an EA request for an adaptive break out strategy | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Adaptive Moving Average -- Kaufman 13 replies

Adaptive RSI & Pivot Point Moving Average, anyone? 1 reply

(Non-)Adaptive? 0 replies

Fractive Adaptive 1 reply

Adaptive Price Zone/TASC indicator 4 replies

Attachments: an EA request for an adaptive break out strategy
Exit Attachments

an EA request for an adaptive break out strategy

  • #1
  • First Post: Edited 3:37pm Oct 16, 2010 3:18pm | Edited 3:37pm
Hi, maybe someone would be interested to make an ea for this strategy. It is a simple break out strategy with ability to adapt to changing volatility.

the logic

1.break out distance 50 persent of daily ATR(120)
2. buy or sell when the price reach this break out distance from the daily open price.
3. due to diferent daily open on diferent brokers it should be posible to set the time when the ea considers as daily open.
4. sl 25 persent of daily ATR, tp 50 persent of daily ATR
5. filter, only buy trades if daily open is above 120ema on 1H time frame and only sell trades when daily open is below 120ema on 1H time frame
6. Fast lot size/acording to persentage of the account size.



It is simple and profitable ,not magic, but anyway a profitable method on longer run. To have an EA would be an advantage to test it more with diferent settings and pairs. My declaration for profitability is based on eur/usd pair.
  • #2
  • Edited Oct 18, 2010 11:18am Oct 17, 2010 6:06pm | Edited Oct 18, 2010 11:18am
Saintsurfer contacted me earlier today to see if I could code the ea, so here it is.

Frankly, I did so because I can, not because I wanted to. I hate breakout systems; the number of times breakout turns out to be fakeout is ridiculous.

To my utter blank amazement, the attached has shown up profitably in backtestint. Now, we all know that backtesting is unreliable, to put it kindly but I have discovered one thing: it a robot is unprofitable in backtesting then it will get murdered on demo, but if it survives backtesting then it stands a chance on demo.

So, let's give it a whirl and see what happens. There are a few issues with the bot at the mo:

 

  1. as coded, the sl calculation appears to give the occasional zero stop loss

Inputs, or at least those that are not screamingly obvious to all but the unremittingly simple:

 

  1. SlAtrPercentage/TpAtrPercentage equates to point 4 in SS's post 1 logic.
  2. TradeExpiryHours: I have coded the bot to send a pending order in the direction of the current market movement, so the trade needs a delete time if it has not filled. The default deletes the trade after 12 hours if it remains unfilled.
  3. TradeSendHour:this equates to the start of your trading day and can be whatever you choose it to be.

All I did in coding this bot was adapt my generic auto-trader, so there are a number of management features included. These are mptm features; if you do not understand them then follow the link in my signature and download the mptm user guide.

Have fun. Given that this bot has survived its first hurdle, it might have a future. We shall see.



Edit: alski spotted a bug; the fix is a few posts down, so I have removed this version.

  • #3
  • Oct 17, 2010 6:20pm Oct 17, 2010 6:20pm
SL is 25% atr, tp is 50% atr, is this coded correctly steve (only ask because you mentioned tp wasn't given, but I think it was?)
I see the tpatrpercent input, but I don't see the slatrpercent input
  • #4
  • Oct 17, 2010 6:44pm Oct 17, 2010 6:44pm
Quoting tazmaan
SL is 25% atr, tp is 50% atr, is this coded correctly steve (only ask because you mentioned tp wasn't given, but I think it was?)
I see the tpatrpercent input, but I don't see the slatrpercent input
oops. I really should not code under the affluence of incohol.

I have corrected the code in post 2 and added an Atr take profit instead of the fixed pips thingy.

  • #5
  • Oct 17, 2010 6:52pm Oct 17, 2010 6:52pm
Hehe. I just looked up Saintsurfer to see where he lives - Lithuania. As I write, it is 2.50 AM in L, so he is probably in bed right now.

In for a surprise when he gets up, isn't he?
  • #6
  • Oct 17, 2010 7:06pm Oct 17, 2010 7:06pm
thx, keep enjoyin your booze
Will let u know if it backtests ok for me on tickdata, first version wasn't putting in any stoploss at all, so we'll see how this one does

Well it's putting in stops now, but tons of invalid stop error 130's. Will give it a few days so things can get ironed out
  • #7
  • Oct 17, 2010 8:04pm Oct 17, 2010 8:04pm
Steve,

Just using default settings and I get a ton of these:

Quote
2010.10.17 18:02:132010.10.18 00:59 Saintsurfer Breakout auto trading robot by Steve Hopwood EURUSDFXF,Daily: EURUSDFXF Saintsurfer order send failed with error(130): invalid stops

But, still profitable on EURUSD over 10 years. Just slightly though.
We are our own best indicator.
  • #8
  • Edited 8:21am Oct 18, 2010 6:02am | Edited 8:21am
Quoting SteveHopwood
Hehe. I just looked up Saintsurfer to see where he lives - Lithuania. As I write, it is 2.50 AM in L, so he is probably in bed right now.

In for a surprise when he gets up, isn't he?
Oh nise to see,oh you not sleeping at all. And we already have a product. Thanks so much, time to check it out.
  • #9
  • Oct 18, 2010 7:42am Oct 18, 2010 7:42am
Quoting SteveHopwood
Hehe. I just looked up Saintsurfer to see where he lives - Lithuania. As I write, it is 2.50 AM in L, so he is probably in bed right now.

In for a surprise when he gets up, isn't he?
I have tried to test a little bit and some things that needs to be figured out.

1. Is it important on which time frame to test, because ATR 120 values should be taken from daily and mooving average filter is on 1h time frame
2. The EA takes only long trades (at lest for me)
3. Is it posible to make the break out distance, daily ATR value and moving average filter period adjustable in the settings.
4.For now the sl and tp seems to be much smaller then they should be acording to the persentage of the daily ATR120 I set, while break out distance is right.
  • #10
  • Oct 18, 2010 7:54am Oct 18, 2010 7:54am
Hey all,

it seems that there is a typo (easy on me Steve) in the code for sending short trades.

//Short
if (TradeDirection == down)
{
price = NormalizeDouble(Bid - AtrVal, Digits);
//if (TakeProfit > 0) take = NormalizeDouble(price - (TakeProfit * Point), Digits);
//stop = NormalizeDouble(price + (AtrVal * (TpAtrPercentage / 100) ), Digits);
take = NormalizeDouble(price - (AtrVal * TpAtrPercentage) / 100, Digits);
stop = NormalizeDouble(price - (AtrVal * SlAtrPercentage) / 100, Digits);
type = OP_SELLSTOP;
SendTrade = true;
}//if (Ask > BbUpper)

should be :

stop = NormalizeDouble(price + (AtrVal * SlAtrPercentage) / 100,

Hope this helps.

St
  • #11
  • Oct 18, 2010 9:27am Oct 18, 2010 9:27am
Quoting alski317
Hey all,

it seems that there is a typo (easy on me Steve) in the code for sending short trades.

//Short
if (TradeDirection == down)
{
price = NormalizeDouble(Bid - AtrVal, Digits);
//if (TakeProfit > 0) take = NormalizeDouble(price - (TakeProfit * Point), Digits);
//stop = NormalizeDouble(price + (AtrVal * (TpAtrPercentage / 100) ), Digits);
take = NormalizeDouble(price - (AtrVal * TpAtrPercentage) / 100, Digits);
stop = NormalizeDouble(price - (AtrVal * SlAtrPercentage) / 100, Digits);
type = OP_SELLSTOP;
SendTrade = true;...
Thanks for your input, I wish I could understand coding. Could you upload the version with this issue fixed?
  • #12
  • Edited 11:40am Oct 18, 2010 11:16am | Edited 11:40am
Quoting alski317
Hey all,

it seems that there is a typo (easy on me Steve) in the code for sending short trades.

//Short...
Thanks. Nice bit of bloop-spotting. 'Tis me and my copy/pasting again.

Fixed another couple of bugs as well, to do with the too-tight tp/sl's.

Added the features requested by Saintsurfer in post 9.

Attached File(s)
File Type: mq4 Saintsurfer Breakout auto trading robot by Steve Hopwood.mq4   35 KB | 762 downloads
  • #13
  • Oct 18, 2010 11:44am Oct 18, 2010 11:44am
Quoting Saintsurfer
I have tried to test a little bit and some things that needs to be figured out.

1. Is it important on which time frame to test, because ATR 120 values should be taken from daily and mooving average filter is on 1h time frame
No. Test on any tf.
Quote
2. The EA takes only long trades (at lest for me)
No idea why. It takes trades in both directions here.
Quote
3. Is it posible to make the break out distance, daily ATR value and moving average filter period adjustable in the settings.
Done. The inputs are in the 'Atr and MA' block.

Quote
4.For now the sl and tp seems to be much smaller then they should be acording to the persentage of the daily ATR120 I set, while break out distance is right.
This needed a bit of adjustment. It should be correct now. Sing out if not.

All this in my latest update just above.

Mind, now the code has fewer mistakes, the robot is a sure-fire loser every time.

  • #14
  • Oct 18, 2010 12:10pm Oct 18, 2010 12:10pm
Quoting SteveHopwood
No. Test on any tf.

No idea why. It takes trades in both directions here.

Done. The inputs are in the 'Atr and MA' block.


This needed a bit of adjustment. It should be correct now. Sing out if not.

All this in my latest update just above.

Mind, now the code has fewer mistakes, the robot is a sure-fire loser every time.

just in case you didn`t know, you are a great man, thank you.
  • #15
  • Edited 2:10pm Oct 18, 2010 12:27pm | Edited 2:10pm
It seems it does now almoust everithing what it has to do, just on thing, I noticed it misses mondays trades. Obviously it has something to do with my broker I test on, becouse it`s opening time is the same it has to send the trade. Sure it sends the trade if I set one hour ahead, but it is wrong time then. I use NY close broke such as fxlite for testing. Could it be solved? Maybe if there was posible to set the time also in minutes, so I could set it for example just one minute ahead and the problem would be solved without big inpact for the results...

edit
I see that the EA do not take another trade if previous is not closed yet, could that limitation be taken off. Just tried some other setings for longer term trading and that would help much.
  • #16
  • Oct 19, 2010 10:44am Oct 19, 2010 10:44am
Quoting tazmaan
thx, keep enjoyin your booze
Will let u know if it backtests ok for me on tickdata, first version wasn't putting in any stoploss at all, so we'll see how this one does

Well it's putting in stops now, but tons of invalid stop error 130's. Will give it a few days so things can get ironed out
Hi tazmaan, could you test the second ea with these setings on tick data if it is not too much trouble? Thank you.

SL ATR persentage 25
TP ATR persentage 50
ATR period 120
ATR break out 50
ema period 480
trade send hour GMT 09 pm
trade expiryhours 17
  • #17
  • Oct 19, 2010 11:16am Oct 19, 2010 11:16am
will do, it's second in my backtest list so not sure how long, but i'll let u know when its done.
  • #18
  • Oct 19, 2010 11:35am Oct 19, 2010 11:35am
Here you go SS. Only took a sec, as I realized I wasn't optimizing as I usually am
Attached Image (click to enlarge)
Click to Enlarge Name: SSbreakout.gif Size: 6 KB
Attached File(s)
File Type: zip SSbreakout.zip   17 KB | 569 downloads
  • #19
  • Edited 12:15pm Oct 19, 2010 11:38am | Edited 12:15pm
Quoting tazmaan
Here you go SS. Only took a sec, as I realized I wasn't optimizing as I usually am
Oh thanks a lot, just wonder if it also works on longer period, than a year...

can you tell me where did you get/buy the tester you using or that data?
  • #20
  • Oct 19, 2010 12:26pm Oct 19, 2010 12:26pm
http://eareview.net/tick-data

it's a long, annoying and tedious process, but to me it's worth it and necessary if you put any weight on backtesting at all.
The data provided by metaquotes is pretty much worthless imo

Later today I will find some time to process older tickdata for you, and run longer backtests. There's a limitation on file size for mt4, so i might have to break it into yearly backtests, but should still give u enough info.
0 traders viewing now