Daily datafeed works BUT intraday doesn't

[chhrissi2909]

Hello,

I downloaded a daily data from yahoo and my intraday from tingoo.

My Code looks like this:

import os, sys, argparse
import pandas as pd 
import backtrader as bt 
from Strategien.GoldenCross import GoldenCross
import datetime
from tiingo import TiingoClient

cerebro = bt.Cerebro()
cerebro.broker.setcash(100000)

symbol = "AAPL"
path = "/Users/me/Desktop/allgemein/Visual Studio/Stock Data/S&P500 Aktien 1H/" + symbol + ".csv"
pathIndex = "/Users/me/Desktop/allgemein/Visual Studio/Stock Data/Index EOD Data/^GSPC.csv"
# read csv, parse date, use dates col as index
stock_prices = pd.read_csv(pathIndex, index_col=[0], parse_dates=[0])
#stock_prices['date'] = pd.to_datetime(stock_prices['date']) #object to datetime
#stock_prices.date = pd.to_datetime(stock_prices.date, format='%Y-%m-%dT%H:%M:%S.%fZ')

print(stock_prices.dtypes)
print(stock_prices)
# create data feed pandas df

feed = bt.feeds.PandasData(dataname=stock_prices)
# add feed
cerebro.adddata(feed)

#cerebro.addstrategy(GoldenCross)
cerebro.run()
cerebro.plot(style='candle')

If I use the pathIndex as feed, then I get plotted my chart with cerebro. But if I use the path of my tiingo data, which is an intraday data (1H data), than I get the output gives me an error called: 'numpy.int64' object has no attribute 'to_pydatetime'.
I also have to mention, that my intraday format looks like this...

dtype: object
                          date    close     high      low     open
0     2017-01-03T15:00:00.000Z  115.450  115.815  115.400  115.600
1     2017-01-03T16:00:00.000Z  115.370  115.670  115.135  115.450
2     2017-01-03T17:00:00.000Z  115.470  115.525  115.270  115.365
3     2017-01-03T18:00:00.000Z  115.235  115.495  115.235  115.475
4     2017-01-03T19:00:00.000Z  115.435  115.445  115.160  115.235
...                        ...      ...      ...      ...      ...

thank you for your help,
best regards Christian

[run-out]

Backtrader defaults to daily. Backtrader cannot tell your timeframe and compression. Try to add in the timeframe and compression in the adddata.

cerebro.adddata(feed, timeframe=bt.Timeframe.Minutes, compression=60)

timeframe (default: TimeFrame.Days)

Potential values: Ticks, Seconds, Minutes, Days, Weeks, Months and Years

compression (default: 1)

Number of actual bars per bar. Informative. Only effective in Data Resampling/Replaying.

[run-out]

@run-out I may have this wrong, you might have to use resample.

[chhrissi2909]

Hey @run-out thank you for your answer. I tried to use:

cerebro.adddata(feed, timeframe=bt.Timeframe.Minutes, compression=60)

Now the error sounds like:
"TypeError: adddata() got an unexpected keyword argument 'timeframe'"

[run-out]

@chhrissi2909 said in Daily datafeed works BUT intraday doesn't:

'numpy.int64' object has no attribute 'to_pydatetime'

Yeah, that first response was my mistake. I suspect your error is form the tingo date format going to pandas then to backtrader? I'm really not sure but maybe give this a try.

    df = pd.read_csv("data/tiingo.csv", index_col=0, parse_dates=True, infer_datetime_format=True)
    df.index.name = "datetime"
    data = bt.feeds.PandasData(dataname=df)
    cerebro.adddata(data)
    cerebro.run()

[chhrissi2909]

@run-out I tired it, but this ist more or less my old code^^.. So it does not work.. Does anyone have an other idea? I think you first thinking of resample could be a good way, but I don't know how this works.

Because you are completely right, the defaults value is for daily data (and this data also works for mer) but the intraday data doesn't..

[run-out]

@chhrissi2909 Could you paste all of your code in please including the new method and show your errors and output? Thanks.

[chhrissi2909]

@run-out here the code:

import os, sys, argparse
import pandas as pd 
import backtrader as bt 
from Strategien.GoldenCross import GoldenCross
import datetime
from tiingo import TiingoClient

cerebro = bt.Cerebro()
cerebro.broker.setcash(100000)

symbol = "AAPL"
path = "/Users/me/Desktop/allgemein/Visual Studio/Stock Data/S&P500 Aktien 1H/" + symbol + ".csv"
pathIndex = "/Users/me/Desktop/allgemein/Visual Studio/Stock Data/Index EOD Data/^GSPC.csv"

# read csv, parse date, use dates col as index
stock_prices = pd.read_csv(path, index_col=0, parse_dates=True, infer_datetime_format=True)

##print(stock_prices.dtypes)
print(stock_prices)

# create data feed pandas df

stock_prices.index.name = "datetime"
data = bt.feeds.PandasData(dataname=stock_prices)
cerebro.adddata(data)
cerebro.run()
cerebro.plot(style='candle')

[chhrissi2909]

For everybody in the Future :D (I fixed the problem to my self):

import os, sys, argparse
import pandas as pd 
import backtrader as bt 
from Strategien.GoldenCross import GoldenCross
import datetime
from tiingo import TiingoClient

cerebro = bt.Cerebro()
cerebro.broker.setcash(100000)

symbol = "AAPL"
path = "/Users/me/Desktop/allgemein/Visual Studio/Stock Data/S&P500 Aktien 1H/" + symbol + ".csv"
pathIndex = "/Users/me/Desktop/allgemein/Visual Studio/Stock Data/Index EOD Data/^GSPC.csv"

# read csv, parse date, use dates col as index
stock_prices = pd.read_csv(path, index_col=1, parse_dates=True, infer_datetime_format=True)

# create data feed pandas df

stock_prices.index.name = "datetime"
data = bt.feeds.PandasData(dataname=stock_prices)
cerebro.adddata(data)
cerebro.run()
cerebro.plot(volume=False)
cerebro.plot(style='candle')

there are 2 key lines, the first is, that for intraday data, if there is no volume you have to tell that backtrader like this:

cerebro.plot(volume=False)

And the second key line was how to read the csv:

stock_prices = pd.read_csv(path, index_col=1, parse_dates=True, infer_datetime_format=True)


Thanks for all your help :)

[DJ Deo]

@chhrissi2909 Why is there no volume in intraday ? we do get it for each candle ! ?