Author Message:
General Description
The Nadaraya-Watson Oscillator (NWO) will give the same information as the Nadaraya-Watson Envelope , but as an oscillator off the main chart, by plotting the relationship between price and the Kernel and its bands. This also means that we can now detect divergences between price and the NWO.
You can think of this indicator as the kernel envelope version of a Bollinger Band %B . In ranging markets the bands are perfect for mean reversion trades, but in certain situations the break of one of the bands can signal the begining of a strong trend and price will remain close to the bands for a long period and will only give small pullbacks. As with any indicator, confluence with price and other tools must be taken into account.
CODE: Update - Added MTF option
CSS:
#// This source code is subject to the terms of the Mozilla Public License 2.0 at https://mozilla.org/MPL/2.0/
#// ©FractalTrade15, original Nadaraya_Watson Kernel code and library by ©jdehorty
#// @FractalTrade15
#indicator(title="Nadaraya-Watson Oscillator", shorttitle = "NWO" , overlay=false, timeframe='')
# converted by Sam4Cok@Samer800 - 04/2023
# Update: Added MTF option by Sam4Cok@Samer800 - 11/2023
declare lower;
#// Kernel Settings
input ColorBars = yes;
input selectTimeframe = {Default "Chart", "Custom"};
input customTimeframe = AggregationPeriod.FIFTEEN_MIN;
input LookbackWindow = 8; # 'Lookback Window'. Recommended range: 3-50'
input alpha = 8.0; # 'Relative Weighting'. Recommended range: 0.25-25'
input RegressionLength = 25;# "Start Regression at Bar". Recommended range: 5-25'
input atr_length = 21; # 'ATR Length'
input nearFactor = 2.25; # 'Near ATR Factor'. Recommended range: 0.5-2.0'
input farFactor = 3.0; # 'Far ATR Factor'. Recommended range: 6.0-8.0'
input mid_width = 0.5; # "Middle Band Width"
input hull_on = yes; # "Hull NW MA"
input hull_length = 55; # "Hull Smoothing Length"
def na = Double.NaN;
def last = IsNaN(close);
def chart = selectTimeframe==selectTimeframe."Chart";
def pos = Double.POSITIVE_INFINITY;
def neg = Double.NEGATIVE_INFINITY;
def x_0 = RegressionLength;
def h = LookbackWindow;
def tfC = close;
def tfH = high;
def tfL = low;
def mtfC = close(Period = customTimeframe);
def mtfH = high(Period = customTimeframe);
def mtfL = low(Period = customTimeframe);
def cl = if chart then tfC else mtfC;
def hi = if chart then tfH else mtfH;
def lo = if chart then tfL else mtfL;
#-- Colors
DefineGlobalColor("bullc" , CreateColor(76, 175, 80));
DefineGlobalColor("bearc" , CreateColor(255, 82, 82));
#-- Functions
#getBounds(_atr, _nearFactor, _farFactor, _yhat) =>
script getBounds {
input _atr = 0;
input _nearFactor = 0;
input _farFactor = 0;
input _yhat = close;
def _upper_far = _yhat + _farFactor * _atr;
def _upper_near = _yhat + _nearFactor * _atr;
def _lower_near = _yhat - _nearFactor * _atr;
def _lower_far = _yhat - _farFactor * _atr;
plot up_near = _upper_near;
plot up_far = _upper_far;
plot lo_near = _lower_near;
plot lo_far = _lower_far;
}
#kernel_atr(length, _high, _low, _close) =>
script kernel_atr {
input length = 10;
input _high = high;
input _low = low;
input _close = close;
def trueRange = TrueRange(_high,_close,_low);
def kernel_atr = WildersAverage(trueRange, length);
plot out = kernel_atr;
}
#rationalQuadratic(float _src, _lookback, _relativeWeight, startAtBar) =>
script rationalQuadratic {
input _src = close;
input _lookback = 8;
input _relativeWeight = 8;
input startAtBar = 25;
def _currentWeight;# = 0.
def _cumulativeWeight;# = 0.
def size = AbsValue(BarNumber());
def _size = if size > 2000 then 2000 else size;
_currentWeight = fold i = 0 to _size + startAtBar with p do
p + GetValue(_src, i) * (Power(1 + (Power(i, 2) / ((Power(_lookback, 2) * 2 * _relativeWeight))), -_relativeWeight));
_cumulativeWeight = fold i1 = 0 to _size + startAtBar with p1 do
p1 + (Power(1 + (Power(i1, 2) / ((Power(_lookback, 2) * 2 * _relativeWeight))), -_relativeWeight));
def yhat = _currentWeight / _cumulativeWeight;
plot out = yhat;
}
#f_c_gradientAdvDecPro(_source, _center, _steps, _c_bearWeak, _c_bearStrong, _c_bullWeak, _c_bullStrong) =>
script f_c_gradientAdvDecPro {
input _source = close;
input _center = close;
input _steps = 5;
def _qtyAdvDec;
def _return;
def _maxSteps = Max(1, _steps);
def _xUp = (_source > _center and _source[1] <= _center[1]);
def _xDn = (_source < _center and _source[1] >= _center[1]);
def _chg = _source - _source[1];
def _up = _chg > 0;
def _dn = _chg < 0;
def _srcBull = _source > _center;
def _srcBear = _source < _center;
_qtyAdvDec = if _srcBull then if _xUp then 1 else
if _up then Min(_maxSteps, _qtyAdvDec[1] + 1) else
if _dn then Max(1, _qtyAdvDec[1] - 1) else _qtyAdvDec[1] else
if _srcBear then if _xDn then 1 else
if _dn then Min(_maxSteps, _qtyAdvDec[1] + 1) else
if _up then Max(1, _qtyAdvDec[1] - 1) else _qtyAdvDec[1] else _qtyAdvDec[1];
_return = if _srcBull then _qtyAdvDec else if _srcBear then -_qtyAdvDec else _return[1];
plot out = _return;
}
#//Calculations
def yhat_close = rationalQuadratic(cl, h, alpha, x_0);
def yhat_high = rationalQuadratic(hi, h, alpha, x_0);
def yhat_low = rationalQuadratic(lo, h, alpha, x_0);
def yhat = yhat_close;
def ktr = kernel_atr(atr_length, yhat_high, yhat_low, yhat_close);
def upper_near = getBounds(ktr, nearFactor, farFactor, yhat_close).up_near;
def upper_far = getBounds(ktr, nearFactor, farFactor, yhat_close).up_far;
def lower_near = getBounds(ktr, nearFactor, farFactor, yhat_close).lo_near;
def lower_far = getBounds(ktr, nearFactor, farFactor, yhat_close).lo_far;
def avg_factor = Average(nearFactor, farFactor);
def z_nad = (cl - yhat) / ktr;
def hullma = HullMovingAvg(z_nad, hull_length);
def emaYhat = ExpAverage(yhat, 2);
def grad2 = f_c_gradientAdvDecPro(yhat_close, emaYhat, 5);
#// Plots
plot p_zn = z_nad; # 'Nadaraya-Watson Estimator'
plot p_hull = if hull_on then hullma else na;
def p_upper_far = if last then na else farFactor; # 'Upper Boundary: Far'
def p_upper_near = if last then na else nearFactor; # 'Upper Boundary: Near'
plot p_yhath = if last then na else mid_width; # 'Nadaraya-Watson Estimation'
plot p_yhat = if last then na else 0; # 'Nadaraya-Watson Estimation'
plot p_yhatl = if last then na else -mid_width; # 'Nadaraya-Watson Estimation'
def p_lower_near = if last then na else -nearFactor; # 'Lower Boundary: Near'
def p_lower_far = if last then na else -farFactor; # 'Lower Boundary: Far'
p_zn.SetLineWeight(2);
p_zn.AssignValueColor(if z_nad>p_yhath then Color.GREEN else
if z_nad<p_yhatl then Color.RED else Color.GRAY);
p_hull.AssignValueColor(if hullma > hullma[2] then GlobalColor("bullc") else GlobalColor("bearc"));
p_yhat.AssignValueColor(if grad2>0 then Color.DARK_GREEN else
if grad2<0 then Color.DARK_RED else Color.GRAY);
p_yhath.SetDefaultColor(Color.DARK_GREEN);
p_yhatl.SetDefaultColor(Color.DARK_RED);
p_yhath.SetPaintingStrategy(PaintingStrategy.DASHES);
p_yhatl.SetPaintingStrategy(PaintingStrategy.DASHES);
AddCloud(if z_nad>p_yhath then z_nad else na, p_yhath, Color.DARK_GREEN);
AddCloud(if z_nad<p_yhatl then p_yhatl else na, z_nad, Color.DARK_RED);
AddCloud(p_upper_far, p_upper_near, Color.DARK_RED);
AddCloud(p_lower_near, p_lower_far, Color.DARK_GREEN);
AddCloud(if z_nad < -farFactor then pos else na, neg, Color.DARK_GREEN);
AddChart(high = if z_nad < -farFactor then pos else na , low = neg , open = pos, close = neg,
type = ChartType.CANDLE, growcolor = CreateColor(0, 36, 0));
AddChart(high = if z_nad > farFactor then pos else na , low = neg , open = pos, close = neg,
type = ChartType.CANDLE, growcolor = CreateColor(36, 0, 0));
AssignPriceColor(if !ColorBars then Color.CURRENT else
if z_nad>p_yhath then Color.GREEN else
if z_nad<p_yhatl then Color.RED else Color.GRAY);
#--- END of CODE
Last edited:
