ValueError: ['End'] not contained in the index

[brunocapelao]

Hello,

I'm using the lib with a dataset like this:

and I have the following error:

C:\Users\capel\Anaconda3\lib\site-packages\backtesting\_util.py:32: FutureWarning: Converting timezone-aware DatetimeArray to timezone-naive ndarray with 'datetime64[ns]' dtype. In the future, this will return an ndarray with 'object' dtype where each element is a 'pandas.Timestamp' with the correct 'tz'.
	To accept the future behavior, pass 'dtype=object'.
	To keep the old behavior, pass 'dtype="datetime64[ns]"'.
  obj = np.asarray(array).view(cls)
C:\Users\capel\Anaconda3\lib\site-packages\backtesting\_util.py:23: FutureWarning: The default of the 'keep_tz' keyword will change to True in a future release. You can set 'keep_tz=True' to obtain the future behaviour and silence this warning.
  return df.index[:100].to_series().diff().median()
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-7-b0362348a33a> in <module>
      1 backtest = Backtest(dados, BarStrategy, commission=.0075, cash=100)
----> 2 backtest.run()

~\Anaconda3\lib\site-packages\backtesting\backtesting.py in run(self, **kwargs)
    703                 strategy.next()
    704 
--> 705         self._results = self._compute_stats(broker, strategy)
    706         return self._results
    707 

~\Anaconda3\lib\site-packages\backtesting\backtesting.py in _compute_stats(self, broker, strategy)
    870         s = pd.Series()
    871         s['Start'] = df.index[0]
--> 872         s['End'] = df.index[-1]
    873         # Assigning Timedeltas needs the key to exist beforehand,
    874         # otherwise the value is interpreted as nanosec *int*. See:

~\Anaconda3\lib\site-packages\pandas\core\series.py in __setitem__(self, key, value)
   1037         # do the setitem
   1038         cacher_needs_updating = self._check_is_chained_assignment_possible()
-> 1039         setitem(key, value)
   1040         if cacher_needs_updating:
   1041             self._maybe_update_cacher()

~\Anaconda3\lib\site-packages\pandas\core\series.py in setitem(key, value)
   1033                     pass
   1034 
-> 1035             self._set_with(key, value)
   1036 
   1037         # do the setitem

~\Anaconda3\lib\site-packages\pandas\core\series.py in _set_with(self, key, value)
   1083                 self._set_values(key.astype(np.bool_), value)
   1084             else:
-> 1085                 self._set_labels(key, value)
   1086 
   1087     def _set_labels(self, key, value):

~\Anaconda3\lib\site-packages\pandas\core\series.py in _set_labels(self, key, value)
   1093         mask = indexer == -1
   1094         if mask.any():
-> 1095             raise ValueError('%s not contained in the index' % str(key[mask]))
   1096         self._set_values(indexer, value)
   1097 

ValueError: ['End'] not contained in the index

I suggest change this code to use the .loc method expressly, like this:

 871         s.loc['Start'] = df.index[0]
 872         s.loc['End'] = df.index[-1]

[kernc]

Please update to backtesting 0.1.2 released yesterday. This should work.

pip install -U backtesting

[brunocapelao]

This does not work either.

[kernc]

Thanks. Any way I could get ahold of some of your data that reproduces this?

[brunocapelao]

I tested on Linux and it works.

[kernc]

But it still doesn't work on Windos or what? 😕

An explicit .loc sure can't hurt. Care to open a PR for it?

[marvinbelfort]

Found same bug. Apparently is some problem with timedelta64 having an UTC info attached.

s['Start'] = df.index[0]
# FIXME: works for df.index having dtype: datetime64[ns] but not for dtype: datetime64[ns, UTC].
s['End'] = df.index[-1]	

if you do remove timezones from your dataframe, it will work:

mydataframe.index = mydataframe.index.tz_localize(None)

It is a hack, but will have you going until a better solution is found.