oandav20 resampling from 15m to 4h shows extra bars.

ahoyahoy

I'm using oandav20 and resampling from 15m up to 4h and the 1D. The problem is in 4h chart data I see non-aligned bars and it looks like the time axis is broken:

eur_usd_EUR_USD_4h copy.png

I've tried various combinations of the bar2edge and rightedge and adjbartime flags but still has the problem. I've also tried shifting the sessionstart but the daily chart needs this to be at 23 (aligning with broker time).

When I search through the data dump (using the writer) I see extra bars coming out of the resampling aligning to the yellow arrows:

2020-05-14 00:00:00
2020-05-14 04:00:00
2020-05-14 08:00:00
2020-05-14 12:00:00
2020-05-14 13:45:00. <-- what?
2020-05-14 16:00:00
2020-05-14 20:00:00
...

2020-05-19 04:00:00
2020-05-19 08:00:00
2020-05-19 11:15:00 <-- what?
2020-05-19 12:00:00

any pointer appreciated, I've been a coder for 15 years so hit me with the hardcore detail if needed

Here's my setup, if this is of help. Cheers

datakwargs = dict(
    timeframe=bt.TimeFrame.Minutes,
    compression=15,
    stream_timeout=60,
    historical=args.historical,
    fromdate=fromdate,
    tz=args.timezone,
    backfill_start=True,
    backfill=True,
    sessionstart=datetime.time(23, 00), #oanda uses singapore time so 23 in BST.
    sessionend=datetime.time(23, 00)
)

rekwargs4h = dict(
    timeframe=bt.TimeFrame.Minutes, compression=240,
    bar2edge=False, 
    adjbartime=False,
    rightedge=False,
    takelate=False
)

rekwargs1d = dict(
    timeframe=bt.TimeFrame.Days, compression=1,
    bar2edge=False, 
    adjbartime=False, 
    rightedge=False,
    takelate=False
)

#Loop through the list adding to cerebro.
for i in range(len(datalist)):
    data = DataFactory(dataname=datalist[i], **datakwargs)
    # enabling this seems to break daily graph
    cerebro.resampledata(data, name=datalist[i], timeframe=datakwargs['timeframe'], compression=datakwargs['compression'])
        
    cerebro.resampledata(data, name=datalist[i], **rekwargs4h)
    
    # day
    cerebro.resampledata(data, name=datalist[i], **rekwargs1d)
   
config = dict()
``

ahoyahoy

ok I've moved this to the right topic, admins feel free to delete

dasch

this may be due to the sessionend value and the first time you receive a data update. but this is just a guess. If this does not appear without the session end value, investigate this.

amulya agarwal

Can you provide the entire source code or a link to it?

dasch

These two settings:

sessionstart=datetime.time(23, 00), #oanda uses singapore time so 23 in BST.
sessionend=datetime.time(23, 00)

amulya agarwal

@dasch How are we supposed to get data in different timeframe like 30mins or 1h rather than 1d which is a default.

bigdavediode

@amulya-agarwal

Use compression, as above, compression 1 for 1 minute bars if you wish, as below:

        timeframe=bt.TimeFrame.Minutes,
        compression=barsize,
        useRTH=False,
        #outsideRth=True,    fill orders outside rth -- use for orders, not here...
        sessionstart=datetime.time(9, 30, 0),
        sessionend=datetime.time(16, 30, 0),

...and then your adddata or resample line using these arguments.