Heiken Ashi Smoothed Cloud – Day Trading System

• 03/03/2021 at 4:45 PM #162998
  SB-FO

  With the help of the community (Nonetheless, Nicolas, & Vonasi) I was able to turn the smoothed Heiken Ashi smoothed : Forum ProBuilder support – ProRealTime (prorealcode.com) from the blog into an equity strategy.  I typically trade the leveraged small cap ETF TNA in the USA and so the backtest is on it.

  I had luck going back 3 years on a 5 minute chart, but could not get to perform in other timeframes or longer and shorter horizons.  I think something is here, but t needs works and I am open to ideas.   I am thinking good meat, need sauce…

  Code below and attached.

  DEFPARAM CumulateOrders = False // Cumulating positions deactivated DEFPARAM FLATBEFORE = 093000 DEFPARAM FlatAfter = 154500 capital = 100000 + strategyprofit Equity = capital / close myCurrentProfit = STRATEGYPROFIT Price = CustomClose //t and the deviation of the mean: //t = 0 = average // Simple //t = 1 = exponentialaverage //t = 2 = weightedaverage //t = 3 = wilderaverage //t = 4 = triangularaverage //t = 5 = endpointaverage //t = 6 = timeseriesaverage //t = 7 = Hull (PRT v11 only) //t = 8 = ZeroLag (PRT v11 only) P = 19 // p P1 = 14 // p1 T = 5 // endpointaverage T1 = 4 // triangularaverage if barindex > p + 1 then mo = average[p,t](open) mc = average[p,t](close) ml = average[p,t](low) mh = average[p,t](high) //endif once haopen=mo haclose=(mo+mc+ml+mh)/4 haopen=(haopen[1]+haclose[1])/2 endif HAopen1 = average[p1,t1](haopen) HAclose1 = average[p1,t1](haclose) // Draw indicator Graph HAclose1 COLOURED(34,139,3) AS "Heiken Smoothed Cloud Open" Graph HAopen1 COLOURED(225,0,0) AS "Heiken Smoothed Cloud Close" // Conditions to enter long positions IF NOT LongOnMarket AND HAopen1 Crosses Over HAclose1 THEN BUY Equity SHARES AT MARKET ENDIF // Conditions to exit long positions If LongOnMarket AND HAopen1 Crosses Under HAclose1 THEN SELL AT MARKET ENDIF // //Conditions to enter short positions IF NOT ShortOnMarket AND HAopen1 Crosses Under HAclose1 THEN Sellshort Equity SHARES AT MARKET ENDIF // Conditions to exit short positions IF ShortOnMarket AND HAopen1 Crosses Over HAclose1 THEN EXITSHORT AT MARKET ENDIF

  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 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

  DEFPARAM CumulateOrders = False // Cumulating positions deactivated DEFPARAM FLATBEFORE = 093000 DEFPARAM FlatAfter = 154500 capital = 100000 + strategyprofit Equity = capital / close myCurrentProfit = STRATEGYPROFIT   Price = CustomClose   //t and the deviation of the mean: //t = 0 = average // Simple //t = 1 = exponentialaverage //t = 2 = weightedaverage //t = 3 = wilderaverage //t = 4 = triangularaverage //t = 5 = endpointaverage //t = 6 = timeseriesaverage //t = 7 = Hull (PRT v11 only) //t = 8 = ZeroLag (PRT v11 only)   P = 19 // p P1 = 14 // p1 T = 5 // endpointaverage T1 = 4 // triangularaverage   if barindex > p + 1 then mo = average[p,t](open) mc = average[p,t](close) ml = average[p,t](low) mh = average[p,t](high) //endif   once haopen=mo haclose=(mo+mc+ml+mh)/4 haopen=(haopen[1]+haclose[1])/2 endif   HAopen1 = average[p1,t1](haopen) HAclose1 = average[p1,t1](haclose)   // Draw indicator Graph HAclose1 COLOURED(34,139,3) AS "Heiken Smoothed Cloud Open" Graph HAopen1 COLOURED(225,0,0) AS "Heiken Smoothed Cloud Close"   // Conditions to enter long positions   IF NOT LongOnMarket AND HAopen1 Crosses Over HAclose1 THEN BUY Equity SHARES AT MARKET ENDIF   // Conditions to exit long positions If LongOnMarket AND HAopen1 Crosses Under HAclose1 THEN SELL AT MARKET ENDIF // //Conditions to enter short positions IF NOT ShortOnMarket AND HAopen1 Crosses Under HAclose1 THEN Sellshort Equity SHARES AT MARKET ENDIF   // Conditions to exit short positions IF ShortOnMarket AND HAopen1 Crosses Over HAclose1 THEN EXITSHORT AT MARKET ENDIF

   

  2 users thanked author for this post.

  LaurentBZH35, Paul

  

  SB-FO

  Participant

  Topics: 2

  Replies: 8

  Been thanked: 2 times
  03/03/2021 at 5:25 PM #163064
  nonetheless

  If I were you I’d try to re-optimize it for a major index – DJ or NASDAQ for example. Low liquidity  on TNA means the spread is enormous and the margin requirement comparatively high.

  Attached is a quick test with your MM switched off and Spread = 2. Position size = 20, roughly equivalent margin to $1 on the DJI

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1237

  Been thanked: 978 times
  03/03/2021 at 5:59 PM #163068
  nonetheless

  Actually you can ignore the result I attached as I forgot to alter the time settings for the UK. Running it on Wall St hours is better, but not great.

  I think the spread is too high to make this work on that ETF.

  

  nonetheless

  Participant

  Topics: 83

  Replies: 1237

  Been thanked: 978 times
  03/03/2021 at 6:35 PM #163069
  SB-FO

  I did not think of the spread as I normally trade on 4H charts and thus the spreads don’t bother me much as I’m not trying to pluck small gains, instead stay in a trade for days and catch the swing.

  As I said prior, I like the idea of HA Cloud but it needs additional coding to work. I was my first code I have published.

   

  

  SB-FO

  Participant

  Topics: 2

  Replies: 8

  Been thanked: 2 times
  03/06/2021 at 1:57 AM #163252
  Paul

  @SB-FO thanks for publishing.

  I ran it in ORT layout (optimised in step2), for 100k. Dow 1m. long&short, 0.25%sl

  Added a criteria to reduce number of trades. I like to use i.e. close>totalprice

  The thing is, disabling short shows a poor performance for long, disabling long shows a good result for short. Together they have the best results.

  Here are results on 200k on step4. Don’t ditch your setup just yet. Above you have your ingredient for the sauce 🙂

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  03/06/2021 at 3:17 AM #163256
  Paul

  looking at the charts, it’s pretty annoying that using a profittarget limits it’s potential. (I used sl* 1.5 in optimisation)

  Optimising the ts unleashed extra performance while removing the pt.

  Basically it makes sense. A wave of moving averages are always lagging, so if a wave start to move down, generally the index is already down the average. If the additional criteria happens there is enough room for an upswing with a low risk. You are then profiting from fake signals.

   

   

   

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  03/06/2021 at 3:07 PM #163299
  Paul

  there’s still room for improvement.

  2 bad days stood out, dec 31 and nov 03 us election

  Ignoring those 2 days and looking at the start& endtime improved it further. Check the stoplos for both up to 0.5%

  So it was optimised (core) on 100k and values applied on 200k  I didn’t use the equitypart and it’s all based on 1 positionsize.

  There is still doubt, why are results in months (besides november) different on 100k compared to 200k. Probably because a once function and could be a problem.

  Anyway I hope this is a reason to look at the ORT layout because on that idea there was not much interest and the strategy posted by the SB-FO .

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  03/15/2021 at 9:25 PM #164267
  SB-FO

  All, thanks for your feedback and my apologies for not responding sooner. It is ski season here in Utah and that keeping me fit.

  Anyway, Paul, thanks for your input.  Could you be so kind and post your revises code.

  I’ll be posting another strategy which is also promising.

  SB-FO

  

  SB-FO

  Participant

  Topics: 2

  Replies: 8

  Been thanked: 2 times
  03/16/2021 at 4:15 AM #164284
  Paul

  Hi, that’s a nice way to keep fit!  Here’s one that is optimised on up to 200k dow 1minute. It really  shows if loading it on 1M bars.

  defparam cumulateorders = false defparam preloadbars = 2000 ONCE a1 = 50 ONCE a2 = 40 ONCE a3 = 6 once a4 = 8 once positionsize=1 // excluding trading days if year=2020 and month=11 and day=3 then // election us tradeday = 0 elsif month=12 and day=31 then tradeday = 0 else tradeday = 1 endif // strategy //t and the deviation of the mean: //t = 0 = average // Simple //t = 1 = exponentialaverage //t = 2 = weightedaverage //t = 3 = wilderaverage //t = 4 = triangularaverage //t = 5 = endpointaverage //t = 6 = timeseriesaverage //t = 7 = Hull (PRT v11 only) //t = 8 = ZeroLag (PRT v11 only) P = a1 // p P1 = a2 // p1 T = a3 // endpointaverage T1 = a4 // triangularaverage if barindex > p+1 then mo = average[p,t](open) mc = average[p,t](close) ml = average[p,t](low) mh = average[p,t](high) //endif haopen=mo haclose=(mo+mc+ml+mh)/4 haopen=(haopen[1]+haclose[1])/2 endif HAopen1 = average[p1,t1](haopen) HAclose1 = average[p1,t1](haclose) condbuy =HAopen1 Crosses Over HAclose1 and close>totalprice and close[1]<totalprice[1] condsell=HAopen1 Crosses Under HAclose1 and close<totalprice and close[1]>totalprice[1] // ctime = time >= 060500 and time < 215400 ctime = ctime and tradeday // entry if ctime then if condbuy then buy positionsize contract at market set stop %loss 0.5 set target %profit 0.75 elsif condsell then sellshort positionsize contract at market set stop %loss 0.5 set target %profit 0.75 endif endif // if dayofweek=5 and opentime>=215400 then sell at market exitshort at market endif // if positionperf(0)*100>0.4 then if haclose1 crosses over haopen1 then exitshort at market endif if haclose1 crosses under haopen1 then sell at market endif endif

  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 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87

  defparam cumulateorders = false defparam preloadbars    = 2000    ONCE a1 = 50 ONCE a2 = 40 ONCE a3 = 6 once a4 = 8   once positionsize=1   // excluding trading days if year=2020 and month=11 and day=3 then // election us tradeday = 0 elsif month=12 and day=31 then tradeday = 0 else tradeday = 1 endif   // strategy //t and the deviation of the mean: //t = 0 = average // Simple //t = 1 = exponentialaverage //t = 2 = weightedaverage //t = 3 = wilderaverage //t = 4 = triangularaverage //t = 5 = endpointaverage //t = 6 = timeseriesaverage //t = 7 = Hull (PRT v11 only) //t = 8 = ZeroLag (PRT v11 only) P = a1 // p P1 = a2 // p1 T = a3 // endpointaverage T1 = a4 // triangularaverage   if barindex > p+1  then mo = average[p,t](open) mc = average[p,t](close) ml = average[p,t](low) mh = average[p,t](high) //endif haopen=mo haclose=(mo+mc+ml+mh)/4 haopen=(haopen[1]+haclose[1])/2 endif HAopen1 = average[p1,t1](haopen) HAclose1 = average[p1,t1](haclose)   condbuy =HAopen1 Crosses Over HAclose1 and close>totalprice and close[1]<totalprice[1] condsell=HAopen1 Crosses Under HAclose1 and close<totalprice and close[1]>totalprice[1]   // ctime = time >= 060500 and time < 215400 ctime = ctime and tradeday   // entry if ctime then if condbuy then buy positionsize contract at market set stop %loss 0.5 set target %profit 0.75 elsif condsell then sellshort positionsize contract at market set stop %loss 0.5 set target %profit 0.75 endif endif   // if dayofweek=5 and opentime>=215400 then sell at market exitshort at market endif   // if positionperf(0)*100>0.4 then if haclose1 crosses over haopen1  then exitshort at market endif   if haclose1 crosses under haopen1 then sell at market endif endif

   

   

  3 users thanked author for this post.

  Midlanddave, Ryugin

  

  Paul

  Participant

  Topics: 40

  Replies: 812

  Been thanked: 660 times
  03/16/2021 at 9:20 PM #164383
  SB-FO

  Thanks Paul.  I will play with it.

  

  SB-FO

  Participant

  Topics: 2

  Replies: 8

  Been thanked: 2 times
  03/17/2021 at 11:50 AM #164453
  Ryugin

  Thank you very much for sharing this strategy with us, it looks very promising and the overfit risk is really low as log as this is a simple stragegy without indicators (assumming that Heiken Ashi is not an indiciator).

   

  I wonder if it could be interesting to include a vectorial component to this strategy, for example the angle and slope of the crossing between the upper and the lower Heiken Ashi Clould lines in order to avoid false signals when the angle is not big enough. What do you think? I’ll try to program this vectorial Heikin Cloud by myself but as long as my programming skills are very limited, maybe someone wants to take a step ahead and program it by him/herself and could share it whith us.

  

  Ryugin

  Participant

  Topics: 1

  Replies: 21

  Been thanked: 4 times
  03/17/2021 at 3:23 PM #164498
  SB-FO

  Ryugin, that is a great idea.  I have thought of using the slop curve / crossing, alas my coding skills are not good enough to know how to do that.  Perhaps indeed another member of the community can help out.  Nicolas & Vonasi might have some insights.

  Cheers, SB-FO

   

  

  SB-FO

  Participant

  Topics: 2

  Replies: 8

  Been thanked: 2 times
• Author
  Posts