Skip to content

Latest commit

 

History

History
329 lines (268 loc) · 41.2 KB

README.md

File metadata and controls

329 lines (268 loc) · 41.2 KB
Raw

Captor Fund Management AB



openseries

PyPI version Conda Version Conda platforms Python version GitHub Action Test Suite Coverage Poetry Ruff License

This is a project with tools to analyze financial timeseries of a single asset or a group of assets. It is solely made for daily or less frequent data.

Basic Usage

To install:

pip install openseries

or:

conda install -c conda-forge openseries

An example of how to make use of the OpenTimeSeries is shown below. The design aligns with how we within our fund company's code base have a subclass of the OpenTimeSeries with class methods for our different data sources. Combined with some additional tools it allows us to efficiently present investment cases to clients.

The code snippet can be pasted into a Python console to run it. Install openseries and yfinance first.

from openseries import OpenTimeSeries
import yfinance as yf

move=yf.Ticker("^MOVE")
history=move.history(period="max")
series=OpenTimeSeries.from_df(history.loc[:, "Close"])
_=series.set_new_label("ICE BofAML MOVE Index")
_,_=series.plot_series()

Sample output using the OpenFrame.all_properties() method:

                       Scilla Global Equity C (simulation fund) Global Low Volatility index, SEK
                                                ValueType.PRICE                  ValueType.PRICE
Total return                                           3.641282                         1.946319
Arithmetic return                                      0.096271                         0.069636
Geometric return                                       0.093057                          0.06464
Volatility                                             0.120279                         0.117866
Return vol ratio                                       0.800396                          0.59081
Downside deviation                                     0.085956                         0.086723
Sortino ratio                                          1.119993                         0.802975
Positive share                                         0.541783                         0.551996
Worst                                                 -0.071616                        -0.089415
Worst month                                           -0.122503                        -0.154485
Max drawdown                                          -0.309849                        -0.435444
Max drawdown in cal yr                                -0.309849                        -0.348681
Max drawdown dates                                   2020-03-23                       2009-03-09
CVaR 95.0%                                             -0.01793                        -0.018429
VaR 95.0%                                             -0.011365                        -0.010807
Imp vol from VaR 95%                                   0.109204                         0.103834
Z-score                                                0.587905                         0.103241
Skew                                                  -0.650782                        -0.888109
Kurtosis                                               8.511166                        17.527367
observations                                               4309                             4309
span of days                                               6301                             6301
first indices                                        2006-01-03                       2006-01-03
last indices                                         2023-04-05                       2023-04-05

Usage example on Jupyter Nbviewer

nbviewer

Development Instructions

These instructions assume that you have a compatible Python version installed on your machine and that you are OK to install this project in a virtual environment.

The OpenTimeSeries and OpenFrame classes are both subclasses of the Pydantic BaseModel. Please refer to its documentation for information on any attributes or methods inherited from this model.

Windows Powershell

git clone https://github.com/CaptorAB/openseries.git
cd openseries
./make.ps1 make

Mac Terminal/Linux

git clone https://github.com/CaptorAB/openseries.git
cd openseries
make
source source_me
make install

Testing and Linting / Type-checking

Ruff and Mypy checking is embedded in the pre-commit hook. Both are also used in the project's GitHub workflows and are run when the lint alternative is chosen in the below commands. Any silenced error codes can be found in the pyproject.toml file or in in-line comments.

Windows Powershell

./make.ps1 test
./make.ps1 lint

Mac Terminal/Linux

make test
make lint

Table of Contents

On some files in the project

File Description
series.py Defines the class OpenTimeSeries for managing and analyzing a single timeseries. The module also defines a function timeseries_chain that can be used to chain two timeseries objects together.
frame.py Defines the class OpenFrame for managing a group of timeseries, and e.g. calculate a portfolio timeseries from a rebalancing strategy between timeseries. The module also defines functions to simulate, optimize, and plot portfolios.
simulation.py Defines the class ReturnSimulation to create simulated financial timeseries. Used in the project's test suite

Class methods used to construct objects.

Method Applies to Description
from_arrays OpenTimeSeries Class method to create an OpenTimeSeries object from a list of date strings and a list of values.
from_df OpenTimeSeries Class method to create an OpenTimeSeries object from a pandas.DataFrame or pandas.Series.
from_fixed_rate OpenTimeSeries Class method to create an OpenTimeSeries object from a fixed rate, number of days and an end date.
from_deepcopy OpenTimeSeries, OpenFrame Creates a copy of an OpenTimeSeries object.

Non-numerical or "helper" properties that apply only to the OpenTimeSeries class.

Property type Applies to Description
timeseries_id str OpenTimeSeries Placeholder for database identifier for the timeseries. Can be left as empty string.
instrument_id str OpenTimeSeries Placeholder for database identifier for the instrument associated with the timeseries. Can be left as empty string.
dates list[str] OpenTimeSeries Dates of the timeseries. Not edited by any method to allow reversion to original.
values list[float] OpenTimeSeries Values of the timeseries. Not edited by any method to allow reversion to original.
currency str OpenTimeSeries Currency of the timeseries. Only used if conversion/hedging methods are added.
domestic str OpenTimeSeries Domestic currency of the user / investor. Only used if conversion/hedging methods are added.
local_ccy bool OpenTimeSeries Indicates if series should be in its local currency or the domestic currency of the user. Only used if conversion/hedging methods are added.
name str OpenTimeSeries An identifier field.
isin str OpenTimeSeries ISIN code of the associated instrument. If any.
label str OpenTimeSeries Field used in outputs. Derived from name as default.
countries list or str OpenTimeSeries (List of) country code(s) according to ISO 3166-1 alpha-2 used to generate business days.
valuetype ValueType OpenTimeSeries Field identifies the type of values in the series. ValueType is an Enum.

Non-numerical or "helper" properties that apply only to the OpenFrame class.

Property type Applies to Description
constituents list[OpenTimeSeries] OpenFrame A list of the OpenTimeSeries that make up an OpenFrame.
columns_lvl_zero list OpenFrame A list of the level zero column names in the OpenFrame pandas.DataFrame.
columns_lvl_one list OpenFrame A list of the level one column names in the OpenFrame pandas.DataFrame.
item_count int OpenFrame Number of columns in the OpenFrame pandas.DataFrame.
weights list[float] OpenFrame Weights used in the method make_portfolio.
first_indices pandas.Series OpenFrame First dates of all the series in the OpenFrame.
last_indices pandas.Series OpenFrame Last dates of all the series in the OpenFrame.
lengths_of_items pandas.Series OpenFrame Number of items in each of the series in the OpenFrame.
span_of_days_all pandas.Series OpenFrame Number of days from the first to the last in each of the series.

Non-numerical or "helper" properties that apply to both the OpenTimeSeries and the OpenFrame class.

Property type Applies to Description
first_idx datetime.date OpenTimeSeries, OpenFrame First date of the series.
last_idx datetime.date OpenTimeSeries, OpenFrame Last date of the series.
length int OpenTimeSeries, OpenFrame Number of items in the series.
span_of_days int OpenTimeSeries, OpenFrame Number of days from the first to the last date in the series.
tsdf pandas.DataFrame OpenTimeSeries, OpenFrame The Pandas DataFrame which gets edited by the class methods.
max_drawdown_date datetime.date, pandas.Series OpenTimeSeries, OpenFrame Date when the maximum drawdown occurred.
periods_in_a_year float OpenTimeSeries, OpenFrame The number of observations in an average year for all days in the data.
yearfrac float OpenTimeSeries, OpenFrame Length of timeseries expressed as np.float64 fraction of a year with 365.25 days.

Methods that apply only to the OpenTimeSeries class.

Method Applies to Description
pandas_df OpenTimeSeries Method to create the tsdf pandas.DataFrame from the dates and values.
set_new_label OpenTimeSeries Method to change the pandas.DataFrame column MultiIndex.
running_adjustment OpenTimeSeries Adjusts the series performance with a float factor.
ewma_vol_func OpenTimeSeries Returns a pandas.Series with volatility based on Exponentially Weighted Moving Average.
from_1d_rate_to_cumret OpenTimeSeries Converts a series of 1-day rates into a cumulative valueseries. 
                                                                       |

Methods that apply only to the OpenFrame class.

Method Applies to Description
merge_series OpenFrame Merges the Pandas Dataframes of the constituent OpenTimeSeries.
trunc_frame OpenFrame Truncates the OpenFrame to a common period.
add_timeseries OpenFrame Adds a given OpenTimeSeries to the OpenFrame.
delete_timeseries OpenFrame Deletes an OpenTimeSeries from the OpenFrame.
relative OpenFrame Calculates a new series that is the relative performance of two others.
make_portfolio OpenFrame Calculates a portfolio timeseries based on the series and weights. Weights can be provided as a list, or a weight strategy can be set as equal weights or inverted volatility.
ord_least_squares_fit OpenFrame Performs a regression and an Ordinary Least Squares fit.
beta OpenFrame Calculates Beta of an asset relative a market.
jensen_alpha OpenFrame Calculates Jensen's Alpha of an asset relative a market.
tracking_error_func OpenFrame Calculates the tracking errors relative to a selected series in the OpenFrame.
info_ratio_func OpenFrame Calculates the information ratios relative to a selected series in the OpenFrame.
capture_ratio_func OpenFrame Calculates up, down and up/down capture ratios relative to a selected series.
rolling_info_ratio OpenFrame Returns a pandas.DataFrame with the rolling information ratio between two series.
rolling_beta OpenFrame Returns a pandas.DataFrame with the rolling Beta of an asset relative a market.
rolling_corr OpenFrame Calculates and adds a series of rolling correlations between two other series.
correl_matrix OpenFrame Returns a pandas.DataFrame with a correlation matrix.
ewma_risk OpenFrame Returns a pandas.DataFrame with volatility and correlation based on Exponentially Weighted Moving Average.

Methods that apply to both the OpenTimeSeries and the OpenFrame class.

Method Applies to Description
align_index_to_local_cdays OpenTimeSeries, OpenFrame Aligns the series dates to a business calendar. Defaults to Sweden.
resample OpenTimeSeries, OpenFrame Resamples the series to a specific frequency.
resample_to_business_period_ends OpenTimeSeries, OpenFrame Resamples the series to month-end dates with monthly, quarterly or annual frequency.
value_nan_handle OpenTimeSeries, OpenFrame Fills Nan in a value series with the preceding non-Nan value.
return_nan_handle OpenTimeSeries, OpenFrame Replaces Nan in a return series with a 0.0 float.
to_cumret OpenTimeSeries, OpenFrame Converts a return series into a value series and/or resets a value series to be rebased from 1.0.
to_json OpenTimeSeries, OpenFrame Method to export object data to a json file.
to_xlsx OpenTimeSeries, OpenFrame Method to save the data in the .tsdf DataFrame to an Excel file.
value_to_ret OpenTimeSeries, OpenFrame Converts a value series into a percentage return series.
value_to_diff OpenTimeSeries, OpenFrame Converts a value series into a series of differences.
value_to_log OpenTimeSeries, OpenFrame Converts a value series into a logarithmic return series.
value_ret_calendar_period OpenTimeSeries, OpenFrame Returns the series simple return for a specific calendar period.
plot_series OpenTimeSeries, OpenFrame Opens a HTML Plotly Scatter plot of the series in a browser window.
plot_bars OpenTimeSeries, OpenFrame Opens a HTML Plotly Bar plot of the series in a browser window.
to_drawdown_series OpenTimeSeries, OpenFrame Converts the series into drawdown series.
rolling_return OpenTimeSeries, OpenFrame Returns a pandas.DataFrame with rolling returns.
rolling_vol OpenTimeSeries, OpenFrame Returns a pandas.DataFrame with rolling volatilities.
rolling_var_down OpenTimeSeries, OpenFrame Returns a pandas.DataFrame with rolling VaR figures.
rolling_cvar_down OpenTimeSeries, OpenFrame Returns a pandas.DataFrame with rolling CVaR figures.
calc_range OpenTimeSeries, OpenFrame Returns the start and end dates of a range from specific period definitions. Used by the below numerical methods and not meant to be used independently.

Numerical properties available for individual OpenTimeSeries or on all series in an OpenFrame.

Property type Applies to Description
all_properties pandas.DataFrame OpenTimeSeries, OpenFrame Returns most of the properties in one go.
arithmetic_ret float, pandas.Series OpenTimeSeries, OpenFrame Annualized arithmetic mean of returns.
geo_ret float, pandas.Series OpenTimeSeries, OpenFrame Compound Annual Growth Rate(CAGR), a specific implementation of geometric mean.
value_ret float, pandas.Series OpenTimeSeries, OpenFrame Simple return from first to last observation.
vol float, pandas.Series OpenTimeSeries, OpenFrame Annualized volatility. Pandas .std() is the equivalent of stdev.s([...]) in MS excel.
downside_deviation float, pandas.Series OpenTimeSeries, OpenFrame Downside deviation is the volatility of all negative return observations. Minimum Accepted Return (MAR) set to zero.
ret_vol_ratio float, pandas.Series OpenTimeSeries, OpenFrame Ratio of arithmetic mean return and annualized volatility. It is the Sharpe Ratio with the riskfree rate set to zero.
sortino_ratio float, pandas.Series OpenTimeSeries, OpenFrame The Sortino Ratio is the arithmetic mean return divided by the downside deviation. This attribute assumes that the riskfree rate and the MAR are both zero.
omega_ratio float, pandas.Series OpenTimeSeries, OpenFrame The Omega Ratio compares returns above a certain target level (MAR) to the total downside risk below MAR.
var_down float, pandas.Series OpenTimeSeries, OpenFrame Downside 95% Value At Risk, "VaR". The equivalent of percentile.inc([...], 1-level) over returns in MS Excel. For other confidence levels use the corresponding method.
cvar_down float, pandas.Series OpenTimeSeries, OpenFrame Downside 95% Conditional Value At Risk, "CVaR". For other confidence levels use the corresponding method.
worst float, pandas.Series OpenTimeSeries, OpenFrame Most negative percentage change of a single observation.
worst_month float, pandas.Series OpenTimeSeries, OpenFrame Most negative month.
max_drawdown float, pandas.Series OpenTimeSeries, OpenFrame Maximum drawdown.
max_drawdown_cal_year float, pandas.Series OpenTimeSeries, OpenFrame Max drawdown in a single calendar year.
positive_share float, pandas.Series OpenTimeSeries, OpenFrame The share of percentage changes that are positive.
vol_from_var float, pandas.Series OpenTimeSeries, OpenFrame Implied annualized volatility from the Downside VaR using the assumption that returns are normally distributed.
skew float, pandas.Series OpenTimeSeries, OpenFrame Skew of the return distribution.
kurtosis float, pandas.Series OpenTimeSeries, OpenFrame Kurtosis of the return distribution.
z_score float, pandas.Series OpenTimeSeries, OpenFrame Z-score as (last return - mean return) / standard deviation of returns.

Methods below are identical to the Numerical Properties above.

They are simply methods that take different date or length inputs to return the properties for subset periods.

Method type Applies to Description
arithmetic_ret_func float, pandas.Series OpenTimeSeries, OpenFrame Annualized arithmetic mean of returns.
geo_ret_func float, pandas.Series OpenTimeSeries, OpenFrame Compound Annual Growth Rate(CAGR), a specific implementation of geometric mean.
value_ret_func float, pandas.Series OpenTimeSeries, OpenFrame Simple return from first to last observation.
vol_func float, pandas.Series OpenTimeSeries, OpenFrame Annualized volatility. Pandas .std() is the equivalent of stdev.s([...]) in MS excel.
downside_deviation_func float, pandas.Series OpenTimeSeries, OpenFrame Downside deviation is the volatility of all negative return observations. MAR and riskfree rate can be set.
ret_vol_ratio_func float, pandas.Series OpenTimeSeries, OpenFrame Ratio of arithmetic mean return and annualized volatility. It is the Sharpe Ratio with the riskfree rate set to zero. A riskfree rate can be set as a float or a series chosen for the frame function.
sortino_ratio_func float, pandas.Series OpenTimeSeries, OpenFrame The Sortino Ratio is the arithmetic mean return divided by the downside deviation. A riskfree rate can be set as a float or a series chosen for the frame function. MAR is set to zero.
omega_ratio_func float, pandas.Series OpenTimeSeries, OpenFrame The Omega Ratio compares returns above a certain target level (MAR) to the total downside risk below MAR.
var_down_func float, pandas.Series OpenTimeSeries, OpenFrame Downside 95% Value At Risk, "VaR". The equivalent of percentile.inc([...], 1-level) over returns in MS Excel. Default is 95% confidence level.
cvar_down_func float, pandas.Series OpenTimeSeries, OpenFrame Downside 95% Conditional Value At Risk, "CVaR". Default is 95% confidence level.
worst_func float, pandas.Series OpenTimeSeries, OpenFrame Most negative percentage change for a given number of observations (default=1).
max_drawdown_func float, pandas.Series OpenTimeSeries, OpenFrame Maximum drawdown.
positive_share_func float, pandas.Series OpenTimeSeries, OpenFrame The share of percentage changes that are positive.
vol_from_var_func float, pandas.Series OpenTimeSeries, OpenFrame Implied annualized volatility from the Downside VaR using the assumption that returns are normally distributed.
skew_func float, pandas.Series OpenTimeSeries, OpenFrame Skew of the return distribution.
kurtosis_func float, pandas.Series OpenTimeSeries, OpenFrame Kurtosis of the return distribution.
z_score_func float, pandas.Series OpenTimeSeries, OpenFrame Z-score as (last return - mean return) / standard deviation of returns.
target_weight_from_var float, pandas.Series OpenTimeSeries, OpenFrame A position target weight from the ratio between a VaR implied volatility and a given target volatility.