pyndex-fin 0.1.2

Pyndex: A python package for Russell index reconstruction.

pyndex: the python reconstructor for the FTSE Russell US indexes.

What is it?

pyndex is a Python package developed to reconstruct the Russell US indexes. It is based on the results of the paper:

Evidence of Crowding on Russell 3000 Reconstitution Events

Main features

Here are the main features of pyndex:

1. Reconstruction of Russell U.S. indexes to great accuracy
2. Point in time control of index constituents
3. Calendar for Index reconstruction for each year from 1989

Where to get it

The source code is currently hosted in the following GitHub repository folder:

Binary installers for the latest released version are available at the Python package index. To install type on your terminal:

# PyPI
pip install pyndex-fin

Citation

Please use following citation to cite pyndex in scientific publications:

Bibtex entry:

@misc{aless2020evidence,
    title={Evidence of Crowding on Russell 3000 Reconstitution Events},
    author={Alessandro Micheli and Eyal Neuman},
    year={2020},
    eprint={2006.07456},
    archivePrefix={arXiv},
    primaryClass={q-fin.TR}
}

License

The software is distributed under GNU General Public License v3.0.

Usage

The package can reconstruct the Russell 1000, 2000 and 3000 index. The index to be reconstructed is passed via the argument index as "1000", "2000" and "3000", respectively. The oldest year supported is 1989.

First, start your connection to the WRDS database.

>>> import wrds
>>> db = wrds.Connection()
Loading library list...
Done

Then pass your WRDS connection to the package along with the parameters year and index.

>>> import pyndex as px
>>> index = px.Index.from_wrds(db, year = 2010, index = "3000")
>>> calendar = px.Index.get_calendar(year = 2010)

The method px.Index.from_wrds() will return a pandas MultiIndex DataFrame containing the index weights identified by permno, permco and cusip. The method px.Index.get_calendar() will return the index reconstruction calendar for the corresponding year.

One can join a sequence of year in a single DataFrame using px.join.

>>> index_2010 = px.Index.from_wrds(db, year = 2010, index = "3000")
>>> index_2011 = px.Index.from_wrds(db, year = 2011, index = "3000")
>>> new_index = px.join([index_2010,index_2011])

To check the difference of index constituents between two points in time you can use px.diff as follows,

>>> index_2010 = px.Index.from_wrds(db, year = 2010, index = "3000")
>>> slice_1 = index_2010["2010-08-20","2010-08-20"]
>>> slice_2 = index_2010["2010-09-20","2010-09-20"]
>>> additions, deletions = px.diff(slice_1, slice_2)

The first value contains the index additions from slice_1 to slice_2 while the second one contains the index deletions. This is particularly useful if one has to find the index addition between two index events, e.g. the annual rebalance and the Q3 quarterly additions.

In this case one would use px.diff as follows.

>>> index_2010 = px.Index.from_wrds(db, year = 2010, index = "3000")
>>> annual_rebalance = index_2010["2010-06-25":"2010-06-25"]
>>> q3_rebalance = index_2010["2010-09-17":"2010-09-17"]
>>> additions, deletions = px.diff(annual_rebalance, q3_rebalance)

Getting Help

For any usage or installation questions, please get in touch with Alessandro Micheli at [e-mail] .

Changelog

2-Jul-2020 : Debugged duplicated dates in Index time series.