Re: CFB (Composite Fractal Behavior) indicators for MT4

51
太虚一毫 wrote: Mon Sep 13, 2021 2:45 am right. But it is hoped that the histo version will reach the high quality and high level of the cfb adaptive qqe histo alerts + arrows indicator. cfb adaptive qqe histo alerts + arrows.mq4, in my opinion, may be the highest accuracy indicator of the forum in 2021.

Thank you teacher!

#property indicator_minimum 0
#property indicator_maximum 1

The advantage of this type of indicator is that it is easy to adjust the layout of the MT4 computer display screen.
Something like this?
These users thanked the author mrtools for the post (total 2):
太虚一毫, Jedidiah




Re: CFB (Composite Fractal Behavior) indicators for MT4

58
太虚一毫 wrote: Wed Jun 01, 2022 11:11 am This is an example image putting together cfb adaptive qqe histo, Adxm averages vhf adaptive. cool! :thumbup:

If teachers put them in a metric in coding, it should reduce computer resource consumption right? Cooler!
Image

Image

So far not able to join the 2 but think I was able to make the cfb adaptive qqe lighter and hopefully somewhat faster, with more ma smoothing choices for the rsi/qqe and think a couple more rsi type choices.
These users thanked the author mrtools for the post (total 7):
BeatlemaniaSA, kvak, 太虚一毫, thomdel, Jimmy, Jedidiah, Chickenspicy

Re: CFB (Composite Fractal Behavior) indicators for MT4

59
RodrigoRT7 wrote: Sun Jun 26, 2022 5:12 pm hello everyone, how are you? sorry for insisting on this topic, but this combination of averages really seems to be very promising.

I had already liked Cora Wave a lot and was researching some Jurik indicators on Trading View and I came across this one.

I would like to see how they would look on Renko, but even on candles it already looks interesting. if you can help me to convert at least the Cora Wave, I would be super grateful.

+ if you can kindly include Jurik Composite Fractal Behavior (CFB) on EMA [Loxx], that would be awesome.

thank you very much in advance :D

Code: Select all

// This source code is subject to the terms of the Mozilla Public License 2.0 at https://mozilla.org/MPL/2.0/
// © loxx

//@version=5
indicator("Jurik Composite Fractal Behavior (CFB) on EMA [Loxx]", shorttitle = "JCFBEMA [Loxx]", overlay = true, timeframe="", timeframe_gaps = true, max_bars_back = 2000)

greencolor = #2DD204
redcolor = #D2042D 

EMA(x, t) =>
    _ema = x
    _ema := na(_ema[1]) ? x : (x - nz(_ema[1])) * (2 / (t + 1)) + nz(_ema[1])
    _ema

_jcfbaux(src, depth)=>
    jrc04 = 0.0, jrc05 = 0.0, jrc06 = 0.0, jrc13 = 0.0, jrc03 = 0.0, jrc08 = 0.0
    if (bar_index >= bar_index - depth * 2) 
        for k = depth - 1 to 0
            jrc04 += math.abs(nz(src[k]) - nz(src[k+1]))
            jrc05 += (depth + k) * math.abs(nz(src[k]) - nz(src[k+1]))
            jrc06 += nz(src[k+1])
    if(bar_index < bar_index - depth * 2)
        jrc03 := math.abs(src - nz(src[1]))
        jrc13 := math.abs(nz(src[depth]) - nz(src[depth+1]))
        jrc04 := jrc04 - jrc13 + jrc03
        jrc05 := jrc05 - jrc04 + jrc03 * depth
        jrc06 := jrc06 - nz(src[depth+1]) + nz(src[1])
    jrc08 := math.abs(depth * src - jrc06)
    jcfbaux = jrc05 == 0.0 ? 0.0 : jrc08/jrc05
    jcfbaux

_jcfb(src, len, smth) =>
    a1 = array.new<float>(0, 0)
    c1 = array.new<float>(0, 0)
    d1 = array.new<float>(0, 0)
    cumsum = 2.0, result = 0.0

    //crete an array of 2^index, pushed onto itself
    //max values with injest of depth = 10: [1, 1, 2, 2, 4, 4, 8, 8, 16, 16, 32, 32, 64, 64, 128, 128, 256, 256, 512, 512]
    for i = 0 to len -1
        array.push(a1, math.pow(2, i))
        array.push(a1, math.pow(2, i))
        
    //cumulative sum of 2^index array
    //max values with injest of depth = 10: [2, 3, 4, 6, 8, 12, 16, 24, 32, 48, 64, 96, 128, 192, 256, 384, 512, 768, 1024, 1536]
    for i = 0 to array.size(a1) - 1 
        array.push(c1, cumsum)
        cumsum += array.get(a1, i)
        
    //add values from jcfbaux calculation using the cumumulative sum arrary above
    for i = 0 to array.size(c1) - 1 
        array.push(d1, _jcfbaux(src, array.get(c1, i)))
    
    baux = array.copy(d1)
    cnt = array.size(baux)
    arrE = array.new<float>(cnt, 0)
    arrX = array.new<float>(cnt, 0)
    
    if bar_index <= smth
        for j = 0 to bar_index - 1
            for k = 0 to cnt - 1
                eget = array.get(arrE, k)
                array.set(arrE, k, eget + nz(array.get(baux, k)[bar_index - j]))
        for k = 0 to cnt - 1
            eget = nz(array.get(arrE, k))
            array.set(arrE, k, eget / bar_index)
    else
        for k = 0 to cnt - 1
            eget = nz(array.get(arrE, k))
            array.set(arrE, k, eget + (nz(array.get(baux, k)) - nz(array.get(baux, k)[smth])) / smth)
    if bar_index > 5
        a = 1.0, b = 1.0
        for k = 0 to cnt - 1
            if (k % 2 == 0)
                array.set(arrX, k, b * nz(array.get(arrE, k)))
                b := b * (1 - nz(array.get(arrX, k)))
            else
                array.set(arrX, k, a * nz(array.get(arrE, k)))
                a := a * (1 - nz(array.get(arrX, k)))
        sq = 0.0, sqw = 0.0 
        for i = 0 to cnt - 1
            sqw += nz(array.get(arrX, i)) * nz(array.get(arrX, i)) * nz(array.get(c1, i))
        for i = 0 to array.size(arrX) - 1
            sq += math.pow(nz(array.get(arrX, i)), 2)
        result := sq == 0.0 ? 0.0 : sqw / sq
        result
            
_a_jurik_filt(src, len, phase) =>
    //static variales
    volty = 0.0, avolty = 0.0, vsum = 0.0, bsmax = src, bsmin = src
    len1 = math.max(math.log(math.sqrt(0.5 * (len-1))) / math.log(2.0) + 2.0, 0)
    len2 = math.sqrt(0.5 * (len - 1)) * len1
    pow1 = math.max(len1 - 2.0, 0.5)
    beta = 0.45 * (len - 1) / (0.45 * (len - 1) + 2)
    div = 1.0 / (10.0 + 10.0 * (math.min(math.max(len-10, 0), 100)) / 100)
    phaseRatio = phase < -100 ? 0.5 : phase > 100 ? 2.5 : 1.5 + phase * 0.01
    bet = len2 / (len2 + 1)

    //Price volatility
    del1 = src - nz(bsmax[1])
    del2 = src - nz(bsmin[1])
    volty := math.abs(del1) > math.abs(del2) ? math.abs(del1) : math.abs(del2)

    //Relative price volatility factor
    vsum := nz(vsum[1]) + div * (volty - nz(volty[10]))
    avolty := nz(avolty[1]) + (2.0 / (math.max(4.0 * len, 30) + 1.0)) * (vsum - nz(avolty[1]))
    dVolty = avolty > 0 ? volty / avolty : 0
    dVolty := math.max(1, math.min(math.pow(len1, 1.0/pow1), dVolty))
    
    //Jurik volatility bands
    pow2 = math.pow(dVolty, pow1)
    Kv = math.pow(bet, math.sqrt(pow2))
    bsmax := del1 > 0 ? src : src - Kv * del1
    bsmin := del2 < 0 ? src : src - Kv * del2
    
    //Jurik Dynamic Factor
    alpha = math.pow(beta, pow2)

    //1st stage - prelimimary smoothing by adaptive EMA
    jma = 0.0, ma1 = 0.0, det0 = 0.0, e2 = 0.0
    ma1 := (1 - alpha) * src + alpha * nz(ma1[1])
    
    //2nd stage - one more prelimimary smoothing by Kalman filter
    det0 := (src - ma1) * (1 - beta) + beta * nz(det0[1])
    ma2 = ma1 + phaseRatio * det0

    //3rd stage - final smoothing by unique Jurik adaptive filter
    e2 := (ma2 - nz(jma[1])) * math.pow(1 - alpha, 2) + math.pow(alpha, 2) * nz(e2[1])
    jma := e2 + nz(jma[1])
    jma


src = input.source(hlcc4, "Source", group = "Basic Settings")

cfb_src = input.source(hlcc4, "CFB Source", group = "CFB Ingest Settings")

// backsamping period for highs/lows
nlen = input.int(50, "CFB Normal Length", minval = 1, group = "CFB Ingest Settings")

// cfb depth, max 10 since that's around 5000 bars
cfb_len = input.int(10, "CFB Depth", maxval = 10, group = "CFB Ingest Settings") 

// internal cfb calc smoothing
smth = input.int(8, "CFB Smooth Length", minval = 1, group = "CFB Ingest Settings")

 // lower bound of samples returned from the nlen rolling window
slim = input.int(10, "CFB Short Limit", minval = 1, group = "CFB Ingest Settings")

// upper bound of samples returned from the nlen rolling window
llim = input.int(20, "CFB Long Limit", minval = 1, group = "CFB Ingest Settings") 

// for jurik filter post cfb calcuation smoothing length
jcfbsmlen = input.int(10, "CFB Jurik Smooth Length", minval = 1, group = "CFB Ingest Settings")

// for jurik filter post cfb calcuation smoothing phase, generall a number between -100 and 100
jcfbsmph = input.float(0, "CFB Jurik Smooth Phase", group = "CFB Ingest Settings")


colorbars = input.bool(true, title='Color bars?', group = "UI Options")


cfb_draft = _jcfb(cfb_src, cfb_len, smth)
cfb_pre = _a_jurik_filt(_a_jurik_filt(cfb_draft, jcfbsmlen, jcfbsmph), jcfbsmlen, jcfbsmph)

max = ta.highest(cfb_pre, nlen)
min = ta.lowest(cfb_pre, nlen)
denom = max - min
ratio = (denom > 0) ? (cfb_pre - min) / denom : 0.5
len_out_cfb = math.ceil(slim + ratio * (llim - slim))

emaout = EMA(src, int(len_out_cfb))  

plot(emaout, color = src >= emaout ? greencolor : redcolor, linewidth = 3)

barcolor(color = src >= emaout ? greencolor : redcolor)

Code: Select all

//@version=4
// This source code is subject to the terms of the Mozilla Public License 2.0 at https://mozilla.org/MPL/2.0/
// © RedKTrader

study("Comp_Ratio_MA", shorttitle = "CoRa Wave", overlay = true, resolution ="")

// ======================================================================    
// Compound Ratio Weight MA function
// Compound Ratio Weight is where the weight increases in a "logarithmicly linear" way (i.e., linear when plotted on a log chart) - similar to compound ratio
// the "step ratio" between weights is consistent - that's not the case with linear-weight moving average (WMA), or EMA 
// another advantage is we can significantly reduce the "tail weight" - which is "relatively" large in other MAs and contributes to lag
//
// Compound Weight ratio     r = (A/P)^1/t - 1
// Weight at time t         A = P(1 + r)^t 
//                            = Start_val * (1 + r) ^ index
// Note: index is 0 at the furthest point back -- num periods = length -1
//
f_adj_crwma(source, length, Start_Wt, r_multi) =>
    numerator = 0.0, denom = 0.0, c_weight = 0.0
    //Start_Wt = 1.0    // Start Wight is an input in this version - can also be set to a basic value here.
    End_Wt = length     // use length as initial End Weight to calculate base "r"
    r = pow((End_Wt / Start_Wt),(1 / (length - 1))) - 1
    base = 1 + r * r_multi
    for i = 0 to length -1
        c_weight    := Start_Wt * pow(base,(length - i))
        numerator   := numerator + source[i] * c_weight 
        denom       := denom + c_weight
    numerator / denom    
// ====================================================================== ==   

data        = input(title = "Source",                 type = input.source,      defval = hlc3)
length      = input(title = "length",                 type = input.integer,     defval = 20,  minval = 1)
r_multi     = input(title = "Comp Ratio Multiplier",  type = input.float,       defval = 2.0, minval = 0, step = .1)
smooth      = input(title = "Auto Smoothing",         type = input.bool,        defval = true,                    group = "Smoothing")
man_smooth  = input(title = "Manual Smoothing",       type = input.integer,     defval = 1, minval = 1, step = 1, group = "Smoothing")

s           = smooth ? max(round(sqrt(length)),1) : man_smooth
cora_raw    = f_adj_crwma(data, length, 0.01, r_multi)    
cora_wave   = wma(cora_raw, s)

c_up        = color.new(color.aqua, 0)
c_dn        = color.new(#FF9800   , 0)
cora_up     = cora_wave > cora_wave[1]
plot(cora_wave, title="Adjustible CoRa_Wave", color = cora_up ? c_up : c_dn, linewidth = 3)
Image

Image

Image
Think this would be the cfb adaptive ema.
These users thanked the author mrtools for the post (total 5):
Chickenspicy, kvak, RodrigoRT7, 太虚一毫, josi


Who is online

Users browsing this forum: Amazon [Bot], boytoy, bypy, Grapeshot [Bot], kvak, LittleCaro, Sogou [Bot], specialkey, TheJurgFX and 86 guests