Wrong Percentage Calculation

• 10/21/2017 at 8:36 PM #50153
  Vonasi

  BodyPerc = ((ABS(close - open))/(High-Low))*100 IF BodyPerc <= 25 THEN R = 255 G = 0 B = 0 ELSE IF BodyPerc > 25 and BodyPerc < 75 THEN R = 0 G = 0 B = 255 ELSE R = 0 G = 255 B = 0 ENDIF ENDIF Return BodyPerc coloured (R,G,B) Style(Histogram,2) as "Body Percentage"

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

  BodyPerc = ((ABS(close - open))/(High-Low))*100   IF BodyPerc <= 25 THEN R = 255 G = 0 B = 0 ELSE IF BodyPerc > 25 and BodyPerc < 75 THEN R = 0 G = 0 B = 255 ELSE R = 0 G = 255 B = 0 ENDIF ENDIF     Return BodyPerc coloured (R,G,B) Style(Histogram,2) as "Body Percentage"

  I wrote a simple indicator to show what percentage the body of a candle is compared to the total range but I’m getting some rogue percentages as the data for some candle is showing their range to be smaller than the candle body – which is impossible. Why is this?

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  10/21/2017 at 9:11 PM #50159
  Despair

  Your code is fine. This is obviously an error in the data stream. Was this on an demo account?

  

  Despair

  Blocked

  Topics: 97

  Replies: 1034

  Been thanked: 217 times
  10/21/2017 at 9:32 PM #50160
  robertogozzi

  Try replacing line 8 with:

  IF BodyPerc < 75 THEN

  Sorry for not being able to insert PRT code from My smartphone.

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 16454

  Been thanked: 3636 times
  10/21/2017 at 9:57 PM #50162
  Vonasi

  Despair – No this was a live account – I gave up on demo account data a long time ago!

  RobertoGozzi – Line 8 is just about colour choice and nothing to do with the calculation.

  The screenshot shows the fault in the high low close and open values that lead to the bad calculation. There are several of these over the years. Makes coding a strategy a little difficult if the data stream is rubbish!

   

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  10/21/2017 at 10:47 PM #50170
  AVT

  just a quick idea: abs(high-low)

  

  AVT

  Participant

  Topics: 17

  Replies: 186

  Been thanked: 58 times
  10/21/2017 at 11:04 PM #50171
  Vonasi

  Thanks for the suggestion AVT but a high will always be higher than a low so the result will always be a positive from (high-low). The fault is in the data supplied from IG/PRT as if you check the numbers in the cursor details panel shown in the image the range is smaller than the body of the candle, which is impossible.

  1 user thanked author for this post.

  AVT

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  10/21/2017 at 11:07 PM #50172
  robertogozzi

  High-Low cannot be negative, so using ABS() would not affect the outcome.

  1 user thanked author for this post.

  AVT

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 16454

  Been thanked: 3636 times
  10/21/2017 at 11:12 PM #50174
  AVT

  Damned guys, you’re right !! – already too late in the evening

  

  AVT

  Participant

  Topics: 17

  Replies: 186

  Been thanked: 58 times
  10/21/2017 at 11:16 PM #50175
  JC_Bywan

  Are the wrong ones you found only sundays like the one in your attached graph? If yes, at least it would be contained to the usual sunday night data issue, rather than affecting normal week days… If no it might be worth graphing the day close in smaller timeframe and check for the instrument you look at where actual day close is relative to midnight, problem being with 24h dataflow that open and close might not be sctritly “on the same day” and generate weird stuff if high and low are on the same day…

  

  JC_Bywan

  Moderator

  Topics: 20

  Replies: 2542

  Been thanked: 836 times
  10/22/2017 at 8:23 AM #50183
  Vonasi

  No they are not all Sundays. Screenshot attached of a particularly bad period in 2003. Not sure how that data could be used for making decisions on when to trade if you can’t trust the basic data of high, low and open,close to be correct.

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  10/22/2017 at 9:39 AM #50194
  Despair

  I think you should report this to PRT. If you write a ticket directly via the platform you can even select data-errors.

  

  Despair

  Blocked

  Topics: 97

  Replies: 1034

  Been thanked: 217 times
  10/22/2017 at 1:35 PM #50210
  Vonasi

  Report sent.

  1 user thanked author for this post.

  Despair

  

  Vonasi

  Moderator

  Topics: 192

  Replies: 5088

  Been thanked: 1810 times
  10/22/2017 at 7:51 PM #50254
  robertogozzi

  I run it on DAX 1-hour TF and everything worked fine, but not as fine when I run it on a Daily and 4-hour TF, so I think the problem is the one Noobywan talked about, that is candles across the end of the day.

  

  robertogozzi

  Moderator

  Topics: 69

  Replies: 16454

  Been thanked: 3636 times
• Author
  Posts