simple-ranker
An extremely simple, multi-value ranking module for pandas DataFrames.
Based on the algorithm from The Little Book That Beats the Market, and the backbone behind MagicRanker.
Usage
The Ranker class takes in a timeseries indexed pandas DataFrame and a list of rank and filter dicts. For example, let's say I have timeseries DataFrame that looks like this:
>>> df
code pe roe
2014-06-20 ANZ 5 0.70
2014-06-20 CBA 4 0.25
2014-06-20 NAB 10 0.14
[3 rows x 3 columns]
Perhaps I wanted to simply rank it by roe? I would do so like this:
>>> from simple_ranker import Ranker
>>> roe_rank = {'name': 'roe'}
>>> r = Ranker(df, [roe_rank])
>>> print r.process()
code pe roe total_rank
2014-06-20 ANZ 5 0.70 1
2014-06-20 CBA 4 0.25 2
2014-06-20 NAB 10 0.14 3
[3 rows x 4 columns]
Now, what if we wanted to rank them by both roe and pe. We'd do this:
>>> roe_rank = {'name': 'roe'}
>>> pe_rank = {'name': 'pe', 'ascending': True}
>>> r = Ranker(df, [roe_rank, pe_rank])
>>> print r.process()
code pe roe total_rank
2014-06-20 ANZ 5 0.70 3
2014-06-20 NAB 10 0.14 4
2014-06-20 CBA 4 0.25 5
[3 rows x 4 columns]
Note that we add the ascending key to the pe_rank dict in order to rank from lowest to highest (descending is default).
Perhaps, I have timeseries data that spanned more than one year. I could choose to average the data over a timeframe before ranking:
>>> roe_rank = {'name': 'roe', 'average': 3}
>>> pe_rank = {'name': 'pe', 'ascending': True}
>>> r = Ranker(df, [roe_rank, pe_rank])
>>> print r.process()
Or I could filter out certain values by adding a max and min key to the rank dict
>>> roe_rank = {'name': 'roe', 'average': 3, 'max': 0.70}
>>> pe_rank = {'name': 'pe', 'ascending': True, 'min': 5}
>>> r = Ranker(df, [roe_rank, pe_rank])
>>> print r.process()
Ranking algorithm
The ranking algorithm is extremely straight-forward:
- Take the first column and rank.
- Take the second column and rank.
- Create a total rank by adding the ranks from the first column to the second column.
Tests
Firstly, ensure the dev-requirements are installed (probably best to put it in a virtualenv):
> virtualenv env
> . env/bin/activate
(env)> pip install -r dev-requirements.txt
Run tests via nose, like so:
(env)> nosetests
Nose is configured to integrate with coverage (on the 100% test coverage tip).