Heerozh
commented
4 years ago CsvDirLoader was not designed to cover all situations when designing (too difficult), you can refer to QuandlLoader (It has fewer lines) to rewrite one for your needs.
Answer your questions:
My broker provides me data feeds named (for instance) AAPL.txt (read: not .csv); can CsvDirLoader access those (without having to rename/link them?)
No, you need rename to *.csv. I will add a parameter in the future.
My broker "csv" data doesn't have a csv header, how can I tell CsvDirLoader which column is what?
This can be done by passing the pandas.read_csv parameters, as an example:
loader = CsvDirLoader('path_to_files', parse_dates=True, prices_index='date', header=None, names=['date','open','high','low','close','volume'])
print(loader._load())| date | asset | open | high | low | close | volume | _time_cat_id |
|---|---|---|---|---|---|---|---|
| 12/31/1980 | TEST | 0.4891 | 0.4891 | 0.4873 | 0.4873 | 11177100 | 0 |
| 12/31/1980 | TEST | 0.4926 | 0.4962 | 0.4926 | 0.4926 | 6772100 | 1 |
| 12/31/1980 | TEST | 0.4837 | 0.4837 | 0.4819 | 0.4819 | 11170100 | 2 |
Does spectre supports intraday data
Yes. But CsvDirLoader needs time and date in one column. For your data, you may have to rewrite a DataLoader.
jibanes
My broker provides me data feeds named (for instance) AAPL.txt (read: not .csv); can CsvDirLoader access those (without having to rename/link them?)
My broker "csv" data doesn't have a csv header, how can I tell CsvDirLoader which column is what? (note: the data is adjusted, there is no need for split adjustments)
My broker formats daily historical data as:
Do you have an example of read_csv arguments to pass to CsvDirLoader to parse this date format, or is parse_dates=True sufficient here?
Does spectre supports intraday data (aka minute-data here), and how should it be loaded?
Thank you.