DAX 5 min long+short
Forums › ProRealTime English forum › ProOrder support › DAX 5 min long+short
- This topic has 32 replies, 13 voices, and was last updated 6 years ago by T-rader.
-
-
11/15/2017 at 3:38 PM #52846
Has made a simple code for DAX 30 which runs between 09:00 and 17:20.
According to the 100k backtest, is looks ok..
It would be interesting to see a 200k backtest on it 🙂
Spread 1p.
1234567891011121314151617181920212223242526272829303132333435363738394041DEFPARAM CumulateOrders = FalseDEFPARAM FLATBEFORE = 090000DEFPARAM FLATAFTER = 172000// Conditions to enter long positionsindicator1 = BollingerDown[19](close)c1 = (close <= indicator1)indicator2 = Stochastic[15,3](close)c2 = (indicator2 CROSSES UNDER 13)IF c1 AND c2 THENBUY 1 CONTRACT AT MARKETENDIF// Conditions to enter short positionsindicator10 = BollingerUp[21](close)c10 = (close >= indicator10)indicator20 = Stochastic[14,3](close)c20 = (indicator20 CROSSES OVER 70)IF c10 AND c20 THENSELLSHORT 1 CONTRACT AT MARKETENDIF// Stops and targetsSLL = 27TPL = 14if longonmarket thenSET STOP pLOSS SLLsET TARGET pPROFIT TPLENDIFSLS = 30TPS = 10if SHORTonmarket thenSET STOP pLOSS SLSsET TARGET pPROFIT TPSENDIF2 users thanked author for this post.
11/15/2017 at 6:37 PM #5285811/15/2017 at 9:31 PM #5287311/15/2017 at 9:48 PM #5287711/15/2017 at 9:51 PM #52879Sorry Barney, couldn’t resist 🙂
11/16/2017 at 7:13 AM #52905Sorry Barney, couldn’t resist
Haha!!! It so early in the morning and it is awesome to read a joke in the forum! Hahah! XD
11/16/2017 at 7:59 AM #5290811/16/2017 at 2:44 PM #52992Thanks for the 200k backtest!
I do not feel completely satisfied with the result though.
Seems to be difficult to get simple codes to work, it feels like you need a 200 k test to see if algon is something to have or not(especially at 5 min algos).
11/16/2017 at 3:00 PM #52996If it was easy….. 😉
I know exactly how you feel. Before i got the 200K backtest premium for myself, i had several systems that looked so good in 100K, when in 200K tho…
Even made a post about it: https://www.prorealcode.com/topic/creating-strategies-questions-about-backtest-and-periods/In short, it just does not feel sexy when its HORRIBLE before 2016, but SEXY after 😀
I guess that if you curve-fit something there might be a chance that the same curve might work tomorrow.. I mean you can be pretty sure that it will STOP working i guess… and thats why i dont want to run stuff that only looks good on 100K… Even though i have for example the linked system going in demo, making money, i just know that its going to stop working maybe tomorrow, or maybe in a week or maybe in a month. I’d rather have a more robust system than that…
Ps: i do not have alot of experience with live automated systems trading 🙂
11/17/2017 at 11:52 AM #53099All periods and stoploss / takeprofit values have been optimized, that’s why it is curvefitted. You can do optimisation of course, but do WFA to prove robustness of it! The past 100k bars before your own test are pure Out Of Sample data that could validate or not your strategy, but it doesnt.
You have 2 ways to make things better:
- develop your idea on at least 1 In Sample period (70% of data) and test in the next 30%
- if you are in a optimisation process, do a simple WFA with one or many IS/OOS iterations to find the good edge
Come one Barney, I know you can make it!
11/21/2017 at 5:30 PM #53482what is WFA? @Nicolas
11/22/2017 at 9:07 AM #53548https://www.prorealcode.com/blog/learning/prorealtime-walk-analysis-tool/
This was one of the reasons i came to this forum!
Big thanks to nicolas for the good work 🙂11/22/2017 at 4:47 PM #53641Ok, here coms a code that is not over optimized. Simple code again thugh!
DAX 5 min spread 1.
123456789101112131415161718192021DEFPARAM CumulateOrders=Falsedefparam flatbefore = 090000defparam flatafter = 173000indicator1 = ExponentialAverage[46](close)indicator2 = ExponentialAverage[10](close)c1 = (indicator1 > indicator2[1])indicator3 = AroonUp[20]indicator4 = Aroondown[76]c2 = (indicator3 CROSSES OVER indicator4)indicator100 = closeindicator101 = Supertrend[7,1]c100 = indicator100 > indicator101IF c1 and c2 and c100 THENBUY 1 CONTRACTs AT MARKETENDIFset stop ploss 20set target pprofit 10Would be nice if someone would try 200k on it 🙂
11/22/2017 at 10:33 PM #5366111/22/2017 at 11:34 PM #53665 -
AuthorPosts
Find exclusive trading pro-tools on