CONVERSION INDICADOR TRADINGVIEW:Árbol de Decisión

This topic has 4 replies, 2 voices, and was last updated 1 month ago by Iván.

Viewing 5 posts - 1 through 5 (of 5 total)

08/06/2024 at 12:51 PM #236167

Fr7

¿Qué es un Árbol de Decisión?

Un árbol de decisión es una herramienta de modelado predictivo que se utiliza en estadísticas, minería de datos y aprendizaje automático. Se llama “árbol” porque su estructura se asemeja a un árbol con ramas. Aquí tienes una explicación básica:

Nodos: Cada nodo del árbol representa una pregunta o condición sobre los datos.
Ramas: Cada rama representa una posible respuesta o resultado de la pregunta.
Hojas: Las hojas son los nodos finales que representan la decisión o clasificación final.

¿Cómo Funciona?

Inicio: Comienza en el nodo raíz (la primera pregunta).
Decisiones: En cada nodo, se hace una pregunta y se sigue la rama correspondiente según la respuesta.
Resultado: Se continúa hasta llegar a una hoja, que da la decisión final.

¿Qué Hace el Código Anterior?

El código anterior crea un indicador en TradingView que utiliza un árbol de decisión para generar señales de compra y venta basadas en varios indicadores técnicos (RSI, MACD, medias móviles) y los caminos de decisión generados por la librería FunctionDecisionTree.

Desglose del Código:

Importar Librería: Importa la librería FunctionDecisionTree que contiene la función para generar caminos de decisión.
Parámetros de Entrada: Define los parámetros de entrada como la profundidad del árbol, el número de caminos y el número de pesos.
Inicialización de Pesos: Crea un array de pesos que se utilizarán en el árbol de decisión.
Generación de Caminos de Decisión: Utiliza la función decision_tree para generar caminos de decisión basados en los pesos y la profundidad especificada.
Indicadores Técnicos: Calcula varios indicadores técnicos como las medias móviles, RSI y MACD.
Generar Señales: Genera señales de compra y venta basadas en los caminos de decisión y los indicadores técnicos.
Plotear Señales: Muestra las señales de compra y venta en el gráfico con flechas hacia arriba (BUY) y hacia abajo (SELL).

Fr7

Participant

Topics: 47

Replies: 150

Been thanked: 17 times

08/06/2024 at 12:53 PM #236168

Fr7

// © RicardoSantos

//@version=5

// @description Method to generate decision tree based on weights.
library(title='FunctionDecisionTree')

// @function Method to generate decision tree based on weights.
// @param weights float array, weights for decision consideration.
// @param depth int, depth of the tree.
// @returns int array
export decision_tree(float[] weights, int depth) => //{
    int _size_w = array.size(id=weights)
    //
    int[] _path = array.new_int(size=0, initial_value=0)
    int _sumweights = math.ceil(array.sum(id=weights))
    if _size_w > 0 and depth > 0 and _sumweights == 1
        for _d = 1 to depth by 1
            for _w = 0 to 999 by 1
                int _select_weight_index = int(math.random(max=_size_w))
                float _rng = math.random(max=1.0)
                float _weight = array.get(id=weights, index=_select_weight_index)
                if _weight >= _rng
                    array.push(id=_path, value=_select_weight_index)
                    break
    _path
//{ usage:
// Input Parameters:
int depth = input.int(defval=1, minval=1)
int paths = input.int(defval=2, minval=1)
int number_of_weights = input.int(defval=2, minval=1)
// Array to contain weights, sum of array values must be 1.
var float[] node_weights = array.new_float(size=number_of_weights, initial_value=0)
if barstate.isfirst
    float _weight = 1.0 / number_of_weights
    for _i=1 to number_of_weights
        array.set(node_weights, _i-1, _weight)
    string t = ''
    for _p = 1 to paths by 1
        _path = decision_tree(node_weights, depth)
        t := t + '\nPath ' + str.tostring(_p, '#') + ': ' + str.tostring(_path)
        t
    var label la = label.new(bar_index, 0.0, t)
//{ remarks:
//}}}

// © RicardoSantos

//@version=5

// @description Method to generate decision tree based on weights.

library(title='FunctionDecisionTree')

// @function Method to generate decision tree based on weights.

// @param weights float array, weights for decision consideration.

// @param depth int, depth of the tree.

// @returns int array

export decision_tree(float[] weights, int depth) => //{

int _size_w = array.size(id=weights)

int[] _path = array.new_int(size=0, initial_value=0)

int _sumweights = math.ceil(array.sum(id=weights))

if _size_w > 0 and depth > 0 and _sumweights == 1

for _d = 1 to depth by 1

for _w = 0 to 999 by 1

int _select_weight_index = int(math.random(max=_size_w))

float _rng = math.random(max=1.0)

float _weight = array.get(id=weights, index=_select_weight_index)

if _weight >= _rng

array.push(id=_path, value=_select_weight_index)

break

_path

//{ usage:

// Input Parameters:

int depth = input.int(defval=1, minval=1)

int paths = input.int(defval=2, minval=1)

int number_of_weights = input.int(defval=2, minval=1)

// Array to contain weights, sum of array values must be 1.

var float[] node_weights = array.new_float(size=number_of_weights, initial_value=0)

if barstate.isfirst

float _weight = 1.0 / number_of_weights

for _i=1 to number_of_weights

array.set(node_weights, _i-1, _weight)

string t = ''

for _p = 1 to paths by 1

_path = decision_tree(node_weights, depth)

t := t + '\nPath ' + str.tostring(_p, '#') + ': ' + str.tostring(_path)

var label la = label.new(bar_index, 0.0, t)

//{ remarks:

//}}}

//@version=5
indicator("Decision Tree Indicator with Technical Indicators and Signals", overlay=true)

// Importar la librería
import RicardoSantos/FunctionDecisionTree/1 as dt

// Parámetros de entrada
depth = input.int(defval=3, minval=1, title="Depth of the Tree")
paths = input.int(defval=2, minval=1, title="Number of Paths")
number_of_weights = input.int(defval=2, minval=1, title="Number of Weights")

// Array para contener los pesos, la suma de los valores del array debe ser 1.
var float[] node_weights = array.new_float(size=number_of_weights, initial_value=0)
if barstate.isfirst
    float _weight = 1.0 / number_of_weights
    for _i=0 to number_of_weights - 1
        array.set(node_weights, _i, _weight)

// Generar caminos de decisión
string t = ''
var int[] decision_paths = na
if barstate.islast
    decision_paths := array.new_int(0)
    for _p = 1 to paths by 1
        _path = dt.decision_tree(node_weights, depth)
        for val in _path
            array.push(decision_paths, val)
        t := t + '\nPath ' + str.tostring(_p, '#') + ': ' + str.tostring(_path)

// Mostrar los caminos en el gráfico
var label la = na
if barstate.islast
    la := label.new(bar_index, na, t, color=color.blue, textcolor=color.white, style=label.style_label_down, size=size.small)
    label.set_xy(la, bar_index, high)

// Indicadores técnicos
sma_short = ta.sma(close, 14)
sma_long = ta.sma(close, 50)
rsi = ta.rsi(close, 14)
[macdLine, signalLine, _] = ta.macd(close, 12, 26, 9)

// Plotear indicadores técnicos
plot(sma_short, color=color.green, title="SMA 14")
plot(sma_long, color=color.red, title="SMA 50")
hline(70, "RSI Overbought", color=color.red)
hline(30, "RSI Oversold", color=color.green)
plot(rsi, color=color.blue, title="RSI")
plot(macdLine, color=color.orange, title="MACD Line")
plot(signalLine, color=color.purple, title="Signal Line")

// Generar señales de compra y venta basadas en indicadores técnicos y caminos de decisión
buy_signal = (not na(decision_paths)) and (array.size(decision_paths) > 0) and (array.get(decision_paths, 0) == 1) and (rsi < 30) and (macdLine > signalLine) and (sma_short > sma_long)
sell_signal = (not na(decision_paths)) and (array.size(decision_paths) > 0) and (array.get(decision_paths, 0) == 0) and (rsi > 70) and (macdLine < signalLine) and (sma_short < sma_long)

// Plotear señales de compra y venta
plotshape(series=buy_signal, location=location.belowbar, color=color.green, style=shape.labelup, title="Buy Signal", text="BUY")
plotshape(series=sell_signal, location=location.abovebar, color=color.red, style=shape.labeldown, title="Sell Signal", text="SELL")

//@version=5

indicator("Decision Tree Indicator with Technical Indicators and Signals", overlay=true)

// Importar la librería

import RicardoSantos/FunctionDecisionTree/1 as dt

// Parámetros de entrada

depth = input.int(defval=3, minval=1, title="Depth of the Tree")

paths = input.int(defval=2, minval=1, title="Number of Paths")

number_of_weights = input.int(defval=2, minval=1, title="Number of Weights")

// Array para contener los pesos, la suma de los valores del array debe ser 1.

var float[] node_weights = array.new_float(size=number_of_weights, initial_value=0)

if barstate.isfirst

float _weight = 1.0 / number_of_weights

for _i=0 to number_of_weights - 1

array.set(node_weights, _i, _weight)

// Generar caminos de decisión

string t = ''

var int[] decision_paths = na

if barstate.islast

decision_paths := array.new_int(0)

for _p = 1 to paths by 1

_path = dt.decision_tree(node_weights, depth)

for val in _path

array.push(decision_paths, val)

t := t + '\nPath ' + str.tostring(_p, '#') + ': ' + str.tostring(_path)

// Mostrar los caminos en el gráfico

var label la = na

if barstate.islast

la := label.new(bar_index, na, t, color=color.blue, textcolor=color.white, style=label.style_label_down, size=size.small)

label.set_xy(la, bar_index, high)

// Indicadores técnicos

sma_short = ta.sma(close, 14)

sma_long = ta.sma(close, 50)

rsi = ta.rsi(close, 14)

[macdLine, signalLine, _] = ta.macd(close, 12, 26, 9)

// Plotear indicadores técnicos

plot(sma_short, color=color.green, title="SMA 14")

plot(sma_long, color=color.red, title="SMA 50")

hline(70, "RSI Overbought", color=color.red)

hline(30, "RSI Oversold", color=color.green)

plot(rsi, color=color.blue, title="RSI")

plot(macdLine, color=color.orange, title="MACD Line")

plot(signalLine, color=color.purple, title="Signal Line")

// Generar señales de compra y venta basadas en indicadores técnicos y caminos de decisión

buy_signal = (not na(decision_paths)) and (array.size(decision_paths) > 0) and (array.get(decision_paths, 0) == 1) and (rsi < 30) and (macdLine > signalLine) and (sma_short > sma_long)

sell_signal = (not na(decision_paths)) and (array.size(decision_paths) > 0) and (array.get(decision_paths, 0) == 0) and (rsi > 70) and (macdLine < signalLine) and (sma_short < sma_long)

// Plotear señales de compra y venta

plotshape(series=buy_signal, location=location.belowbar, color=color.green, style=shape.labelup, title="Buy Signal", text="BUY")

plotshape(series=sell_signal, location=location.abovebar, color=color.red, style=shape.labeldown, title="Sell Signal", text="SELL")

Fr7

Participant

Topics: 47

Replies: 150

Been thanked: 17 times

08/06/2024 at 12:57 PM #236169

Fr7

Iván, puedes pasar la librería y el indicador a PRT?

GRACIAS

Fr7

Participant

Topics: 47

Replies: 150

Been thanked: 17 times

08/06/2024 at 10:24 PM #236186

Fr7

// Function Decision Tree Indicator with Technical Indicators and Signals
// Input Parameters
DepthOfTree = 3
NumberOfPaths = 2
NumberOfWeights = 2

// Array to contain weights, sum of array values must be 1.
NodeWeights = []
for i = 1 to NumberOfWeights do
    NodeWeights[i] = 1.0 / NumberOfWeights
next

// Function to generate decision tree based on weights
decisionTree = []
sizeW = 0
sumWeights = 0

if arraySize(NodeWeights) > 0 and DepthOfTree > 0 then
    sizeW = arraySize(NodeWeights)
    sumWeights = round(arraySum(NodeWeights))
    if sumWeights == 1 then
        for d = 1 to DepthOfTree do
            for w = 0 to 999 do
                selectWeightIndex = round(random * (sizeW - 1)) + 1
                rng = random
                weight = NodeWeights[selectWeightIndex]
                if weight >= rng then
                    decisionTree[size(decisionTree)+1] = selectWeightIndex
                    break
                endif
            next
        next
    endif
endif

// Generate decision paths
DecisionPaths = []
pathStr = ""
for p = 1 to NumberOfPaths do
    path = decisionTree
    for val = 1 to size(path) do
        DecisionPaths[size(DecisionPaths)+1] = path[val]
    next
    pathStr = pathStr + "\nPath " + p + ": " + arrayToString(path)
next

// Technical Indicators
SmaShort = Average[14](Close)
SmaLong = Average[50](Close)
Rsi = RSI[14](Close)
MacdLine = MACDLine[12, 26, 9](Close)
SignalLine = Signal[12, 26, 9](Close)

// Generate Buy and Sell signals based on technical indicators and decision paths
BuySignal = 0
SellSignal = 0
if size(DecisionPaths) > 0 then
    if DecisionPaths[1] == 1 and Rsi < 30 and MacdLine > SignalLine and SmaShort > SmaLong then
        BuySignal = 1
    endif
    if DecisionPaths[1] == 0 and Rsi > 70 and MacdLine < SignalLine and SmaShort < SmaLong then
        SellSignal = 1
    endif
endif

// Plot Buy and Sell signals
if BuySignal then
    DRAWARROWUP(BarIndex, Low, "BUY", "", Plain, Green)
endif
if SellSignal then
    DRAWARROWDOWN(BarIndex, High, "SELL", "", Plain, Red)
endif
// Plot Technical Indicators 
RETURN SmaShort AS "SMA 14", SmaLong AS "SMA 50"

// Function Decision Tree Indicator with Technical Indicators and Signals

// Input Parameters

DepthOfTree = 3

NumberOfPaths = 2

NumberOfWeights = 2

// Array to contain weights, sum of array values must be 1.

NodeWeights = []

for i = 1 to NumberOfWeights do

NodeWeights[i] = 1.0 / NumberOfWeights

// Function to generate decision tree based on weights

decisionTree = []

sizeW = 0

sumWeights = 0

if arraySize(NodeWeights) > 0 and DepthOfTree > 0 then

sizeW = arraySize(NodeWeights)

sumWeights = round(arraySum(NodeWeights))

if sumWeights == 1 then

for d = 1 to DepthOfTree do

for w = 0 to 999 do

selectWeightIndex = round(random * (sizeW - 1)) + 1

rng = random

weight = NodeWeights[selectWeightIndex]

if weight >= rng then

decisionTree[size(decisionTree)+1] = selectWeightIndex

break

endif

// Generate decision paths

DecisionPaths = []

pathStr = ""

for p = 1 to NumberOfPaths do

path = decisionTree

for val = 1 to size(path) do

DecisionPaths[size(DecisionPaths)+1] = path[val]

pathStr = pathStr + "\nPath " + p + ": " + arrayToString(path)

// Technical Indicators

SmaShort = Average[14](Close)

SmaLong = Average[50](Close)

Rsi = RSI[14](Close)

MacdLine = MACDLine[12, 26, 9](Close)

SignalLine = Signal[12, 26, 9](Close)

// Generate Buy and Sell signals based on technical indicators and decision paths

BuySignal = 0

SellSignal = 0

if size(DecisionPaths) > 0 then

if DecisionPaths[1] == 1 and Rsi < 30 and MacdLine > SignalLine and SmaShort > SmaLong then

BuySignal = 1

endif

if DecisionPaths[1] == 0 and Rsi > 70 and MacdLine < SignalLine and SmaShort < SmaLong then

SellSignal = 1

endif

// Plot Buy and Sell signals

if BuySignal then

DRAWARROWUP(BarIndex, Low, "BUY", "", Plain, Green)

endif

if SellSignal then

DRAWARROWDOWN(BarIndex, High, "SELL", "", Plain, Red)

endif

// Plot Technical Indicators

RETURN SmaShort AS "SMA 14", SmaLong AS "SMA 50"

He intentado la traducción pero aún contiene varios errores…Espero que los foreros puedan corregirlo y mejorarlo.

Fr7

Participant

Topics: 47

Replies: 150

Been thanked: 17 times

08/08/2024 at 12:01 PM #236263

Iván

Sería una cosa así:

// Inicialización de los parámetros de entrada
Depth = 3
Paths = 3
NumberOfWeights = 5
// Inicialización del array de pesos
Weight = 1.0 / NumberOfWeights
SumNodeWeights = 0
FOR i = 0 TO NumberOfWeights - 1 DO
$NodeWeights[i] = Weight
SumNodeWeights = SumNodeWeights + Weight
NEXT
// Generación de los caminos
FOR p = 0 TO Paths - 1 DO
SizeW = NumberOfWeights
SumWeights = CEIL(SumNodeWeights)
    
IF SizeW > 0 AND Depth > 0 AND SumWeights = 1 THEN
FOR d = 1 TO Depth DO
FOUND = 0
FOR w = 0 TO 999 DO
SelectWeightIndex = ROUND(Random (0,SizeW))
Rng = Random(0,1)
Weight = $NodeWeights[SelectWeightIndex]
IF Weight >= Rng THEN
$Paths[p] = SelectWeightIndex
FOUND = 1
BREAK
ENDIF
NEXT
IF FOUND = 0 THEN

ENDIF
NEXT
ENDIF
NEXT
//Mostrar el resultado de path de la última vela en el gráfico
if islastbarupdate then
for i=0 to lastset($paths) do
path=$paths[i]
drawtext("path#i#: #path#",barindex+10,low-i*averagetruerange[14](close))
next
endif
// Indicadores técnicos
SMAShort = Average[14](Close)
SMALong = Average[50](Close)
myMACDLine = MACDLine[12,26,9](Close)
SignalLine = MACDSignal[12,26,9](Close)
// Generar señales de compra y venta basadas en indicadores técnicos y caminos de decisión
BuySignal = ($Paths[0]>0 AND (myMACDLine > SignalLine) AND (SMAShort > SMALong))
SellSignal = ($Paths[0]=0 AND (myMACDLine < SignalLine) AND (SMAShort < SMALong))

// Plotear señales de compra y venta
IF BuySignal THEN
DrawArrowUp(BarIndex, Low)
elsiF SellSignal THEN
DrawArrowDown(BarIndex, High)
ENDIF

return

// Inicialización de los parámetros de entrada

Depth = 3

Paths = 3

NumberOfWeights = 5

// Inicialización del array de pesos

Weight = 1.0 / NumberOfWeights

SumNodeWeights = 0

FOR i = 0 TO NumberOfWeights - 1 DO

$NodeWeights[i] = Weight

SumNodeWeights = SumNodeWeights + Weight

// Generación de los caminos

FOR p = 0 TO Paths - 1 DO

SizeW = NumberOfWeights

SumWeights = CEIL(SumNodeWeights)

IF SizeW > 0 AND Depth > 0 AND SumWeights = 1 THEN

FOR d = 1 TO Depth DO

FOUND = 0

FOR w = 0 TO 999 DO

SelectWeightIndex = ROUND(Random (0,SizeW))

Rng = Random(0,1)

Weight = $NodeWeights[SelectWeightIndex]

IF Weight >= Rng THEN

$Paths[p] = SelectWeightIndex

FOUND = 1

BREAK

ENDIF

IF FOUND = 0 THEN

ENDIF

//Mostrar el resultado de path de la última vela en el gráfico

if islastbarupdate then

for i=0 to lastset($paths) do

path=$paths[i]

drawtext("path#i#: #path#",barindex+10,low-i*averagetruerange[14](close))

endif

// Indicadores técnicos

SMAShort = Average[14](Close)

SMALong = Average[50](Close)

myMACDLine = MACDLine[12,26,9](Close)

SignalLine = MACDSignal[12,26,9](Close)

// Generar señales de compra y venta basadas en indicadores técnicos y caminos de decisión

BuySignal = ($Paths[0]>0 AND (myMACDLine > SignalLine) AND (SMAShort > SMALong))

SellSignal = ($Paths[0]=0 AND (myMACDLine < SignalLine) AND (SMAShort < SMALong))

// Plotear señales de compra y venta

IF BuySignal THEN

DrawArrowUp(BarIndex, Low)

elsiF SellSignal THEN

DrawArrowDown(BarIndex, High)

ENDIF

return

1 user thanked author for this post.

Fr7

Iván

Moderator

Topics: 0

Replies: 797

Been thanked: 271 times

Author

Posts

Viewing 5 posts - 1 through 5 (of 5 total)

Find exclusive trading pro-tools on