Hey Carson,
I guess you are confused between sizer and size.
sizer is a python class that comes with the backtrader module.
size is the quantity that you want to buy during a specific self.buy() call.

In order to derive the size to be bought logically, you must create a sub class that inherits from bt.Sizer base class.

import backtrader as bt

class FixedSize(bt.Sizer):
    params = (('stake', 1),)

    def _getsizing(self, comminfo, cash, data, isbuy):
        return self.params.stake

something like this...

More on bt.Sizers available here.

Tried running the illustration given in the docs.
It works well for destination timeframe 'weekly', compression 1
But fails for timeframes 'daily' and 'minute' using either compression.

Here's the code that was executed.
Note:
-The handy dictionary for timeframe conversion is updated for 'minute'

from __future__ import (absolute_import, division, print_function,
                        unicode_literals)

import argparse

import backtrader as bt
import backtrader.feeds as btfeeds
import backtrader.indicators as btind
from datetime import datetime


class SMAStrategy(bt.Strategy):
    params = (
        ('period', 10),
        ('onlydaily', False),
    )

    def __init__(self):
        self.sma = btind.SMA(self.data, period=self.p.period)

    def start(self):
        self.counter = 0

    def prenext(self):
        self.counter += 1
        print('prenext len %d - counter %d' % (len(self), self.counter))

    def next(self):
        self.counter += 1
        print('---next len %d - counter %d' % (len(self), self.counter))


def runstrat():
    args = parse_args()

    # Create a cerebro entity
    cerebro = bt.Cerebro(stdstats=False)

    cerebro.addstrategy(
        SMAStrategy,
        # args for the strategy
        period=args.period,
    )

    # Load the Data
    datapath = args.dataname or "nifty19-20.csv"
    data = btfeeds.BacktraderCSVData(dataname=datapath, fromdate=datetime.now().replace(year=2020, month=8, day=17, hour=0, minute=0, second=0), timeframe=bt.TimeFrame.Minutes)

    # Handy dictionary for the argument timeframe conversion
    tframes = dict(
        minute=bt.TimeFrame.Minutes,
        daily=bt.TimeFrame.Days,
        weekly=bt.TimeFrame.Weeks,
        monthly=bt.TimeFrame.Months)

    # First add the original data - smaller timeframe
    cerebro.replaydata(data,
                       timeframe=tframes['minute'],
                       compression=5)

    # Run over everything
    cerebro.run()

    # Plot the result
    cerebro.plot(style='bar')


def parse_args():
    parser = argparse.ArgumentParser(
        description='Pandas test script')

    parser.add_argument('--dataname', default='', required=False,
                        help='File Data to Load')

    parser.add_argument('--timeframe', default='minute', required=False,
                        choices=['minute', 'daily', 'weekly', 'monhtly'],
                        help='Timeframe to resample to')

    parser.add_argument('--compression', default=10, required=False, type=int,
                        help='Compress n bars into 1')

    parser.add_argument('--period', default=10, required=False, type=int,
                        help='Period to apply to indicator')

    return parser.parse_args()


if __name__ == '__main__':
    runstrat()

Output:
---next len 1476 - counter 1476
---next len 1477 - counter 1477
---next len 1478 - counter 1478
---next len 1479 - counter 1479
---next len 1480 - counter 1480
---next len 1481 - counter 1481
---next len 1482 - counter 1482
---next len 1483 - counter 1483
---next len 1484 - counter 1484
---next len 1485 - counter 1485
---next len 1486 - counter 1486
---next len 1487 - counter 1487
---next len 1488 - counter 1488
---next len 1489 - counter 1489
---next len 1490 - counter 1490
---next len 1491 - counter 1491
---next len 1492 - counter 1492
---next len 1493 - counter 1493
---next len 1494 - counter 1494
---next len 1495 - counter 1495
---next len 1496 - counter 1496
---next len 1497 - counter 1497
---next len 1498 - counter 1498
---next len 1499 - counter 1499
---next len 1500 - counter 1500

Am I missing something?

@Carson-Lansdowne said in Attribute Error: object has no attribute 'postion':

postion

Hey Carson.
A little modification in the 31st line of your code is required.
Under the class method next within the class buynhold_1, make the following changes:

Old code:

if not self.postion:
            if self.currdate == datetime.date(2016, 1, 3):
                self.buy()
                self.bar_executed = len(self)

New Code:

if not self.position:
            if self.currdate == datetime.date(2016, 1, 3):
                self.buy()
                self.bar_executed = len(self)

It's a mere spelling error self.position and not self.postion. Hope it works now:)

@Carson-Lansdowne said in Accessing variable defined in strategy from main Cerebro:

Hey Carson,

def stop(self):
        global roi
        # calculate the actual returns
        print(self.value)
        roi = ((self.broker.get_value() -self.val_start) / (self.value))
        print('ROI:        {:.2f}%'.format(100.0 * self.roi))
        
        #return self.roi

This snippet is copied from your code as it was.
You have defined a new variable global roi but you are trying to access it as a class attribute. Try using roi instead of self.roi in the print statement of your method. Hope it works. :)

Already done that. Still no Luck.

@rajanprabu said in Compression using daily and minute timeframes unsuccessful during replaydata():

te timefr

I tried it with the additional param compression=1 but the result is still the same.

from __future__ import (absolute_import, division, print_function,
                        unicode_literals)

import argparse

import backtrader as bt
import backtrader.feeds as btfeeds
import backtrader.indicators as btind
from datetime import datetime


class SMAStrategy(bt.Strategy):
    params = (
        ('period', 10),
        ('onlydaily', False),
    )

    def __init__(self):
        pass

    def start(self):
        self.counter = 0

    def prenext(self):
        self.counter += 1
        print(f'{self.datas[0].datetime.datetime(0)} prenext len %d - counter %d' % (len(self), self.counter))

    def next(self):
        self.counter += 1
        print(f'{self.datas[0].datetime.datetime(0)} ---next len %d - counter %d' % (len(self), self.counter))


def runstrat():
    args = parse_args()

    # Create a cerebro entity
    cerebro = bt.Cerebro(stdstats=False)

    cerebro.addstrategy(
        SMAStrategy,
        # args for the strategy
        period=args.period,
    )

    # Load the Data
    datapath = args.dataname or "nifty19-20.csv"
    data = btfeeds.BacktraderCSVData(dataname=datapath, fromdate=datetime.now().replace(year=2020, month=8, day=17, hour=0, minute=0, second=0), timeframe=bt.TimeFrame.Minutes, compression=1)

    # Handy dictionary for the argument timeframe conversion
    tframes = dict(
        minute=bt.TimeFrame.Minutes,
        daily=bt.TimeFrame.Days,
        weekly=bt.TimeFrame.Weeks,
        monthly=bt.TimeFrame.Months)

    # First add the original data - smaller timeframe
    cerebro.replaydata(data,
                       timeframe=tframes['daily'],
                       compression=1)

    # Run over everything
    cerebro.run()

    # Plot the result
    cerebro.plot(style='bar')


def parse_args():
    parser = argparse.ArgumentParser(
        description='Pandas test script')

    parser.add_argument('--dataname', default='', required=False,
                        help='File Data to Load')

    parser.add_argument('--timeframe', default='minute', required=False,
                        choices=['minute', 'daily', 'weekly', 'monhtly'],
                        help='Timeframe to resample to')

    parser.add_argument('--compression', default=10, required=False, type=int,
                        help='Compress n bars into 1')

    parser.add_argument('--period', default=10, required=False, type=int,
                        help='Period to apply to indicator')

    return parser.parse_args()


if __name__ == '__main__':
    runstrat()

2020-08-20 23:59:59.999989 ---next len 1482 - counter 1482
2020-08-20 23:59:59.999989 ---next len 1483 - counter 1483
2020-08-20 23:59:59.999989 ---next len 1484 - counter 1484
2020-08-20 23:59:59.999989 ---next len 1485 - counter 1485
2020-08-20 23:59:59.999989 ---next len 1486 - counter 1486
2020-08-20 23:59:59.999989 ---next len 1487 - counter 1487
2020-08-20 23:59:59.999989 ---next len 1488 - counter 1488
2020-08-20 23:59:59.999989 ---next len 1489 - counter 1489
2020-08-20 23:59:59.999989 ---next len 1490 - counter 1490
2020-08-20 23:59:59.999989 ---next len 1491 - counter 1491
2020-08-20 23:59:59.999989 ---next len 1492 - counter 1492
2020-08-20 23:59:59.999989 ---next len 1493 - counter 1493
2020-08-20 23:59:59.999989 ---next len 1494 - counter 1494
2020-08-20 23:59:59.999989 ---next len 1495 - counter 1495
2020-08-20 23:59:59.999989 ---next len 1496 - counter 1496
2020-08-20 23:59:59.999989 ---next len 1497 - counter 1497
2020-08-20 23:59:59.999989 ---next len 1498 - counter 1498
2020-08-20 23:59:59.999989 ---next len 1499 - counter 1499
2020-08-20 23:59:59.999989 ---next len 1500 - counter 1500

:(

I still can;t figure out where exactly am I going wrong!
Btw, Thanks Rajan for the help.

1. But will it simulate replay data?
2. pandas will just help resample the minute data to a higher timeframe. Am I right?

try to add sessionstart and sessionend parameters. seems logical to have them.

It worked. Can't thank you enough @ab_trader

Datafeed: feed.csv [0_1610445720484_feed.csv](Uploading 100%)
Original timeframe of the feed is 1 minute

Objective:

• To create an indicator that uses data from two different time frames.

from backtrader import Strategy
from backtrader import Indicator
import pandas as pd

class Dummy(Indicator):
    lines = ('l',)

    def next(self):
        # Datetime of the latest completed bars in both Time-Frames
        print(self.data0.datetime.datetime(0), end=": ")
        print(self.data1.datetime.datetime(0))


class TestStrategy(Strategy):
    params = ()

    def __init__(self):
        # Initializing dummy indicator (defined above)
        self.d = Dummy(self.data0, self.data1)
        pass

    def next(self):
        # Datetime of the latest completed bars in both Time-Frames
        print(self.data0.datetime.datetime(0), end=": ")
        print(self.data1.datetime.datetime(0))


df  =  pd.read_csv('feed.csv',
                   index_col='timestamp',
                   parse_dates=True)

data = bt.feeds.PandasData(
    dataname=df,
    openinterest='oi',
    timeframe=bt.TimeFrame.Minutes,
    compression=1,
    sessionstart=sessionstart,
    sessionend=sessionend
)

#  Creating the Cerebro instance
cerebro = bt.Cerebro(runonce=False)

# Adding Strategy: TestStrategy (defined above)
cerebro.addstrategy(TestStrategy)

# Adding feed 1 -> `1 minute` data. Smaller Time-Frame First
cerebro.adddata(data)

# Adding feed 2 -> Resampling `1 minute' to `15 minute`
cerebro.resampledata(data,
                     timeframe=bt.TimeFrame.Minutes,
                     compression=15)

# Executing
cerebro.run(stdstats=False)
cerebro.plot(style='candle', barup='green')

Output using compression 15:

2020-01-01 09:15:00: 2020-01-01 09:15:00
2020-01-01 09:15:00: 2020-01-01 09:15:00
2020-01-01 09:16:00: 2020-01-01 09:15:00
2020-01-01 09:16:00: 2020-01-01 09:15:00
2020-01-01 09:17:00: 2020-01-01 09:15:00
2020-01-01 09:17:00: 2020-01-01 09:15:00
2020-01-01 09:18:00: 2020-01-01 09:15:00
2020-01-01 09:18:00: 2020-01-01 09:15:00
2020-01-01 09:19:00: 2020-01-01 09:15:00
2020-01-01 09:19:00: 2020-01-01 09:15:00
2020-01-01 09:20:00: 2020-01-01 09:15:00
2020-01-01 09:20:00: 2020-01-01 09:15:00
2020-01-01 09:21:00: 2020-01-01 09:15:00
2020-01-01 09:21:00: 2020-01-01 09:15:00
2020-01-01 09:22:00: 2020-01-01 09:15:00

But when the compression is changed to 16, execution throws Index Error

# Adding feed 2 -> Resampling `1 minute' to `15 minute`
cerebro.resampledata(data,
                     timeframe=bt.TimeFrame.Minutes,
                     compression=16)

Output using compression 16:

  File "C:\Users\PC2\PycharmProjects\Backtrader\venv\lib\site-packages\backtrader\lineiterator.py", line 347, in nextstart
    self.next()
  File "C:/Users/PC2/.PyCharmCE2019.1/config/scratches/Test.py", line 10, in next
    print(self.data1.datetime.datetime(0))
  File "C:\Users\PC2\PycharmProjects\Backtrader\venv\lib\site-packages\backtrader\linebuffer.py", line 387, in datetime
    return num2date(self.array[self.idx + ago],
IndexError: array index out of range

Problem: Multi-Timeframe works well with compression=15 but fails with compression=16 (or any other than 15)

Observation:

self.data1.datetime.datetime(0)

works well for all Time-Frames in the next method of Strategy. But throws Index Error for Time-Frames other than 15 in the next method of Indicator

I'd be really grateful if someone can help me understand where am I going wrong.

@rajanprabu said in Multi-Timeframe works for compression=15 but fails for 16:

params = (('period', 16),)

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

Worked like a charm...
Thank you very much @rajanprabu

Typo^
Market hours: 09:15hrs to 15:30hrs

I'm trying to resample 1 minute bars to 20 minutes.
But the resampled bars are produced at absurd intervals.

Source of datafeed: Interactive Brokers API
Market hours: 09:00 hrs to 15:30 hrs
Backfill duration: 10 days

Note: Data provided by IB API is left-edge labeled (forward looking) by default. BackwardLookingFilter is a simple workaround that handles rightedge-ing minute bars. The problem persists with or without it.

from datetime import datetime, timedelta

import backtrader as bt

fromdate = datetime.now().date() - timedelta(days=10)
sessionstart = datetime.now().time().replace(hour=9, minute=15, second=0, microsecond=0)
sessionend = datetime.now().time().replace(hour=15, minute=30, second=0, microsecond=0)


class BackwardLookingFilter(object):

    def __init__(self, data):
        pass

    def __call__(self, data):
        data.datetime[0] = data.date2num(data.datetime.datetime(0) + timedelta(minutes=1))
        return False


class Test(bt.Strategy)

    def next(self):

        print("Minutes 1 ", self.data0.datetime.datetime(0))
        print("Minutes 20 ", self.data1.datetime.datetime(0))
        print("_"*50)



cerebro = bt.Cerebro()
cerebro.addstrategy(Test)

store = bt.stores.IBStore(port=7497, _debug=False)
cerebro.setbroker(store.getbroker())


minutedata = store.getdata(dataname="NIFTY50_IND_NSE", fromdate=fromdate, rtbar=True, sessionstart=sessionstart,
                           historical=True, sessionend=sessionend, timeframe=bt.TimeFrame.Minutes,
                           compression=1)

cerebro.adddata(minutedata)

minutedata.addfilter(BackwardLookingFilter)


cerebro.resampledata(minutedata,
                     timeframe=bt.TimeFrame.Minutes,
                     compression=20,
                     bar2edge=False,
                     boundoff=5)

cerebro.run()

Output:

Server Version: 76
TWS Time at connection:20220226 19:12:01 India Standard Time
Minutes 1  2022-02-16 09:35:00
Minutes 20 2022-02-16 09:35:00
__________________________________________________
Minutes 1  2022-02-16 09:36:00
Minutes 20 2022-02-16 09:35:00
__________________________________________________
Minutes 1  2022-02-16 09:37:00
Minutes 20 2022-02-16 09:35:00
__________________________________________________
Minutes 1  2022-02-16 09:38:00
Minutes 20 2022-02-16 09:37:00
__________________________________________________
Minutes 1  2022-02-16 09:39:00
Minutes 20 2022-02-16 09:37:00
__________________________________________________
Minutes 1  2022-02-16 09:40:00
Minutes 20 2022-02-16 09:37:00
__________________________________________________

Ideally after 09:35:00 the next 20 Minute bar should be produced at 09:55:00.

Unexpected behaviour:
The 20 Minute datafeed produces a bar at time: 09:37:00. (4th row in output)

Please shed some light on where I went wrong.

@rajanprabu said in Multi-Timeframe works for compression=15 but fails for 16:

params = (('period', 16),)

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

Worked like a charm...
Thank you very much @rajanprabu

Datafeed: feed.csv [0_1610445720484_feed.csv](Uploading 100%)
Original timeframe of the feed is 1 minute

Objective:

• To create an indicator that uses data from two different time frames.

from backtrader import Strategy
from backtrader import Indicator
import pandas as pd

class Dummy(Indicator):
    lines = ('l',)

    def next(self):
        # Datetime of the latest completed bars in both Time-Frames
        print(self.data0.datetime.datetime(0), end=": ")
        print(self.data1.datetime.datetime(0))


class TestStrategy(Strategy):
    params = ()

    def __init__(self):
        # Initializing dummy indicator (defined above)
        self.d = Dummy(self.data0, self.data1)
        pass

    def next(self):
        # Datetime of the latest completed bars in both Time-Frames
        print(self.data0.datetime.datetime(0), end=": ")
        print(self.data1.datetime.datetime(0))


df  =  pd.read_csv('feed.csv',
                   index_col='timestamp',
                   parse_dates=True)

data = bt.feeds.PandasData(
    dataname=df,
    openinterest='oi',
    timeframe=bt.TimeFrame.Minutes,
    compression=1,
    sessionstart=sessionstart,
    sessionend=sessionend
)

#  Creating the Cerebro instance
cerebro = bt.Cerebro(runonce=False)

# Adding Strategy: TestStrategy (defined above)
cerebro.addstrategy(TestStrategy)

# Adding feed 1 -> `1 minute` data. Smaller Time-Frame First
cerebro.adddata(data)

# Adding feed 2 -> Resampling `1 minute' to `15 minute`
cerebro.resampledata(data,
                     timeframe=bt.TimeFrame.Minutes,
                     compression=15)

# Executing
cerebro.run(stdstats=False)
cerebro.plot(style='candle', barup='green')

Output using compression 15:

2020-01-01 09:15:00: 2020-01-01 09:15:00
2020-01-01 09:15:00: 2020-01-01 09:15:00
2020-01-01 09:16:00: 2020-01-01 09:15:00
2020-01-01 09:16:00: 2020-01-01 09:15:00
2020-01-01 09:17:00: 2020-01-01 09:15:00
2020-01-01 09:17:00: 2020-01-01 09:15:00
2020-01-01 09:18:00: 2020-01-01 09:15:00
2020-01-01 09:18:00: 2020-01-01 09:15:00
2020-01-01 09:19:00: 2020-01-01 09:15:00
2020-01-01 09:19:00: 2020-01-01 09:15:00
2020-01-01 09:20:00: 2020-01-01 09:15:00
2020-01-01 09:20:00: 2020-01-01 09:15:00
2020-01-01 09:21:00: 2020-01-01 09:15:00
2020-01-01 09:21:00: 2020-01-01 09:15:00
2020-01-01 09:22:00: 2020-01-01 09:15:00

But when the compression is changed to 16, execution throws Index Error

# Adding feed 2 -> Resampling `1 minute' to `15 minute`
cerebro.resampledata(data,
                     timeframe=bt.TimeFrame.Minutes,
                     compression=16)

Output using compression 16:

  File "C:\Users\PC2\PycharmProjects\Backtrader\venv\lib\site-packages\backtrader\lineiterator.py", line 347, in nextstart
    self.next()
  File "C:/Users/PC2/.PyCharmCE2019.1/config/scratches/Test.py", line 10, in next
    print(self.data1.datetime.datetime(0))
  File "C:\Users\PC2\PycharmProjects\Backtrader\venv\lib\site-packages\backtrader\linebuffer.py", line 387, in datetime
    return num2date(self.array[self.idx + ago],
IndexError: array index out of range

Problem: Multi-Timeframe works well with compression=15 but fails with compression=16 (or any other than 15)

Observation:

self.data1.datetime.datetime(0)

works well for all Time-Frames in the next method of Strategy. But throws Index Error for Time-Frames other than 15 in the next method of Indicator

I'd be really grateful if someone can help me understand where am I going wrong.

try to add sessionstart and sessionend parameters. seems logical to have them.

It worked. Can't thank you enough @ab_trader

1. But will it simulate replay data?
2. pandas will just help resample the minute data to a higher timeframe. Am I right?

Hello, I've tried replaying data containing one minute bars. But there's something I'm missing. I've used the example given in the docs. It works well for destination timeframe weekly but fails for daily and intraday.

Here is a brief explanation of what I've done.
Please have a look into it.
Thank you in advance:)

@Tom-Porter said in typo found in indicators documentation.:

class MyStrategy(bt.Strategy):

    def __init__(self):

        sma1 = btind.SimpleMovingAverage(self.data)
        ema1 = btind.ExponentialMovingAverage()

        close_over_sma = self.data.close > sma1
        close_over_ema = self.data.close > ema1
        sma_ema_diff = sma1 - ema1

        self.buy_sig = bt.And(...)

    def next(self):

        if self.buy_sig:
            self.buy()

Absolutely. Thanks for pointing it out Tom.

@Carson-Lansdowne said in Accessing variable defined in strategy from main Cerebro:

Hey Carson,

def stop(self):
        global roi
        # calculate the actual returns
        print(self.value)
        roi = ((self.broker.get_value() -self.val_start) / (self.value))
        print('ROI:        {:.2f}%'.format(100.0 * self.roi))
        
        #return self.roi

This snippet is copied from your code as it was.
You have defined a new variable global roi but you are trying to access it as a class attribute. Try using roi instead of self.roi in the print statement of your method. Hope it works. :)

Hey Carson,
I guess you are confused between sizer and size.
sizer is a python class that comes with the backtrader module.
size is the quantity that you want to buy during a specific self.buy() call.

In order to derive the size to be bought logically, you must create a sub class that inherits from bt.Sizer base class.

import backtrader as bt

class FixedSize(bt.Sizer):
    params = (('stake', 1),)

    def _getsizing(self, comminfo, cash, data, isbuy):
        return self.params.stake

something like this...

More on bt.Sizers available here.