Fix drawdown behavior

[brdsio]

Drawdown function does not work properly.
If one series had a drawdown from 2 to 1 and a drawdown from 10 to 6, pyfolio was considering the second one as the higher drawdown, but in fact the first drawdown was -50% and the second one was -40%.

This commit resolves this issue: #179

I added the tests for the both cases.

[twiecki]

Thanks @flaviodrt! Still some PEP8 issues that should be easy to fix:

pyfolio/tests/test_timeseries.py:22:1: W293 blank line contains whitespace
pyfolio/tests/test_timeseries.py:24:1: W293 blank line contains whitespace
pyfolio/tests/test_timeseries.py:28:1: W293 blank line contains whitespace
pyfolio/tests/test_timeseries.py:29:5: E304 blank lines found after function decorator
pyfolio/tests/test_timeseries.py:35:49: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:36:49: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:40:1: W293 blank line contains whitespace
pyfolio/tests/test_timeseries.py:42:1: W293 blank line contains whitespace
pyfolio/tests/test_timeseries.py:56:22: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:57:56: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:58:57: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:59:58: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:62:1: W293 blank line contains whitespace
pyfolio/tests/test_timeseries.py:64:1: W293 blank line contains whitespace
pyfolio/tests/test_timeseries.py:66:1: W293 blank line contains whitespace
pyfolio/tests/test_timeseries.py:67:74: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:69:56: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:71:58: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:73:60: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:75:1: W293 blank line contains whitespace
pyfolio/tests/test_timeseries.py:76:74: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:78:56: W291 trailing whitespace
pyfolio/tests/test_timeseries.py:80:58: W291 trailing whitespace

(you can use flake8 to check and autopep8 to autofix)

[brdsio]

Hi @twiecki!

Sorry for that.
I've sent a new commit with the fix.

[twiecki]

Thanks. Some of the failures seem unrelated and are probably due to pandas 0.18.1. However, these two are related:

======================================================================
ERROR: test_gen_drawdown_table_relative_0 (pyfolio.tests.test_timeseries.TestDrawdown)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/travis/miniconda/envs/testenv/lib/python2.7/site-packages/nose_parameterized/parameterized.py", line 365, in standalone_func
    return func(*(a + p.args), **p.kwargs)
  File "/home/travis/build/quantopian/pyfolio/pyfolio/tests/test_timeseries.py", line 64, in test_gen_drawdown_table_relative
    drawdowns = timeseries.gen_drawdown_table(rets, top=2)
  File "/home/travis/build/quantopian/pyfolio/pyfolio/timeseries.py", line 1072, in gen_drawdown_table
    unit='D')
  File "/home/travis/miniconda/envs/testenv/lib/python2.7/site-packages/pandas/util/decorators.py", line 91, in wrapper
    return func(*args, **kwargs)
  File "/home/travis/miniconda/envs/testenv/lib/python2.7/site-packages/pandas/tseries/tools.py", line 291, in to_datetime
    unit=unit, infer_datetime_format=infer_datetime_format)
  File "/home/travis/miniconda/envs/testenv/lib/python2.7/site-packages/pandas/tseries/tools.py", line 420, in _to_datetime
    values = _convert_listlike(arg._values, False, format)
  File "/home/travis/miniconda/envs/testenv/lib/python2.7/site-packages/pandas/tseries/tools.py", line 331, in _convert_listlike
    errors=errors)
  File "pandas/tslib.pyx", line 1981, in pandas.tslib.array_with_unit_to_datetime (pandas/tslib.c:38273)
  File "pandas/tslib.pyx", line 2071, in pandas.tslib.array_with_unit_to_datetime (pandas/tslib.c:37194)
ValueError: non convertible value 2000-01-08with the unit 'D'
======================================================================
ERROR: test_get_max_drawdown_begins_first_day_0 (pyfolio.tests.test_timeseries.TestDrawdown)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/travis/miniconda/envs/testenv/lib/python2.7/site-packages/nose_parameterized/parameterized.py", line 365, in standalone_func
    return func(*(a + p.args), **p.kwargs)
  File "/home/travis/build/quantopian/pyfolio/pyfolio/tests/test_timeseries.py", line 30, in test_get_max_drawdown_begins_first_day
    drawdowns = timeseries.gen_drawdown_table(rets, top=1)
  File "/home/travis/build/quantopian/pyfolio/pyfolio/timeseries.py", line 1072, in gen_drawdown_table
    unit='D')
  File "/home/travis/miniconda/envs/testenv/lib/python2.7/site-packages/pandas/util/decorators.py", line 91, in wrapper
    return func(*args, **kwargs)
  File "/home/travis/miniconda/envs/testenv/lib/python2.7/site-packages/pandas/tseries/tools.py", line 291, in to_datetime
    unit=unit, infer_datetime_format=infer_datetime_format)
  File "/home/travis/miniconda/envs/testenv/lib/python2.7/site-packages/pandas/tseries/tools.py", line 420, in _to_datetime
    values = _convert_listlike(arg._values, False, format)
  File "/home/travis/miniconda/envs/testenv/lib/python2.7/site-packages/pandas/tseries/tools.py", line 331, in _convert_listlike
    errors=errors)
  File "pandas/tslib.pyx", line 1981, in pandas.tslib.array_with_unit_to_datetime (pandas/tslib.c:38273)
  File "pandas/tslib.pyx", line 2071, in pandas.tslib.array_with_unit_to_datetime (pandas/tslib.c:37194)
ValueError: non convertible value 2000-01-03with the unit 'D'

[brdsio]

I think that were related to pandas 0.18.1 because the other tests in previous commits before my changes were failing too.
Removing the attribute unit from pd.to_datetime does not raise the exception.
Do you think that is a good solution?

[twiecki]

Yep, that looks great. Thanks for the fix!