How to determine if consecutive fractals form an horizontal level?

New comment
 

I'm trying to build an indicator using multiple fractals in order to spot support and resistance levels. The thing is I want multiple consecutive fractals to be close to each other in order to form a valid level, like if we draw a line linking them; it needs to appear nearly horizontal.

But "close to each other" change a lot depending on the time frame; 0.5% difference can be considered close on the daily but it would be way too big for the M5 chart, you see the problem?

What can I use which would work across all time frames?

 
Loïc B.:

I'm trying to build an indicator using multiple fractals in order to spot support and resistance levels. The thing is I want multiple consecutive fractals to be close to each other in order to form a valid level, like if we draw a line linking them; it needs to appear nearly horizontal.

But "close to each other" change a lot depending on the time frame; 0.5% difference can be considered close on the daily but it would be way too big for the M5 chart, you see the problem?

What can I use which would work across all time frames?

For such dynamic values, the ATR is a very good value to use. It adapts automatically to the symbol and timeframe.

The default period of 14 is a good compromise of fast enough and not spikey. Use a factor as input parameter to make this value adjustable.

Something like ATR * 0.8 to give you 80% of the ATR value.
 
Totally agree with Dominik. ATR is the solution if you want a dynamic and auto adjusting reference.
 
Dominik Egert #:
For such dynamic values, the ATR is a very good value to use. It adapts automatically to the symbol and timeframe.

The default period of 14 is a good compromise of fast enough and not spikey. Use a factor as input parameter to make this value adjustable.

Something like ATR * 0.8 to give you 80% of the ATR value.
That's a good idea, is ATR better than StdDev or fractals clustering technique for defining accurate levels though?
 
Loïc B. #:
That's a good idea, is ATR better than StdDev or fractals clustering technique for defining accurate levels though?
Mathematically, it depends. ATR is an average, with stddev you can "squash" non-stationary data into stationary range on top of using an average.

Both approaches are valid, since ATR is a somewhat more direct correlated value to your chart, it is probably easier to use than stddev.
 
I don't think consecutive fractals are going to be close to each other, and if they are, they're not decent fractals...it makes more sense to draw horizontal lines on timeframe candle extremums or highest/lowest price of the day
 

If you like the idea of support and resistance with ATR like what was discussed here, you can check this indicator

https://www.mql5.com/en/code/47048


I changed the colours personally as I hate colorful shit. But the indicator itself...very very good, tried and tested

Rainbow indicator for mt5
Rainbow indicator for mt5
  • www.mql5.com
This is an indicator with a rainbow-like display, both for trend or pullback prediction.
 
Conor Mcnamara #:
I don't think consecutive fractals are going to be close to each other, and if they are, they're not decent fractals...it makes more sense to draw horizontal lines on timeframe candle extremums or highest/lowest price of the day

What do you mean? See example below

fractals ATR resistance example

The highest is 75.53 while the lowest is 75.50 so max 0.03 difference.

Hence, the last 6 up fractals (from red line) are within 1/3*ATR from each other.

 
Loïc B. #:

What do you mean? See example below

The highest is 75.53 while the lowest is 75.50 so max 0.03 difference.

Hence, the last 6 up fractals (from red line) are within 1/3*ATR from each other.

I mean too close to each other with time (not price). Because too many fractals isn't that good. A strong reliable fractal is better, for example, a price where a strong price divergence is more likely to occur (or is occurring)
 
Conor Mcnamara #:
I mean too close to each other with time (not price). Because too many fractals isn't that good. A strong reliable fractal is better, for example, a price where a strong price divergence is more likely to occur (or is occurring)

It depends on your objective.

A lot of consecutive aligned fractals in the last 15 bars or so often shows that the price is about to break it.

New comment