My project for testing and automation

• 06/17/2024 at 1:46 PM #234007
  Morpheous

  Hi guys,

  I have an algorithm I initially want to back test. It uses a 5 minute timeframe with Bollinger bands with 1,2 and 3 deviations from the hourly timescale.

  If my back testing is successful I will want to automate the algorithm for live trading, I trade with IG.com

  A few questions :-

   

  1. Are the ProRealtime programming tools suitable for this project?
  2. How much back testing data is available, on IG on the 5 minute timeframe I can only go back to November 2021, ideally I would like to backtest 5 years
  3. I am an experienced programmer and I’ve done the introductory course for indicator building, what other training would you recommend to prepare me for my project? is the training free of charge or should I pay for it?
  4. Does the Bollinger function support setting different standard deviations from the programming interface as it does from the user interface, if so, how are these arguments specified?

   

  Thanks in anticipation.

   

   

  

  Morpheous

  Participant

  Topics: 25

  Replies: 72

  Been thanked: 11 times
  06/18/2024 at 8:55 AM #234023
  Iván

  Hi!
  Congrats for your first strategy!!!
  I will try to answer your questions:
  1.- Yes, PRT tools allow you to run your bot.
  2.- It depends on the instrument and your account. If you have premium account you can open 3 charts of 1 million candles. As an example, in EURUSD in 5 minutes timeframe you can backtest starting on march 2011.
  3.- In the market place you have an advance training (paid course). Of course, this forum is one of the best places to learn.
  4.- By default Bollinger Bands are defined with 2 std. If you want to change the best way is create your own BBs.

  REM Calculamos la media movil del bollinger y la desviación estandard mediaBollinger = Average[p,type](close) REM Determinamos la desviación estandard desde zero. REM En la práctica, puede utilizar directamente la función STD ! //if barindex >= p-1 then // //sumy2 = 0 //sumy = 0 //for i = 0 to p-1 do //sumy2 = sumy2 + square(close[i]) //sumy = sumy + close[i] //next //desviacion =sqrt(sumy2 / p - square(sumy / p)) // //else // //desviacion = undefined // //endif desviacion = STD[p](close) REM Deducimos las bandas de Bollinger bollSup = mediaBollinger + s * desviacion bollInf =mediaBollinger - s * desviacion return bollSup AS "Boll+", bollInf AS "Boll-"

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

  REM Calculamos la media movil del bollinger y la desviación estandard   mediaBollinger = Average[p,type](close)   REM Determinamos la desviación estandard desde zero. REM  En la práctica, puede utilizar directamente la función STD !   //if barindex >= p-1 then // //sumy2 = 0 //sumy = 0 //for i = 0 to p-1 do //sumy2 = sumy2 + square(close[i]) //sumy = sumy + close[i] //next //desviacion =sqrt(sumy2 / p - square(sumy / p)) // //else // //desviacion = undefined // //endif   desviacion = STD[p](close) REM Deducimos las bandas de Bollinger   bollSup = mediaBollinger + s * desviacion bollInf =mediaBollinger - s * desviacion   return bollSup AS "Boll+", bollInf AS "Boll-"

  1 user thanked author for this post.

  Morpheous

  

  Iván

  Moderator

  Topics: 0

  Replies: 978

  Been thanked: 338 times
  06/18/2024 at 9:48 AM #234032
  Morpheous

  I do not know what kind of account I have, I trade with IG and they give me access to ProRealtime for free, if I execute a certain number of trades per month. I will ask their support, but currently on USD/JPY I can only go back 200k 5 minute periods.

   

  I will probably go ahead and take the advanced coding course, I think for the cost it will give me a faster start.

  

  Morpheous

  Participant

  Topics: 25

  Replies: 72

  Been thanked: 11 times
  06/18/2024 at 12:23 PM #234037
  Morpheous

  I spoke with IG and they told me that to get an account which supports 1 million periods, I must cancel my account with them and sign up with you, then re-open my IG account, is there any way around having to do this?

  

  Morpheous

  Participant

  Topics: 25

  Replies: 72

  Been thanked: 11 times
  06/18/2024 at 2:29 PM #234046
  Iván

  You can contact with PRT by phone and they will answer all your doubts.

  1 user thanked author for this post.

  Morpheous

  

  Iván

  Moderator

  Topics: 0

  Replies: 978

  Been thanked: 338 times
  06/18/2024 at 4:14 PM #234056
  GraHal

  I’ve been using PRT for 8 years and never felt the need for 1,000,000 bars!

  Maybe it’s my age, but too much of life goes by waiting for 1,000,000 bars to optimise.

  Have you tried to optimise your Algo on even 200k bars  … you okay with how long it takes  Morpheous?

  You could likely open a PRT Account on Premium PRT ( to get 1,000,000 bars) using a different email address than you used on your IG Account?

  Get a few Algos coded up and running and then take a view if you even need 1,000,000 bars?

  if you code an Algo to run on 30 mins TF … historical data over 200k bars at 30 mins TF would be more than 1,000,000 bars at 5 Mins.

  Just a few thoughts to consider.

  Happy Coding! 🙂

  1 user thanked author for this post.

  Morpheous

  

  GraHal

  Participant

  Topics: 212

  Replies: 8785

  Been thanked: 2185 times
  06/18/2024 at 4:28 PM #234059
  Morpheous

  It’s a good point about the demo account, thanks for that.

   

  I’m currently back testing 1 year of 5 minute candles by hand, as I want to optimise by actually seeing the chart with my eyes, once I am satisfied with that I will code and indeed run a back test on 200k bars of 5 minute, maybe that will be enough, I will take a view after I have tried it. Probably the more data the better with quantitative  analysis.

  1 user thanked author for this post.

  GraHal

  

  Morpheous

  Participant

  Topics: 25

  Replies: 72

  Been thanked: 11 times
  06/18/2024 at 6:36 PM #234064
  Morpheous

  I tried the code you provided, cool 🙂

   

   

   

  

  Morpheous

  Participant

  Topics: 25

  Replies: 72

  Been thanked: 11 times
• Author
  Posts