GenericCSVdata timeframe

[bodober]

Hello,

I am trying to import some datafeeds from CSV files, which have one line per day. If I set the timeframe of the class to Days, as it should be, the strategy only goes through the next() method once. Weirdly enough, if I set the timeframe to Weeks instead, the strategy runs once per day, although then I get issues with the resampling.

Any hint where this weird behaviour could come from?

First lines of an example CSV data file:

date,open,close,high,low,turnoverPieces,turnoverEuro
2022-08-10,4.012,3.976,4.012,3.942,50311.0,199699.04
2022-08-09,4.022,4.008,4.09,3.994,94701.0,381702.55
2022-08-08,4.006,4.046,4.074,3.95,167475.0,671826.34
2022-08-05,3.998,3.97,3.998,3.89,93959.0,370817.28
2022-08-04,3.898,3.98,4.0,3.894,143614.0,569000.31
2022-08-03,3.868,3.896,3.896,3.82,162676.0,626748.42
2022-08-02,3.884,3.822,3.884,3.776,110984.0,423815.62
2022-08-01,3.848,3.842,3.902,3.826,126052.0,486242.22

My class:

class my_custom_csv(bt.feeds.GenericCSVData):
    params = (
        ('headers', True),
        ('separator', ','),
                    
        ('nullvalue', 0.0),
    
        ('dtformat', ('%Y-%m-%d')),
        
        ('time', -1),
        ('datetime', 0),
        ('high', 3),
        ('low', 4),
        ('open', 1),
        ('close', 2),
        ('volume', 5),
        ('openinterest', -1),
        
        ('timeframe', bt.TimeFrame.Days), #this is the parameters that can be set to weeks to make the code work
        )
        

And this is how I add and resample the data:

days = 365# how many days to backtest

start = datetime.date.today() - datetime.timedelta(days=days) 
end = datetime.date.today() 

data = my_custom_csv(dataname = '.\History\DE0005493092.txt', fromdate = start, todate = end)
            
data._name = symbol
data._dataname = [symbol]

cerebro_instance.adddata(data)    # Give the data to cerebro
rdata = cerebro_instance.resampledata(data, timeframe=bt.TimeFrame.Weeks) #this is where I resample to weeks my daily data
rdata.plotinfo.plot = False