Donchian Channel

[KT]

Hi, I have been trying to write donchian channel indicator using two sets of data: daily and hourly data. I referred to rorymack's post on simple channel indicator. However, I realised there are some errors in my code :

1. the indicator is based on hourly data instead of the daily data
2. consecutive buys and sells are executed instead of the desired selling after buying one stock.

I would also like to know how to use data1 in the custom built indicator.

Thanks for your help @backtrader and others :)

Below is my code.

from __future__ import (absolute_import, division, print_function,
                        unicode_literals)
import os.path
import argparse
import datetime

import backtrader as bt
import backtrader.feeds as btfeeds

class hilo6(bt.Indicator):
    lines = ('maxi','mini',)
    params = (
            ('period', 6),
            ('data',None),
            )

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

    def next(self):
        highs = self.data.high.get(size=self.p.period)
        lows = self.data.low.get(size=self.p.period)
        self.lines.maxi[0] = max(highs)
        self.lines.mini[0] = min(lows)
                

class St(bt.Strategy):
    
    def log(self, txt, dt=None):
        ''' Logging function fot this strategy'''
        dt = dt or self.data0.datetime.date(0)
        dt1 = self.data0.datetime.time(0)
        print('%s,%s, %s' % (dt.isoformat(),dt1.isoformat(), txt))
        
    def __init__(self):
        self.dataclose = self.data0.close
        self.channel = hilo6(data=self.data1)
        
    def notify_order(self, order):
        if order.status in [order.Submitted, order.Accepted]:
            return
        
        if order.status in [order.Completed]:
            if order.isbuy():
                self.log(
                    'BUY EXECUTED, Price: %.2f, Cost: %.2f' %
                    (order.executed.price,
                     order.executed.value))
                
                self.buyprice = order.executed.price
                
            else:  # Sell
                self.log('SELL EXECUTED, Price: %.2f, Cost: %.2f' %
                         (order.executed.price,
                          order.executed.value))
        
            self.bar_executed = len(self)
            
        elif order.status in [order.Canceled, order.Margin, order.Rejected]:
            self.log('Order Canceled/Margin/Rejected')
        
        self.order = None
        
    def notify_trade(self, trade):
        if not trade.isclosed:
            return

        self.log('OPERATION PROFIT, GROSS %.5f, NET %.5f' %
                 (trade.pnl, trade.pnlcomm))
        
    def next(self):
        txt = ','.join(
            ['%04d' % len(self.data0),
             '%04d' % len(self.data1),
             self.data0.datetime.date(0).isoformat(),
             self.data0.datetime.time(0).isoformat(),
             '%.5f' % self.data0.close[0],
             '%.5f' % self.data1.close[0]]) 
        print(txt)
        
        if not self.position:
             if self.data0.close[-1] < self.channel.lines.mini[0]:
                if self.data0.close[0] > self.channel.lines.mini[0]:
                    self.log('BUY CREATE, %.5f' % self.data0.close[0])
                    self.order=self.buy(data=self.data0)
        
        else:
            if self.data0.close[-1] > self.channel.lines.maxi[0]:
                if self.data0.close[0] < self.channel.lines.maxi[0]:
                    self.log('SELL CREATE, %.5f' % self.data0.close[0])
                    self.order = self.sell(data=self.data0)

                    
def runstrat():
    args = parse_args()

    cerebro = bt.Cerebro()

data0 = btfeeds.GenericCSVData(dataname=datapath0,
                                   timeframe=bt.TimeFrame.Minutes,
                                  fromdate = datetime.datetime(2017,1,1),
                                  todate=datetime.datetime(2017,4,20),                                  
                                  nullvalue=0.0,
                                  dtformat=('%Y.%m.%d'),
                                  tmformat=('%H:%M'),
                                  datetime=0,
                                  time=1,
                                  high=2,
                                  low=3,
                                  open=4,
                                  close=5,
                                  volume=6,
                                  openinterest=-1)
    data1 = btfeeds.GenericCSVData(dataname=datapath1,
                                   timeframe = bt.TimeFrame.Days,
                                  fromdate =datetime.datetime(2017,1,1),
                                  todate= datetime.datetime(2017,4,20),
                                  nullvalue=0.0,
                                  dtformat=('%Y.%m.%d'),
                                  tmformat=('%H:%M'),
                                  datetime=0,
                                  time=1,
                                  high=2,
                                  low=3,
                                  open=4,
                                  close=5,
                                  volume=6,
                                  openinterest=-1)
    cerebro.adddata(data0,name ="data0")
    cerebro.adddata(data1,name="data1")

    cerebro.addstrategy(St)
    cerebro.broker.setcash(100000.0)
    print('Starting Portfolio Value: %.2f' % cerebro.broker.getvalue())
    cerebro.run(stdstats=False, runonce=False)
    print('Final Portfolio Value: %.2f' % cerebro.broker.getvalue())
        
def parse_args():
    parser = argparse.ArgumentParser(
        formatter_class=argparse.ArgumentDefaultsHelpFormatter,
        description='Sample for pivot point and cross plotting')
    parser.add_argument('--plot', required=False, action='store_true',
                        help=('Plot the result'))

    return parser.parse_args()


if __name__ == '__main__':
    runstrat()

[backtrader]

@KT said in Donchian Channel:

the indicator is based on hourly data instead of the daily data

Indicators don't care about the nature of the data unless they need to do something like looking at the time of day and the indicator then becomes only usable with data feeds which carry that payload, making it impossible to create something like this (SMA of an SMA)

ma1 = bt.ind.SMA()
ma2 = bt.ind.SMA(ma1)

@KT said in Donchian Channel:

I would also like to know how to use data1 in the custom built indicator.

self.data1 ? (or self.datas[1])

The indicator per-se seems fine. It's simply not written in the way the platform was envisioned, but an end user can do it. Actually this "channel" indicator is for sure in some of the samples.

class hilo6(bt.ind.PeriodN):
    lines = ('maxi', 'mini',)
    params = (('period', 6),)

    def __init__(self):
        self.lines.maxi = bt.ind.Highest(self.data.high, period=self.p.period)
        self.lines.mini = bt.ind.Lowest(self.data.low, period=self.p.period)

This would be the canonical way, which is declarative rather than imperative.

[KT]

@backtrader I tried using self.data1 to access the first data in the instantiation of the indicator. However, I get this error.

AttributeError: 'Lines_LineSeries_LineIterator_DataAccessor_Indicat' object has no attribute 'data1'.

If I use self.datas[1], the following error occurs.

IndexError: list index out of range

[backtrader]

Ok, problem understood

@KT said in Donchian Channel in __init__:

self.channel = hilo6(data=self.data1)

As you may see above in the reworked version of the indicator, there is no such thing as a data parameter defined. The data feeds are passed automatically (if none is specified) or explicitly (as unnamed arguments).

See Docs - Platform Concepts

As such, use:

self.channel = hilo6(self.data1)

Inside the indicator, this is in this case the only data feed. Which keeps the definition of the indicator generic.

[ab_trader]

You may want to use backtrader's CrossUp and CrossDown indicators to generate Buy and Sell signals. This will simplify your code:

    def __init__(self):
        self.dataclose = self.data0.close
        self.channel = hilo6(data=self.data1)

        self.buysig = bt.indicators.CrossUp(self.data0.close, self.channel.hilo6.mini)
        self.sellsig = bt.indicators.CrossDown(self.data0.close, self.channel.hilo6.maxi)

and then:

    def next(self):
        txt = ','.join(
            ['%04d' % len(self.data0),
             '%04d' % len(self.data1),
             self.data0.datetime.date(0).isoformat(),
             self.data0.datetime.time(0).isoformat(),
             '%.5f' % self.data0.close[0],
             '%.5f' % self.data1.close[0]]) 
        print(txt)
        
        if not self.position:
             if self.buysig[0]:
                    self.log('BUY CREATE, %.5f' % self.data0.close[0])
                    self.order=self.buy(data=self.data0)
        
        else:
            if self.sellsig[0]:
                    self.log('SELL CREATE, %.5f' % self.data0.close[0])
                    self.order = self.sell(data=self.data0)

[KT]

@ab_trader thanks for the suggestion :)

@backtrader thanks, the code works. The indicator only uses the data from data1.

However, I notice there are some errors. When I print out self.data0.datetime.date(0) and self.data1.datetime.date(1), the dates are different.

The snippet below shows data in this format :
len(self.data0),len(self.data1),self.data0.datetime.date(0).isoformat(),
self.data1.datetime.date(0),self.data0.datetime.time(0).isoformat(),....

0551,0024,2017-03-22,2017-03-21,21:00:00,1.24873,1.24707
0552,0024,2017-03-22,2017-03-21,22:00:00,1.24830,1.24707
0553,0024,2017-03-22,2017-03-21,23:00:00,1.24784,1.24707
0553,0025,2017-03-22,2017-03-22,23:00:00,1.24784,1.24784
0554,0025,2017-03-23,2017-03-22,00:00:00,1.24740,1.24784
0555,0025,2017-03-23,2017-03-22,01:00:00,1.24708,1.24784
0556,0025,2017-03-23,2017-03-22,02:00:00,1.24713,1.24784
0557,0025,2017-03-23,2017-03-22,03:00:00,1.24833,1.24784
0558,0025,2017-03-23,2017-03-22,04:00:00,1.24830,1.24784
0559,0025,2017-03-23,2017-03-22,05:00:00,1.24837,1.24784
0560,0025,2017-03-23,2017-03-22,06:00:00,1.24862,1.24784
0561,0025,2017-03-23,2017-03-22,07:00:00,1.24863,1.24784
0562,0025,2017-03-23,2017-03-22,08:00:00,1.25070,1.24784
0563,0025,2017-03-23,2017-03-22,09:00:00,1.24771,1.24784
0564,0025,2017-03-23,2017-03-22,10:00:00,1.24751,1.24784
0565,0025,2017-03-23,2017-03-22,11:00:00,1.24918,1.24784
0566,0025,2017-03-23,2017-03-22,12:00:00,1.25096,1.24784
0567,0025,2017-03-23,2017-03-22,13:00:00,1.24957,1.24784
0568,0025,2017-03-23,2017-03-22,14:00:00,1.24990,1.24784
0569,0025,2017-03-23,2017-03-22,15:00:00,1.25181,1.24784
0570,0025,2017-03-23,2017-03-22,16:00:00,1.25220,1.24784
0571,0025,2017-03-23,2017-03-22,17:00:00,1.25257,1.24784
0572,0025,2017-03-23,2017-03-22,18:00:00,1.25267,1.24784
0573,0025,2017-03-23,2017-03-22,19:00:00,1.25146,1.24784
0574,0025,2017-03-23,2017-03-22,20:00:00,1.25130,1.24784
0575,0025,2017-03-23,2017-03-22,21:00:00,1.25212,1.24784
0576,0025,2017-03-23,2017-03-22,22:00:00,1.25199,1.24784
0577,0025,2017-03-23,2017-03-22,23:00:00,1.25160,1.24784
0577,0026,2017-03-23,2017-03-23,23:00:00,1.25160,1.25160
0578,0026,2017-03-24,2017-03-23,00:00:00,1.25167,1.25160

If i change it to self.data0.datetime.date(-1) and self.data1.datetime.date(1), the date does not tally too.

The snippet below shows data in this format :
len(self.data0),len(self.data1),self.data0.datetime.date(-1).isoformat(),
self.data1.datetime.date(0),self.data0.datetime.time(0).isoformat(),....

0558,0025,2017-03-23,2017-03-22,04:00:00,1.24830,1.24784
0559,0025,2017-03-23,2017-03-22,05:00:00,1.24837,1.24784
0560,0025,2017-03-23,2017-03-22,06:00:00,1.24862,1.24784
0561,0025,2017-03-23,2017-03-22,07:00:00,1.24863,1.24784
0562,0025,2017-03-23,2017-03-22,08:00:00,1.25070,1.24784
0563,0025,2017-03-23,2017-03-22,09:00:00,1.24771,1.24784
0564,0025,2017-03-23,2017-03-22,10:00:00,1.24751,1.24784
0565,0025,2017-03-23,2017-03-22,11:00:00,1.24918,1.24784
0566,0025,2017-03-23,2017-03-22,12:00:00,1.25096,1.24784
0567,0025,2017-03-23,2017-03-22,13:00:00,1.24957,1.24784
0568,0025,2017-03-23,2017-03-22,14:00:00,1.24990,1.24784
0569,0025,2017-03-23,2017-03-22,15:00:00,1.25181,1.24784
0570,0025,2017-03-23,2017-03-22,16:00:00,1.25220,1.24784
0571,0025,2017-03-23,2017-03-22,17:00:00,1.25257,1.24784
0572,0025,2017-03-23,2017-03-22,18:00:00,1.25267,1.24784
0573,0025,2017-03-23,2017-03-22,19:00:00,1.25146,1.24784
0574,0025,2017-03-23,2017-03-22,20:00:00,1.25130,1.24784
0575,0025,2017-03-23,2017-03-22,21:00:00,1.25212,1.24784
0576,0025,2017-03-23,2017-03-22,22:00:00,1.25199,1.24784
0577,0025,2017-03-23,2017-03-22,23:00:00,1.25160,1.24784
0577,0026,2017-03-23,2017-03-23,23:00:00,1.25160,1.25160
0578,0026,2017-03-23,2017-03-23,00:00:00,1.25167,1.25160

Is there anyway to rectify the error such that the dates are synchronized? How do I change such that the time frame is on hourly basis? For now, i am using bt.TimeFrame.Minutes.

Thanks for all your help :)

[backtrader]

@KT said in Donchian Channel:

However, I notice there are some errors. When I print out self.data0.datetime.date(0) and self.data1.datetime.date(1), the dates are different.

The error seems to be on your side. date(0) is the date of the actual moment in time. date(1) is the date of the future next moment in time (which is unknown, but which due to python arithmetic in arrays gives you something)

If i change it to self.data0.datetime.date(-1) and self.data1.datetime.date(1), the date does not tally too.

It shouldn't. One is the previous date, the other is a non-existent date in the future.

Is there anyway to rectify the error such that the dates are synchronized?

With the given information there is for sure not an error. There isn't enough information to actually interpret something really, but it seems you have a daily feed, which moves slower and an intraday feed which moves faster. The daily feed remains static until the next day kicks in. Everything seems perfectly in order.

A daily feed is one that is complete. It cannot be delivered if the intraday feed is from the same day and at the beginning of the day. Or you'd else be cheating.

How do I change such that the time frame is on hourly basis? For now, i am using bt.TimeFrame.Minutes.

60 minutes.

[KT]

Sorry there is a typo there.

    def next(self):
        txt = ','.join(
            ['%04d' % len(self.data0),
             '%04d' % len(self.data1),
             self.data0.datetime.date(-1).isoformat(),
             self.data1.datetime.date(0).isoformat(),
             '%.5f' % self.data0.close[0],
             '%.5f' % self.data1.close[0]]) 
        print(txt)

The output is

1853,0079,2017-04-19,2017-04-18,1.28273,1.28400
1854,0079,2017-04-19,2017-04-18,1.28217,1.28400
1855,0079,2017-04-19,2017-04-18,1.28276,1.28400
1856,0079,2017-04-19,2017-04-18,1.28249,1.28400
1857,0079,2017-04-19,2017-04-18,1.28270,1.28400
1858,0079,2017-04-19,2017-04-18,1.28257,1.28400
1859,0079,2017-04-19,2017-04-18,1.28220,1.28400
1860,0079,2017-04-19,2017-04-18,1.28176,1.28400
1861,0079,2017-04-19,2017-04-18,1.28311,1.28400
1862,0079,2017-04-19,2017-04-18,1.28389,1.28400
1863,0079,2017-04-19,2017-04-18,1.28510,1.28400
1864,0079,2017-04-19,2017-04-18,1.28332,1.28400
1865,0079,2017-04-19,2017-04-18,1.28200,1.28400
1866,0079,2017-04-19,2017-04-18,1.28108,1.28400
1867,0079,2017-04-19,2017-04-18,1.27881,1.28400
1868,0079,2017-04-19,2017-04-18,1.27852,1.28400
1869,0079,2017-04-19,2017-04-18,1.27893,1.28400
1870,0079,2017-04-19,2017-04-18,1.27841,1.28400
1871,0079,2017-04-19,2017-04-18,1.27865,1.28400
1872,0079,2017-04-19,2017-04-18,1.27772,1.28400
1872,0080,2017-04-19,2017-04-19,1.27772,1.27772
1873,0080,2017-04-19,2017-04-19,1.27824,1.27772

If i change it to,

    def next(self):
        txt = ','.join(
            ['%04d' % len(self.data0),
             '%04d' % len(self.data1),
             self.data0.datetime.date(0).isoformat(),
             self.data1.datetime.date(0).isoformat(),
             '%.5f' % self.data0.close[0],
             '%.5f' % self.data1.close[0]]) 
        print(txt)

The output will change into

1837,0078,2017-04-18,2017-04-17,1.25564,1.25631
1838,0078,2017-04-18,2017-04-17,1.26521,1.25631
1839,0078,2017-04-18,2017-04-17,1.26682,1.25631
1840,0078,2017-04-18,2017-04-17,1.27188,1.25631
1841,0078,2017-04-18,2017-04-17,1.27387,1.25631
1842,0078,2017-04-18,2017-04-17,1.27641,1.25631
1843,0078,2017-04-18,2017-04-17,1.27670,1.25631
1844,0078,2017-04-18,2017-04-17,1.27662,1.25631
1845,0078,2017-04-18,2017-04-17,1.28302,1.25631
1846,0078,2017-04-18,2017-04-17,1.28490,1.25631
1847,0078,2017-04-18,2017-04-17,1.28434,1.25631
1848,0078,2017-04-18,2017-04-17,1.28400,1.25631
1848,0079,2017-04-18,2017-04-18,1.28400,1.28400
1849,0079,2017-04-19,2017-04-18,1.28426,1.28400
1850,0079,2017-04-19,2017-04-18,1.28428,1.28400
1851,0079,2017-04-19,2017-04-18,1.28435,1.28400
1852,0079,2017-04-19,2017-04-18,1.28359,1.28400
1853,0079,2017-04-19,2017-04-18,1.28273,1.28400
1854,0079,2017-04-19,2017-04-18,1.28217,1.28400
1855,0079,2017-04-19,2017-04-18,1.28276,1.28400

Based on these two outputs, the dates between data0 and data1 are not synchronized. As such, it will result in an error in terms of the signal generated from the indicator.

Data0 is an intraday feed with hourly observations while data1 is a daily feed with daily observations of high low open close and volume.

Thanks @backtrader !!

[backtrader]

@KT said in Donchian Channel:

Based on these two outputs, the dates between data0 and data1 are not synchronized

Sorry but it would seem you fail to understand how resampling works.

• A resampled bar can only be delivered when the bar is complete.

As such

• The intraday feed will deliver timestamps from the current day (2017-04-19 in your example ... although being it an intraday feed, printing the complete datetime would probably be better)

• The daily feed will be paused in the previous day (2017-04-18 in your example) until resampling is complete and the new complete bar can be delivered (2017-04-19)

In any case there are many missing factors in the code snippets to conclude anything:

• Missing input sample data (the data is actually different in your examples)
• A simple sample which reproduces (with the input data) what you think is wrong.

@KT said in Donchian Channel:

self.data0.datetime.date(-1).isoformat(),

That's in any case pointless, because printing the previous intraday bar is not going to be of much help when comparing to a paused (non-delivering) daily feed.

@KT said in Donchian Channel:

...
1848,0078,2017-04-18,2017-04-17,1.28400,1.25631
1848,0079,2017-04-18,2017-04-18,1.28400,1.28400
1849,0079,2017-04-19,2017-04-18,1.28426,1.28400
...

Even without intraday times, the synchronization seems fine

• In length 1848 of the intraday data (1st on the left), the feed is delivering for 2017-04-18 and the daily data is still frozen on the previous day 2017-04-17

  This seems to be the last price for 2017-04-18

• The intraday feed is frozen (hence the length remaining at 1848) because it was the last price for 2017-04-18, the daily

  And the daily data delivers for 2017-04-18, now that the prices are complete.

  Had you printed the length of the strategy and of data1, both would have advanced in this case.

• Once the daily data has synchronized itself by delivering, the intraday data delivers the 1st price for 2017-04-19 and the daily data remains frozen on the previous day, awaiting for completion.

Everything seems perfectly fine.

[KT]

@backtrader Thank you for your help so far :) Appreciate it.

        txt = ','.join(
            ['%04d' % len(self.data0),
             '%04d' % len(self.data1),
             self.data0.datetime.date(0).isoformat(),
             self.data0.datetime.time(0).isoformat(),
             self.data1.datetime.date(0).isoformat(),
             '%.5f' % self.data0.close[0],
             '%.5f' % self.data1.close[0]]) 
        print(txt)

The data are imported in this way.

data0 = btfeeds.GenericCSVData(dataname=datapath0,
                                   timeframe=bt.TimeFrame.Minutes,
                                  fromdate = datetime.datetime(2017,1,1),
                                  todate=datetime.datetime(2017,4,20),                                  
                                  nullvalue=0.0,
                                  dtformat=('%Y.%m.%d'),
                                  tmformat=('%H:%M'),
                                  datetime=0,
                                  time=1,
                                  open=2,
                                  high=3,
                                  low=4,
                                  close=5,
                                  volume=6,
                                  openinterest=-1)
    data1 = btfeeds.GenericCSVData(dataname=datapath1,
                                   timeframe = bt.TimeFrame.Days,
                                  fromdate =datetime.datetime(2017,1,1),
                                  todate= datetime.datetime(2017,4,20),
                                  nullvalue=0.0,
                                  dtformat=('%Y.%m.%d'),
                                  tmformat=('%H:%M'),
                                  datetime=0,
                                  time=1,
                                  open=2,
                                  high=3,
                                  low=4,
                                  close=5,
                                  volume=6,
                                  openinterest=-1)
    cerebro.adddata(data0,name ="Hourly Data")
    cerebro.adddata(data1,name="Daily Data")

The output is printed as below.

1770,0075,2017-04-13,17:00:00,2017-04-12,1.25171,1.25387
1771,0075,2017-04-13,18:00:00,2017-04-12,1.25210,1.25387
1772,0075,2017-04-13,19:00:00,2017-04-12,1.25155,1.25387
1773,0075,2017-04-13,20:00:00,2017-04-12,1.25104,1.25387
1774,0075,2017-04-13,21:00:00,2017-04-12,1.25097,1.25387
1775,0075,2017-04-13,22:00:00,2017-04-12,1.25053,1.25387
1776,0075,2017-04-13,23:00:00,2017-04-12,1.25008,1.25387
1776,0076,2017-04-13,23:00:00,2017-04-13,1.25008,1.25008
1777,0076,2017-04-14,00:00:00,2017-04-13,1.25060,1.25008
1778,0076,2017-04-14,01:00:00,2017-04-13,1.25068,1.25008
1779,0076,2017-04-14,02:00:00,2017-04-13,1.25040,1.25008
1780,0076,2017-04-14,03:00:00,2017-04-13,1.25047,1.25008
1781,0076,2017-04-14,04:00:00,2017-04-13,1.25071,1.25008
1782,0076,2017-04-14,05:00:00,2017-04-13,1.25043,1.25008
1783,0076,2017-04-14,06:00:00,2017-04-13,1.25065,1.25008
1784,0076,2017-04-14,07:00:00,2017-04-13,1.25099,1.25008
1785,0076,2017-04-14,08:00:00,2017-04-13,1.25054,1.25008
1786,0076,2017-04-14,09:00:00,2017-04-13,1.25142,1.25008
1787,0076,2017-04-14,10:00:00,2017-04-13,1.25124,1.25008
1788,0076,2017-04-14,11:00:00,2017-04-13,1.25114,1.25008
1789,0076,2017-04-14,12:00:00,2017-04-13,1.25115,1.25008
1790,0076,2017-04-14,13:00:00,2017-04-13,1.25095,1.25008
1791,0076,2017-04-14,14:00:00,2017-04-13,1.25084,1.25008
1792,0076,2017-04-14,15:00:00,2017-04-13,1.25299,1.25008
1793,0076,2017-04-14,16:00:00,2017-04-13,1.25271,1.25008
1794,0076,2017-04-14,17:00:00,2017-04-13,1.25325,1.25008
1795,0076,2017-04-14,18:00:00,2017-04-13,1.25272,1.25008
1796,0076,2017-04-14,19:00:00,2017-04-13,1.25264,1.25008
1797,0076,2017-04-14,20:00:00,2017-04-13,1.25262,1.25008
1798,0076,2017-04-14,21:00:00,2017-04-13,1.25257,1.25008
1799,0076,2017-04-14,22:00:00,2017-04-13,1.25273,1.25008
1800,0076,2017-04-14,23:00:00,2017-04-13,1.25172,1.25008
1800,0077,2017-04-14,23:00:00,2017-04-14,1.25172,1.25172
1801,0077,2017-04-17,00:00:00,2017-04-14,1.25331,1.25172
1802,0077,2017-04-17,01:00:00,2017-04-14,1.25301,1.25172
1803,0077,2017-04-17,02:00:00,2017-04-14,1.25352,1.25172
1804,0077,2017-04-17,03:00:00,2017-04-14,1.25416,1.25172
1805,0077,2017-04-17,04:00:00,2017-04-14,1.25370,1.25172
1806,0077,2017-04-17,05:00:00,2017-04-14,1.25423,1.25172
1807,0077,2017-04-17,06:00:00,2017-04-14,1.25410,1.25172
1808,0077,2017-04-17,07:00:00,2017-04-14,1.25357,1.25172
1809,0077,2017-04-17,08:00:00,2017-04-14,1.25323,1.25172
1810,0077,2017-04-17,09:00:00,2017-04-14,1.25369,1.25172
1811,0077,2017-04-17,10:00:00,2017-04-14,1.25351,1.25172
1812,0077,2017-04-17,11:00:00,2017-04-14,1.25386,1.25172
1813,0077,2017-04-17,12:00:00,2017-04-14,1.25470,1.25172
1814,0077,2017-04-17,13:00:00,2017-04-14,1.25440,1.25172
1815,0077,2017-04-17,14:00:00,2017-04-14,1.25472,1.25172
1816,0077,2017-04-17,15:00:00,2017-04-14,1.25531,1.25172
1817,0077,2017-04-17,16:00:00,2017-04-14,1.25678,1.25172
1818,0077,2017-04-17,17:00:00,2017-04-14,1.25933,1.25172
1819,0077,2017-04-17,18:00:00,2017-04-14,1.25863,1.25172
1820,0077,2017-04-17,19:00:00,2017-04-14,1.25855,1.25172
1821,0077,2017-04-17,20:00:00,2017-04-14,1.25816,1.25172
1822,0077,2017-04-17,21:00:00,2017-04-14,1.25630,1.25172
1823,0077,2017-04-17,22:00:00,2017-04-14,1.25620,1.25172
1824,0077,2017-04-17,23:00:00,2017-04-14,1.25631,1.25172
1824,0078,2017-04-17,23:00:00,2017-04-17,1.25631,1.25631
1825,0078,2017-04-18,00:00:00,2017-04-17,1.25649,1.25631
1826,0078,2017-04-18,01:00:00,2017-04-17,1.25590,1.25631
1827,0078,2017-04-18,02:00:00,2017-04-17,1.25565,1.25631
1828,0078,2017-04-18,03:00:00,2017-04-17,1.25633,1.25631
1829,0078,2017-04-18,04:00:00,2017-04-17,1.25675,1.25631
1830,0078,2017-04-18,05:00:00,2017-04-17,1.25673,1.25631
1831,0078,2017-04-18,06:00:00,2017-04-17,1.25635,1.25631
1832,0078,2017-04-18,07:00:00,2017-04-17,1.25634,1.25631
1833,0078,2017-04-18,08:00:00,2017-04-17,1.25692,1.25631
1834,0078,2017-04-18,09:00:00,2017-04-17,1.25857,1.25631
1835,0078,2017-04-18,10:00:00,2017-04-17,1.25948,1.25631
1836,0078,2017-04-18,11:00:00,2017-04-17,1.25864,1.25631
1837,0078,2017-04-18,12:00:00,2017-04-17,1.25564,1.25631
1838,0078,2017-04-18,13:00:00,2017-04-17,1.26521,1.25631
1839,0078,2017-04-18,14:00:00,2017-04-17,1.26682,1.25631
1840,0078,2017-04-18,15:00:00,2017-04-17,1.27188,1.25631
1841,0078,2017-04-18,16:00:00,2017-04-17,1.27387,1.25631
1842,0078,2017-04-18,17:00:00,2017-04-17,1.27641,1.25631
1843,0078,2017-04-18,18:00:00,2017-04-17,1.27670,1.25631
1844,0078,2017-04-18,19:00:00,2017-04-17,1.27662,1.25631
1845,0078,2017-04-18,20:00:00,2017-04-17,1.28302,1.25631
1846,0078,2017-04-18,21:00:00,2017-04-17,1.28490,1.25631
1847,0078,2017-04-18,22:00:00,2017-04-17,1.28434,1.25631
1848,0078,2017-04-18,23:00:00,2017-04-17,1.28400,1.25631
1848,0079,2017-04-18,23:00:00,2017-04-18,1.28400,1.28400
1849,0079,2017-04-19,00:00:00,2017-04-18,1.28426,1.28400
1850,0079,2017-04-19,01:00:00,2017-04-18,1.28428,1.28400
1851,0079,2017-04-19,02:00:00,2017-04-18,1.28435,1.28400
1852,0079,2017-04-19,03:00:00,2017-04-18,1.28359,1.28400
1853,0079,2017-04-19,04:00:00,2017-04-18,1.28273,1.28400
1854,0079,2017-04-19,05:00:00,2017-04-18,1.28217,1.28400
1855,0079,2017-04-19,06:00:00,2017-04-18,1.28276,1.28400
1856,0079,2017-04-19,07:00:00,2017-04-18,1.28249,1.28400
1857,0079,2017-04-19,08:00:00,2017-04-18,1.28270,1.28400
1858,0079,2017-04-19,09:00:00,2017-04-18,1.28257,1.28400
1859,0079,2017-04-19,10:00:00,2017-04-18,1.28220,1.28400
1860,0079,2017-04-19,11:00:00,2017-04-18,1.28176,1.28400
1861,0079,2017-04-19,12:00:00,2017-04-18,1.28311,1.28400
1862,0079,2017-04-19,13:00:00,2017-04-18,1.28389,1.28400
1863,0079,2017-04-19,14:00:00,2017-04-18,1.28510,1.28400
1864,0079,2017-04-19,15:00:00,2017-04-18,1.28332,1.28400
1865,0079,2017-04-19,16:00:00,2017-04-18,1.28200,1.28400
1866,0079,2017-04-19,17:00:00,2017-04-18,1.28108,1.28400
1867,0079,2017-04-19,18:00:00,2017-04-18,1.27881,1.28400
1868,0079,2017-04-19,19:00:00,2017-04-18,1.27852,1.28400
1869,0079,2017-04-19,20:00:00,2017-04-18,1.27893,1.28400
1870,0079,2017-04-19,21:00:00,2017-04-18,1.27841,1.28400
1871,0079,2017-04-19,22:00:00,2017-04-18,1.27865,1.28400
1872,0079,2017-04-19,23:00:00,2017-04-18,1.27772,1.28400
1872,0080,2017-04-19,23:00:00,2017-04-19,1.27772,1.27772
1873,0080,2017-04-20,00:00:00,2017-04-19,1.27824,1.27772

data0's day is one day faster than data1's day.

It seems that 23:00 is double counted for data1 (e.g. 0079 and 0080)

Is there any way to rectify the error?

[backtrader]

@KT said in Donchian Channel:

It seems that 23:00 is double counted for data1 (e.g. 0079 and 0080)
Is there any way to rectify the error?

You keep talking about errors. There is none. And there is again no sample data and an incomplete snippet.

1776,0075,2017-04-13,23:00:00,2017-04-12,1.25008,1.25387
1776,0076,2017-04-13,23:00:00,2017-04-13,1.25008,1.25008
1777,0076,2017-04-14,00:00:00,2017-04-13,1.25060,1.25008

Even in the absence of a complete snippet

• The bar is not repeated. The intraday data stream is frozen (hence the no change in length, which remains at 1776)
• Because the daily data stream can first deliver the daily resampled bar. The only signal available to the resampling filter is the change of the day

The platform cannot magically know when things come to an end. In this case the only signal is the change of date.

Use a trading calendar if you want to offer more information and a signal to the platform. See Docs - Trading Calendar

[KT]

@backtrader Sorry for the misunderstanding. I got it alr. Thankss for the help :)

[Ariel Iván Ruiz Mateos]

Hi @KT ! Could you paste the full source code with all the modifications?

thanks a lot! and good work!