Indicator that checks if the current candle high (as soon as a condition is met) breaks the previous days high (over a specific period ) or not

ayush-somani

I am making an Indicator that calculates how many previous days (period is dynamic i.e. 3, 4, 5, 10 and 20) high is taken out as soon as we met a condition on a bar. I used Highest (MaxN) to find the High in those specified periods (i.e. 3, 4, 5, 10 and 20) in next of my Indicator and stored the results in Lines. Then, I compared them in next of my Strategy with the bar's high (the one which satisfies the condition). But I am not getting the desired results even though on charts I can point out such data points. Please help !!
@run-out @dasch @backtrader @vladisld

ayush-somani

Please refer to this example

Screenshot 2020-09-25 at 3.10.30 PM.png

The 12th Feb bar has met the condition and now I need to check that it's has taken out high of some defined period, which in this case is from 15th Jan to 11th Feb. So, I need to count how many such bars are there in this.

ab_trader

@ayush-somani please don't spam. It was clear that you unable to code something from your very first post. Other posts don't add any value, just pollutes the community.

dasch

It would be great if you provide the code you tried, to see where you stuck. Also please don't open that many posts with the same question.

ayush-somani

@dasch This is my Code -

from datetime import datetime
import os.path
import sys
import backtrader as bt
import pandas as pd
import math

ticker_list = ['SBIN']

class DynamicHighest(bt.Indicator):
lines = ('periodline',)
params = dict(minperiod=3)

def __init__(self):
    self.addminperiod(self.p.minperiod)

def next(self):
    pass

class MyStrategy(bt.Strategy):
params = dict(bearish=-100, bullish=100)

def log(self, txt, dt=None):
    dt = dt or self.datas[0].datetime.date(0)
    print('{0}, {1}'.format(dt.isoformat(), txt))

def __init__(self):
    self.dataopen = self.datas[0].open
    self.datahigh = self.datas[0].high
    self.datalow = self.datas[0].low
    self.dataclose = self.datas[0].close
    self.CDLENGULFING = bt.talib.CDLENGULFING(self.datas[0].open, self.datas[0].high, self.datas[0].low, self.datas[0].close)
    self.CDLDARKCLOUDCOVER = bt.talib.CDLDARKCLOUDCOVER(self.datas[0].open, self.datas[0].high, self.datas[0].low, self.datas[0].close)
    self.dyn_highest = DynamicHighest(self.data.high)

def next(self):
    if math.isclose(self.dataopen[0], self.datahigh[0], abs_tol=0.2):
        #Minimum Criteria to be Considered
        if self.datahigh[0] > self.datahigh[-1]:
            """print('DATE: '+str(self.date(0))+' CDLENGULFING: '+str(self.CDLENGULFING[0])+' CDLDARKCLOUDCOVER: '+str(self.CDLDARKCLOUDCOVER[0]))"""
            if (self.CDLENGULFING[0] == self.p.bearish or self.CDLDARKCLOUDCOVER[0] == self.p.bearish):
                self.log(' SELL CREATED AT PRICE: , %.2f' % self.dataclose[0])

if name == 'main':
for ticker in ticker_list:
print('Ticker: %s' % ticker)
cerebro = bt.Cerebro()
cerebro.broker.set_cash(10000)

    filename = '%s.csv' % ticker
    #modpath = os.path.dirname('/Users/ayushsomani/Desktop/BANKNIFTY/')
    datapath = os.path.join('/Users/ayushsomani/Desktop/BANKNIFTY/', filename)

    if not os.path.isfile(datapath):
        print('File: %s not found' % filename)
        sys.exit()

    dateparse = lambda date: datetime.strptime(date, '%Y-%m-%d')
    data = pd.read_csv(datapath, index_col='Date', parse_dates=True, date_parser=dateparse)
    datafeed = bt.feeds.PandasData(dataname=data)
    cerebro.adddata(datafeed)
    print('Starting Portfolio Value: %.2f' % cerebro.broker.get_value())
    cerebro.addstrategy(MyStrategy)
    cerebro.run()
    print('Final Portfolio Value: %.2f' % cerebro.broker.get_value())

I am unable to write the logic in my Created Indicator (DynamicHighest). I tried various approaches but couldn't get the success. Moreover, some of the rows are skipped when I run my Cerebro. My CSV starts on Date - 17/09/18 but when I debugged it is processing from Date - 04/10/18. Why so?

CSV File

ayush-somani

@dasch
What does exactly FindFirstIndexHighest does (if I use it calculating count of those bars)?

run-out

@ayush-somani said in Indicator that checks if the current candle high (as soon as a condition is met) breaks the previous days high (over a specific period ) or not:

What does exactly FindFirstIndexHighest does (if I use it calculating count of those bars)?

This will return the index location of the first highest value looking back. See docs.

ayush-somani

@run-out
I am not getting the desired output.

Code Snippet -

from future import (absolute_import, division, print_function,
unicode_literals)
from datetime import datetime
import os.path
import sys
import backtrader as bt
import pandas as pd
import math

ticker_list = ['SBIN']

class DynamicHighest(bt.Indicator):
lines = ('firsthighindex','periodline',)
params = dict(minperiod=3, maxperiod=20)

def __init__(self):
    self.addminperiod(self.p.minperiod)
    self.l.firsthighindex = bt.ind.FindFirstIndexHighest(self.datas[0].high, period=self.p.maxperiod)

def next(self):
    if self.datas[0].high[0] < self.datas[0].high[-int(self.l.firsthighindex[0])]:
        self.l.periodline[0] = self.l.firsthighindex[0]

class MyStrategy(bt.Strategy):
params = dict(bearish=-100, bullish=100)

def log(self, txt, dt=None):
    dt = dt or self.datas[0].datetime.date(0)
    print('{0}, {1}'.format(dt.isoformat(), txt))

def __init__(self):
    self.dataopen = self.datas[0].open
    self.datahigh = self.datas[0].high
    self.datalow = self.datas[0].low
    self.dataclose = self.datas[0].close
    self.CDLENGULFING = bt.talib.CDLENGULFING(self.datas[0].open, self.datas[0].high, self.datas[0].low, self.datas[0].close)
    self.CDLDARKCLOUDCOVER = bt.talib.CDLDARKCLOUDCOVER(self.datas[0].open, self.datas[0].high, self.datas[0].low, self.datas[0].close)
    self.dyn_highest = DynamicHighest(self.datas[0])

def next(self):
    if math.isclose(self.dataopen[0], self.datahigh[0], abs_tol=0.2):
        #Minimum Criteria to be Considered
        if self.datahigh[0] > self.datahigh[-1]:
            if (self.CDLENGULFING[0] == self.p.bearish or self.CDLDARKCLOUDCOVER[0] == self.p.bearish):
                if self.dyn_highest.l.periodline[0] >= 3:
                    self.log(' SELL CREATED AT PRICE: , %.2f with %.2f number of days taken out.' % (self.dataclose[0],abs(self.dyn_highest.l.periodline[0])))

if name == 'main':
for ticker in ticker_list:
print('Ticker: %s' % ticker)
cerebro = bt.Cerebro()
cerebro.broker.set_cash(10000)

    filename = '%s.csv' % ticker
    #modpath = os.path.dirname('/Users/ayushsomani/Desktop/BANKNIFTY/')
    datapath = os.path.join('/Users/ayushsomani/Desktop/BANKNIFTY/', filename)

    if not os.path.isfile(datapath):
        print('File: %s not found' % filename)
        sys.exit()

    dateparse = lambda date: datetime.strptime(date, '%Y-%m-%d')
    data = pd.read_csv(datapath, index_col='Date', parse_dates=True, date_parser=dateparse)
    datafeed = bt.feeds.PandasData(dataname=data)

    cerebro.adddata(datafeed)
    print('Starting Portfolio Value: %.2f' % cerebro.broker.get_value())
    cerebro.addstrategy(MyStrategy)
    cerebro.run()
    print('Final Portfolio Value: %.2f' % cerebro.broker.get_value())

Output -
Ticker: SBIN
Starting Portfolio Value: 10000.00
2018-11-19, SELL CREATED AT PRICE: , 288.15 with 9.00 number of days taken out.
2018-12-04, SELL CREATED AT PRICE: , 282.55 with 19.00 number of days taken out.
2020-06-24, SELL CREATED AT PRICE: , 184.60 with 12.00 number of days taken out.
Final Portfolio Value: 10000.00

On Chart (https://in.tradingview.com/chart/K307Cbrw/), we can clearly see that 04/12/2018 High is broken on 30/11/2018 but the Output is showing that it took 19 bars to break this high. Hope, you understood my query what I am trying to achieve.

ayush-somani

@run-out
This is my CSV file.

run-out

@ayush-somani said in Indicator that checks if the current candle high (as soon as a condition is met) breaks the previous days high (over a specific period ) or not:

if self.datas[0].high[0] < self.datas[0].high[-int(self.l.firsthighindex[0])]:

Using MaxN might be more straight forward.