PS, I am not recommending any type of trading and am not responsible in any way for how these will be used or any information disclosed. My advise is my opinion only.[/quote]
does not match up see the charts
Attached Image (click to enlarge)
Click to Enlarge Name: 9.gif Size: 55 KB
PROFITABLE TRADERS ARE NOT LUCKY BUT ARE WELL PREPARED
  • #84
  • Jan 1, 2011 8:18am Jan 1, 2011 8:18am
Quoting netFX
Although during back-testing real spread is not taken into account but this system has SL-10 PIPs and TP-20 PIPs and therefore the win/loss % would be same in demo/live account as back-testing. I am not interested in profit pips but win/loss % at this time and back-testing would reflect the correct win/lost %.

This is for sure...but It's my opinion...and my opinion only that % win/loss is an insufficient data in order to evaluate a strategy..and this strategy in particular doesn't always have a 1:2 risk/reward, even if the stop loss is always 10 pips, Take Profit is variable between 10 an 20 pips (i'm talking about Crodzilla settings), this is the reason why you need to have real expectation of the pips gained/losed taking in account also the spread, a 10 pips SL will never be a 10 pips SL, it will be 10pip+spread...maybe I'm missing something...but this is how I see it...
Market is like a puzzle, YOU have to fit the pieces!!
  • #85
  • Jan 1, 2011 8:27am Jan 1, 2011 8:27am
Quoting Chris_B
This is for sure...but It's my opinion...and my opinion only that % win/loss is an insufficient data in order to evaluate a strategy..and this strategy in particular doesn't always have a 1:2 risk/reward, even if the stop loss is always 10 pips, Take Profit is variable between 10 an 20 pips (i'm talking about Crodzilla settings), this is the reason why you need to have real expectation of the pips gained/losed taking in account also the spread, a 10 pips SL will never be a 10 pips SL, it will be 10pip+spread...maybe I'm missing something...but this...
I agree.
Real pips gain data can be generated only by running demo/live account.
Thanks for your feedback.
AshN
  • #86
  • Jan 1, 2011 9:42am Jan 1, 2011 9:42am
Quoting Chris_B
this strategy in particular doesn't always have a 1:2 risk/reward, even if the stop loss is always 10 pips, Take Profit is variable between 10 an 20 pips (i'm talking about Crodzilla settings), this is the reason why you need to have real expectation of the pips gained/losed taking in account also the spread, a 10 pips SL will never be a 10 pips SL, it will be 10pip+spread
It depends on which strategy you are talking about. There are four now: Dreamliner, zamanib, crodzilla, and Chris_B. They all have different rules - some differences are subtle, some are rather significant.

And it does not appear that crodzilla's take profit is variable. In fact, his take profit can go into the next bar which is completely different from what Dreamliner, zamanib are manually doing in the original thread.

Quote
...maybe I'm missing something...but this is how I see it...

And everyone is entitled to their own opinion. Productive discussions regarding differing opinions can only be helpful.

Quote
In order to have a better quality you need to use real tick data...for example those provided by Dukascopy,

Of course. tazmaan and myself both use Dukascopy tick data - it appears to be very good data, considering it is free.

I have used the Dukascopy tick data to recreate range bars and regular time bars to compare to my broker's charts and the differences are minimal. In fact, the data is surprisingly very realistic.

It is particularly useful when testing during news times. During many high impact news releases, the spreads and price spike. Using tick data, one can really see how these spikes impact the strategy performance. Running in a simulator, one can analyze what happens during these spikes and, for example, see why a system with a 20 pip SL can have a 40 pip loss during news time.

Quote
in any case you will need Bid data and Ask data to take in account the spread

That's usually what tick data includes - the bid, ask, and timestamp for each tick.

Quote
and in any case the spread will never be the same you have with your broker...so the backtest will never be realistic enough...

Of course the spread will not be the same as your broker. However, as you pointed out it is "real" tick data supplied by Dukascopy. Is it 100% perfect? Probably not. Is backtesting with this tick data going to be 100% accurate? Probably not...but it is "realistic enough" to give a good indication of strategy's profitability.
  • #87
  • Edited 10:58am Jan 1, 2011 10:33am | Edited 10:58am
Quoting zznbrm
It depends on which strategy you are talking about. There are four now: Dreamliner, zamanib, crodzilla, and Chris_B. They all have different rules - some differences are subtle, some are rather significant.

And it does not appear that crodzilla's take profit is variable. In fact, his take profit can go into the next bar which is completely different from what Dreamliner, zamanib are manually doing in the original thread.
I don't want to argue with you backtesting experience,,,it's surely better then mine...
In any case I prefer a good forward testing...yes..it takes time...

I don't have a strategy here, I just created an EA that can be configured with all the variations of the trategy, that are only 3:

Dreamliner's: with 50CRB, minimum trigger 17pips, maximum trigger 25 pips, stop placed at the low/high of the bar and TP at the close of the bar.

Zamanib's: with 50CRB, minimum trigger 20pips, maximum trigger 30pips, stop fixed to 30pips and TP at the close of the bar.

Crodzilla's EA: with 30CRB, minimum trigger 10pips, maximum trigger 20pips, stop fixed to 10pips and TP fixed to 20pips (I was wrong on this before)

My EA is just a tool to automate these 3 strategies, you can configure it as you want.

I'm not saying these strategies are surely profitables or that they are not...I'm just putting together an EA...
Market is like a puzzle, YOU have to fit the pieces!!
  • #88
  • Jan 1, 2011 10:52am Jan 1, 2011 10:52am
I did not know we were arguing!

Anyhow, your efforts are greatly appreciated!

Quoting Chris_B
I don't want to argue with you backtesting experience,,,it's surely better then mine...
In any case I prefer a good forward testing...yes..it takes time...

I don't have a strategy here, I just created an EA that can be configured with all the variations of the trategy, that are only 3:

Dreamliner's: with 50CRB, minimum trigger 17pips, maximum trigger 25 pips, stop placed at the low/high of the bar and TP at the close of the bar.

Zamanib's: with 50CRB, minimum trigger 20pips, maximum trigger 30pips, stop fixed to 30pips and TP at the close...
  • #89
  • Jan 1, 2011 11:47am Jan 1, 2011 11:47am
Quoting zznbrm
I did not know we were arguing!

Anyhow, your efforts are greatly appreciated!

No problem mate!!


I'm ready with a new update of the EA...but I need to test it when market is open before release it...
Market is like a puzzle, YOU have to fit the pieces!!
  • #90
  • Jan 1, 2011 12:43pm Jan 1, 2011 12:43pm
After banging my head for over a week i found what seems to be a good range bar indicator. Instruction included. Now just have to figure out how to get the EA's working with it. Thanks a bunch for the work everyone is putting into this system!!!


Quoting Chris_B
No problem mate!!


I'm ready with a new update of the EA...but I need to test it when market is open before release it...
Attached File(s)
File Type: ex4 range_bars_wlas_v09.ex4   9 KB | 281 downloads
File Type: txt instructions.txt   < 1 KB | 424 downloads
  • #91
  • Jan 1, 2011 1:04pm Jan 1, 2011 1:04pm
Quoting netFX
with your solution back-testing on off-line chart is working.
I have noticed one thing though which is the modeling quality is like ~20%. I don't know with this modeling quality how reliable back-testing results would be.
Ignore this. Yes you should use tickdata, but what you are getting is not 20% accurate either, it's still the typical 90%.
The reason it's saying ~20% is because strategy tester can't use m1 to build the 'm5' data, so it's setting this accuracy low. It doesn't know that you already built the m5 data for it based on m1 data. Not sure if that makes sense, but that's the deal
  • #92
  • Jan 1, 2011 1:12pm Jan 1, 2011 1:12pm
Quoting zamanib
you data is wrong wrong it is not possible to do back test that way -- the way to back test is physically look at both chart bars for bar-- match the rangebars chart with a regular chart
Which data is "wrong wrong"?

Why would you say it is not possible to do "back test that way"? Do you even understand how I backtested?

Here is a video of a simulated trade of Dreamliner's strategy using tick data. I used a fast simulation speed since the trade took over 4 hours to close. After viewing it, can you please explain to me why the test is invalid? Why is using ticks to recreate the rangebars and trades incorrect?

Inserted Video
  • #93
  • Jan 1, 2011 2:16pm Jan 1, 2011 2:16pm
Quoting tazmaan
Ignore this. Yes you should use tickdata, but what you are getting is not 20% accurate either, it's still the typical 90%.
The reason it's saying ~20% is because strategy tester can't use m1 to build the 'm5' data, so it's setting this accuracy low. It doesn't know that you already built the m5 data for it based on m1 data. Not sure if that makes sense, but that's the deal
For back-testing we are labeling chart to M5 although it is generated from M1. This chart data generated from M1 is back-tested in M5 on MT4 therefore, platform may be sampleing data as M5 timeframe to determine modelling quality and causing less number of samples and thus reporting almost 1/5th of modeling quality. Perhaps, this is the reason.
AshN
  • #94
  • Jan 1, 2011 2:49pm Jan 1, 2011 2:49pm
Quoting zznbrm
Which data is "wrong wrong"?

Why would you say it is not possible to do "back test that way"? Do you even understand how I backtested?

Here is a video of a simulated trade of Dreamliner's strategy using tick data. I used a fast simulation speed since the trade took over 4 hours to close. After viewing it, can you please explain to me why the test is invalid? Why is using ticks to recreate the rangebars and trades incorrect?

http://www.youtube.com/watch?v=IJwzAowpBFY
Yes..what you are doing is surely the most accurate backtesting possible...
Can you run an optimization to see what would be the better settings?
Market is like a puzzle, YOU have to fit the pieces!!
  • #95
  • Jan 1, 2011 4:01pm Jan 1, 2011 4:01pm
Quoting zamanib
Here are the 2 indicators. "offline ohlc global" goes on the Range Bar chart and "DL Lines and RangeBars" goes on any other chart chosen as long as it is the same symbol.

PS, I am not recommending any type of trading and am not responsible in any way for how these will be used or any information disclosed. My advise is my opinion only.
does not match up see the charts[/quote]

Mine works fine. Also DL Line and Range Bars works as visual on current bar chart as well
Thanks SM Jones
Q
Attached Image(s) (click to enlarge)
Click to Enlarge Name: gu sm jones indy.gif Size: 12 KB
Click to Enlarge Name: gu 5m.gif Size: 14 KB
The Only Constant Is Change!
  • #96
  • Jan 1, 2011 5:41pm Jan 1, 2011 5:41pm
Quoting zznbrm
It depends on which strategy you are talking about. There are four now: Dreamliner, zamanib, crodzilla, and Chris_B. They all have different rules - some differences are subtle, some are rather significant.

And it does not appear that crodzilla's take profit is variable. In fact, his take profit can go into the next bar which is completely different from what Dreamliner, zamanib are manually doing in the original thread.
Based on a 30 pip bar, the take profit is based on the min threashold. If you set it to 10, the stop-loss will be 10 pips and the take-profit will be 20... coded to be the barsize minus min threashold. So, no its not variable per se, its based on your min setting, which is what I interpreted Dreamliners strategy to be as he explained on post #1. Stop-loss plus take-profit will always be your bar size.

If its not, I apologize. Please do not use my EA.
  • #97
  • Jan 1, 2011 9:55pm Jan 1, 2011 9:55pm
Quoting crodzilla
Based on a 30 pip bar, the take profit is based on the min threashold. If you set it to 10, the stop-loss will be 10 pips and the take-profit will be 20... coded to be the barsize minus min threashold. So, no its not variable per se, its based on your min setting, which is what I interpreted Dreamliners strategy to be as he explained on post #1. Stop-loss plus take-profit will always be your bar size.
I think Chris_B correctly summarizes Dreamliner's strategy a few posts back.

Quote
If its not, I apologize. Please do not use my EA.

I don't think an apology is necessary. As far as using your EA...Ok, I won't use it.

Good luck.
  • #98
  • Jan 1, 2011 10:28pm Jan 1, 2011 10:28pm
Quoting crodzilla
I usually don't post code. But I will if asked.
Hi crodzilla

If you agree, I would like to use your base code for the dreamliner ea to experiment with alternate strategies. My Mql skills are v limited and using your code will make it possible to copy/ paste/ implement/ test alternating strategies. I tried going through the other mpas ea posted here but it has too much money management stuff thats intimidating me. Thanks
  • #99
  • Jan 1, 2011 11:36pm Jan 1, 2011 11:36pm
Quoting Chris_B
No problem mate!!


I'm ready with a new update of the EA...but I need to test it when market is open before release it...
Hi Chris

Thanks for your EA v 1.3. I tested this over the w.e and found that it opens long positions much higer than the open price of the current bar. The sells are ok, but there seems to be a problem with stop loss and profit targets on both short and longs. Just feedback for you to consider in your next ver.
  • #100
  • Jan 2, 2011 6:09am Jan 2, 2011 6:09am
Quoting msood
Hi Chris

Thanks for your EA v 1.3. I tested this over the w.e and found that it opens long positions much higer than the open price of the current bar. The sells are ok, but there seems to be a problem with stop loss and profit targets on both short and longs. Just feedback for you to consider in your next ver.
Hi!!

It's not a bug...RangeBars Charts are printed using only BID data, when you open a BUY trade...you open it at the ASK price, you need to take in consideration the spread, so the Buy Stop is placed at Open + spread. The spread exists and you can't ignore it...
For example, if you want to open a trade when price goes above 1.50000 you need to place your order at 1.50000+spread, or your order will trigger before that price (1.50000 - spread).
The same for stoploss and takeprofit, they take in consideration the spread.
Maybe I worry too much about details...but that's how market works...so my EAs take always in consideration the spread...

I can make a swich to turn off the "spread calculation"...if you prefere using it that way...

Please note that by default TakeProfit is setted to 0, this means that TakeProfit is not fixed, but it's projected at the close of the current bar and it can change during the bar life. The same if you set Stoploss to 0, it will change following the high/low of the current bar, until the conditions for trade remain valid.

That said...it's surely possible that the code can contain some errors...if you find something strange report it and I will check...

I think tomorrow I will post the new version of the EA, and I think it will be the final version, because now the EA does what it is supposed to do...I will post also the .set files ready for every variation of the strategy.

Hope this helps...
Market is like a puzzle, YOU have to fit the pieces!!
4 traders viewing now