Buy when price goes up, sell when price goes down…

• 06/25/2018 at 5:45 PM #74337
  AlgoAlex

  Hi all,

  I would like to share my work on what should be a simple strategy to gain profit in market, in particular forex.

  Let’s start from this statement:

  Buy when prices are moving up. Buy each retrace/dip. Keep buying until the last retrace becomes a trend change which is the one trade you lose on.
  
  Sell when prices are moving down. Sell each retrace/rally. Keep selling until the last retrace becomes a trend change which is the second trade you lose on.

   

  Looks easy isn’t it?

  Why forex? Because currencies pairs cannot go to zero, neither raise to the moon, they must stay in ranges determined by central banks politics, government actions etc. They trend for days or weeks, giving multiple opportunities to enter.

  First of all, we must be able to identify if there is a trend. Due to the lack of mtf on PRT, I used a little trick. I use RSIOMA  (https://www.prorealcode.com/prorealtime-indicators/rsioma-indicator/) with a multiplier for higher timeframe.

  Example:

  ma=ExponentialAverage[14*tfx](WeightedClose) rsioma=rsi[14*tfx](ma) marsioma=ExponentialAverage[21*tfx](rsioma)

  1 2 3

  ma=ExponentialAverage[14*tfx](WeightedClose) rsioma=rsi[14*tfx](ma) marsioma=ExponentialAverage[21*tfx](rsioma)

  where “tfx” is a round number thet gives me an higher tf rsioma on my tf. f.e. if I am on a 1H chart, tfx=24 give rsioma of daily tf.

  bullish=rsioma>rsioma[1] and rsioma>marsioma bearish=rsioma<rsioma[1] and rsioma<marsioma longtrendbullish=summation[10](bullish)=10 longtrendbearish=summation[10](bearish)=10

  1 2 3 4

  bullish=rsioma>rsioma[1] and rsioma>marsioma bearish=rsioma<rsioma[1] and rsioma<marsioma longtrendbullish=summation[10](bullish)=10 longtrendbearish=summation[10](bearish)=10

  with these conditions I can determine if overall trend is bullish or bearish.

   

  Now, I wanna get an entry signal in the direction of bigger trend. This is where we can speculate the most (rsi, adx, moving averages crossover…) but I personally pick Leo’s moving average (https://www.prorealcode.com/prorealtime-indicators/leo-moving-average-supportresistance/) because, according to his own words “it is an indicator that do not move so quickly as the Hull Moving Average but no lagging like exponential or weighted moving average”.

  This is the base for developing a strategy (or strategies) discussion:

  //rsioma higher timeframe ma=ExponentialAverage[14*tfx](WeightedClose) rsioma=rsi[14*tfx](ma) marsioma=ExponentialAverage[21*tfx](rsioma) bullish=rsioma>rsioma[1] and rsioma>marsioma bearish=rsioma<rsioma[1] and rsioma<marsioma longtrendbullish=summation[10](bullish)=10 longtrendbearish=summation[10](bearish)=10 //Leo moving average for retracement //period=20 //Formula: LMA=WMA+(WMA-SMA) LMA=2*weightedaverage[period](close)-average[period](close) smoothLMA=weightedaverage[period](LMA) entrylong=LMA crosses over smoothLMA entryshort=LMA crosses under smoothLMA if longtrendbullish and entrylong then DRAWARROWUP(BARINDEX,LOW-0.003) coloured(0,100,155) endif if longtrendbearish and entryshort then DRAWARROWDOWN(BARINDEX,HIGH+0.003) coloured(255,100,155) endif return

  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

  //rsioma higher timeframe ma=ExponentialAverage[14*tfx](WeightedClose) rsioma=rsi[14*tfx](ma) marsioma=ExponentialAverage[21*tfx](rsioma)   bullish=rsioma>rsioma[1] and rsioma>marsioma bearish=rsioma<rsioma[1] and rsioma<marsioma longtrendbullish=summation[10](bullish)=10 longtrendbearish=summation[10](bearish)=10   //Leo moving average for retracement //period=20 //Formula: LMA=WMA+(WMA-SMA) LMA=2*weightedaverage[period](close)-average[period](close) smoothLMA=weightedaverage[period](LMA)   entrylong=LMA crosses over smoothLMA entryshort=LMA crosses under smoothLMA     if longtrendbullish and entrylong then DRAWARROWUP(BARINDEX,LOW-0.003) coloured(0,100,155) endif   if longtrendbearish and entryshort then DRAWARROWDOWN(BARINDEX,HIGH+0.003) coloured(255,100,155) endif   return

  Would like to have feedbacks and ideas about it.

  Thanks!

  

  AlgoAlex

  Participant

  Topics: 33

  Replies: 64

  Been thanked: 25 times
  06/25/2018 at 7:28 PM #74345
  GraHal

  Well it can do what you set out to do and on Indices! 🙂

  If you have 200k bars then adjust the settings on the itf?

  It runs soooo so quick, it’s a pleasure to optimise!! 🙂

  Great Work Alex
  GraHal

   

   

  

  GraHal

  Participant

  Topics: 212

  Replies: 8789

  Been thanked: 2187 times
  06/26/2018 at 9:19 AM #74367
  GraHal

  I’m running the code below in Demo and it keeps getting stopped due to insufficient history.

  The biggest numbers of bars I can see required is 21 x 240 = 5040 so I’ve set preload bars at 5500.

  I have tried preload bars at 7000 and 15000 and it got stopped each time.

  Is there a maximum number of preload bars does anybody know please??

  Anybody any ideas why  the System is being stopped due to not enough History when I have tried 5500, 7000 and 15000  preload bars and System is stopped each time??

  PS / Edit  was just stopped again at 5500 preload bars!
  I’m going to try 30000 preload now!!!!

  //rsioma higher timeframe DEFPARAM CUMULATEORDERS = False DEFPARAM PRELOADBARS = 5500 tfx = 240 ma=ExponentialAverage[14*tfx](WeightedClose) rsioma=rsi[14*tfx](ma) marsioma=ExponentialAverage[21*tfx](rsioma) bullish=rsioma>rsioma[1] and rsioma>marsioma bearish=rsioma<rsioma[1] and rsioma<marsioma longtrendbullish=summation[10](bullish)=10 longtrendbearish=summation[10](bearish)=10 //Leo moving average for retracement period=50 //Formula: LMA=WMA+(WMA-SMA) LMA=2*weightedaverage[period](close)-average[period](close) smoothLMA=weightedaverage[period](LMA) entrylong=LMA crosses over smoothLMA entryshort=LMA crosses under smoothLMA if longtrendbullish and entrylong then Buy at Market endif if longtrendbearish and entryshort then SellShort at Market endif SET TARGET PPROFIT 15 SET STOP PLOSS 330

  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 31 32 33

  //rsioma higher timeframe DEFPARAM CUMULATEORDERS = False DEFPARAM PRELOADBARS = 5500 tfx = 240 ma=ExponentialAverage[14*tfx](WeightedClose) rsioma=rsi[14*tfx](ma) marsioma=ExponentialAverage[21*tfx](rsioma)   bullish=rsioma>rsioma[1] and rsioma>marsioma bearish=rsioma<rsioma[1] and rsioma<marsioma longtrendbullish=summation[10](bullish)=10 longtrendbearish=summation[10](bearish)=10   //Leo moving average for retracement period=50 //Formula: LMA=WMA+(WMA-SMA) LMA=2*weightedaverage[period](close)-average[period](close) smoothLMA=weightedaverage[period](LMA)   entrylong=LMA crosses over smoothLMA entryshort=LMA crosses under smoothLMA     if longtrendbullish and entrylong then Buy at Market endif   if longtrendbearish and entryshort then SellShort at Market endif   SET TARGET PPROFIT 15 SET STOP PLOSS 330

   

   

  

  GraHal

  Participant

  Topics: 212

  Replies: 8789

  Been thanked: 2187 times
  06/26/2018 at 10:53 AM #74398
  GraHal

  Stopped again even with 30000 preload bars.

  Shouldn’t this Topic be on the ProOrder Support Forum as it is discussing code issues?

  PS / Edit

  I’ve sent a Technical Report to PRT.

  

  GraHal

  Participant

  Topics: 212

  Replies: 8789

  Been thanked: 2187 times
  06/26/2018 at 11:18 AM #74402
  Eric

  Not sure but i think i read something about preload bars and that there is a limit of how many you can use?

  maybe Nicolas knows?

  1 user thanked author for this post.

  GraHal

  

  Eric

  Participant

  Topics: 18

  Replies: 441

  Been thanked: 84 times
  06/26/2018 at 11:52 AM #74410
  robertogozzi

  In line 5 MA                 is calculated with 3360 (14 * 240) periods.

  In line 6 RSIOMA       is calculated with 3360 (14 * 240) periods of the MA calculated with 3360 periods!

  In line 7 MARSIOMA is calculated with 5040 (21 * 240) periods of the RSIOMA calculated with 3360 that had been calculated with MA calculated on 3360 periods!!

  I’m  afraid they need to be internally multiplied, hence… no way to run it!!!!!

   

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 17142

  Been thanked: 4032 times
  06/26/2018 at 12:40 PM #74422
  GraHal

  Thank You Roberto

  robertogozzi wrote:

  I’m  afraid they need to be internally multiplied, hence… no way to run it!!!!!

  So the code is flawed then?

  Can you / anybody  suggest any way to write Line 5, 6 and 7 so it will run please?  What do you mean internally multiplied??

  I had already tried below and it got rejected … as expected due to Roberto obsevations.

  ma=ExponentialAverage[3360](WeightedClose)//14*tfx rsioma=rsi[3360](ma)//14*tfx marsioma=ExponentialAverage[5040](rsioma)//21*tfx

  1 2 3

  ma=ExponentialAverage[3360](WeightedClose)//14*tfx rsioma=rsi[3360](ma)//14*tfx marsioma=ExponentialAverage[5040](rsioma)//21*tfx

   

   

  

  GraHal

  Participant

  Topics: 212

  Replies: 8789

  Been thanked: 2187 times
  06/26/2018 at 12:53 PM #74427
  GraHal

  Aha do you mean like this …

  //ma=ExponentialAverage[3360](WeightedClose)//14*tfx rsioma=rsi[3360](ExponentialAverage[3360](WeightedClose))//14*tfx marsioma=ExponentialAverage[5040](rsi[3360](ExponentialAverage[3360](WeightedClose)))//21*tfx

  1 2 3

  //ma=ExponentialAverage[3360](WeightedClose)//14*tfx rsioma=rsi[3360](ExponentialAverage[3360](WeightedClose))//14*tfx marsioma=ExponentialAverage[5040](rsi[3360](ExponentialAverage[3360](WeightedClose)))//21*tfx

   

  

  GraHal

  Participant

  Topics: 212

  Replies: 8789

  Been thanked: 2187 times
  06/26/2018 at 1:16 PM #74428
  GraHal

  No above is no good … rejected again, same reason … not enough historical data.

  

  GraHal

  Participant

  Topics: 212

  Replies: 8789

  Been thanked: 2187 times
  06/26/2018 at 1:34 PM #74431
  robertogozzi

  No GraHal, I meant internally they will probably need to multiply 3360 * 3360 * 5040 = ….. billion bars!

  Should PRT answer giving different reasons, please let us know.

  1 user thanked author for this post.

  GraHal

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 17142

  Been thanked: 4032 times
  06/26/2018 at 1:43 PM #74433
  Nicolas

  Maximum preloadbars is 10 thousands bars (10.000). So, even if you set more than this limit, the maximum bars loaded by ProOrder will always be 10k.

  2 users thanked author for this post.

  robertogozzi, GraHal

  

  Nicolas

  Keymaster

  Topics: 81

  Replies: 20863

  Been thanked: 4773 times
  06/26/2018 at 2:09 PM #74435
  AlgoAlex

  Apart from platform limitations, I don’t see useful using a daily trend for 15M f.e.

  I personally don’t go over 2 superior timeframes.

   

  By the way GraHal, try to remove marsioma from code and use only rsioma’s slope for trend. This should reduce number of bars for computation.

  

  AlgoAlex

  Participant

  Topics: 33

  Replies: 64

  Been thanked: 25 times
  06/26/2018 at 2:16 PM #74436
  JC_Bywan

  That’s the way I simulate mtf in my codes (although with diffrent multipliers and not rsi but that’s not the point). Grahal with those sets of 14 21 and 240 values, it’s almost as Roberto said, except you’dd add rather than multiply: here you’d need a full first set of 3360 bars before being able to have ma for the first time (before that it doesn’t exist), then from bar 3360 it takes another set of 3360 to have your first rsioma calculated on ma, that’s 6720 (should be 6719 because starting from 3360 rather than 3361 but let’s not get hung up on that), and only from then would you start counting the 21tfx=5040 to get your first marsioma, so that’s 6720+5040=11760 (well, more accurately 11758). And you’re above the 10k magic number, but not by much, so doesn’t have to be forgotten completely, just adapted to a bit less as long as you’d be happy with less.

  As a side point, even if not relevant here and now but you might bump into the problem later on, another one to take into account when simulating mtf in such way you’re borderline with limits, it is the total history available in the chosen timeframe of your instrument… you may have your 10k bars available in a smallish timeframe, and use multipliers that keep you below 10k, but when trying it in a bigger timeframe, like (x) days or weekly, IG might not have a long enough history to get your first higher timeframe simulated value, regardless of the preloadbars set to max. For later on, just in case…

  Switching off for world cup now, and going on holidays. Enjoy your summer guys.

  3 users thanked author for this post.

  robertogozzi, Nicolas, GraHal

  

  JC_Bywan

  Moderator

  Topics: 20

  Replies: 2651

  Been thanked: 915 times
  06/26/2018 at 2:46 PM #74442
  GraHal

  Ha it worked @Noobywan … have a well earned holiday! 🙂

  Thank You
  GraHal

  

  GraHal

  Participant

  Topics: 212

  Replies: 8789

  Been thanked: 2187 times
• Author
  Posts