Compression using daily and minute timeframes unsuccessful during replaydata()

Soham Jain

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?

Jinisha Sheth

I faced a similar problem:(
I guess the issue is how backtrader comprehends time frame of the smaller data feed provided. Unless explicitly mentioned, it defaults to daily and hence it can not build a bar with a tf smaller than daily.

Soham Jain

Already done that. Still no Luck.

rajanprabu

@Soham-Jain said in Compression using daily and minute timeframes unsuccessful during replaydata():

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)

@Soham-Jain I think you need to give compression along with timeframe. say compression=1 to indicate its 1 minute timeframe. BT by default takes it as 1 day but not sure if it takes 1 minutes for bt.TimeFrame.Minutes.

Soham Jain

@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.

crunchypickle

might be just easier to compress it yourself with Pandas

Soham Jain

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

crunchypickle

@Soham-Jain I haven't worked on resampling for a while but I'm quite sure pandas should support downsampling. That is a very common data science task.

ab_trader

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

Soham Jain

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

It worked. Can't thank you enough @ab_trader