finagg: Financial Aggregation for Python

finagg is a Python package that provides implementations of popular and free financial APIs, tools for aggregating historical data from those APIs into SQL databases, and tools for transforming aggregated data into features useful for analysis and AI/ML.

• Documentation: https://theogognf.github.io/finagg/
• PyPI: https://pypi.org/project/finagg/
• Repository: https://github.com/theOGognf/finagg

Quick Start

Installation

Install with pip for the latest stable version.

pip install finagg

Install from GitHub for the latest unstable version.

git clone https://github.com/theOGognf/finagg.git
pip install ./finagg/

Optionally install the recommended datasets (economic data, company financials, stock histories, etc.) from 3rd party APIs into a local SQL database.

finagg install -ss economic -ts indices -z -r

The installation will point you where to get free API keys for each API that requires one and will write those API keys to a local .env file for storage. Run finagg install --help for more installation options and details.

Basic Usage

These are just finagg usage samples. See the documentation for all the supported APIs and features.

Explore the APIs directly

These methods require internet access and API keys/user agent declarations.

Get Bureau of Economic Analysis (BEA) data.

>>> finagg.bea.api.gdp_by_industry.get(year=[2019]).head(5)
   table_id freq  year quarter industry                         industry_description ...
0         1    Q  2019       1       11  Agriculture, forestry, fishing, and hunting ...
1         1    Q  2019       1    111CA                                        Farms ...
2         1    Q  2019       1    113FF    Forestry, fishing, and related activities ...
3         1    Q  2019       1       21                                       Mining ...
4         1    Q  2019       1      211                       Oil and gas extraction ...

Get Federal Reserve Economic Data (FRED).

>>> finagg.fred.api.series.observations.get(
...   "CPIAUCNS",
...   realtime_start=0,
...   realtime_end=-1,
...   output_type=4
... ).head(5)
  realtime_start realtime_end        date  value series_id
0     1949-04-22   1953-02-26  1949-03-01  169.5  CPIAUCNS
1     1949-05-23   1953-02-26  1949-04-01  169.7  CPIAUCNS
2     1949-06-24   1953-02-26  1949-05-01  169.2  CPIAUCNS
3     1949-07-22   1953-02-26  1949-06-01  169.6  CPIAUCNS
4     1949-08-26   1953-02-26  1949-07-01  168.5  CPIAUCNS

Get Securities and Exchange Commission (SEC) filings.

>>> finagg.sec.api.company_facts.get(ticker="AAPL").head(5)
          end        value                  accn    fy  fp    form       filed ...
0  2009-06-27  895816758.0  0001193125-09-153165  2009  Q3    10-Q  2009-07-22 ...
1  2009-10-16  900678473.0  0001193125-09-214859  2009  FY    10-K  2009-10-27 ...
2  2009-10-16  900678473.0  0001193125-10-012091  2009  FY  10-K/A  2010-01-25 ...
3  2010-01-15  906794589.0  0001193125-10-012085  2010  Q1    10-Q  2010-01-25 ...
4  2010-04-09  909938383.0  0001193125-10-088957  2010  Q2    10-Q  2010-04-21 ...

Use installed raw data for exploring the most popular features

These methods require internet access, API keys/user agent declarations, and downloading and installing raw data through the finagg install or finagg <api/subpackage> install commands.

Get the most popular FRED features all in one dataframe.

>>> finagg.fred.feat.economic.from_raw().head(5)
            CIVPART  LOG_CHANGE(CPIAUCNS)  LOG_CHANGE(CSUSHPINSA)  FEDFUNDS ...
date                                                                        ...
2014-10-06     62.8                   0.0                     0.0      0.09 ...
2014-10-08     62.8                   0.0                     0.0      0.09 ...
2014-10-13     62.8                   0.0                     0.0      0.09 ...
2014-10-15     62.8                   0.0                     0.0      0.09 ...
2014-10-20     62.8                   0.0                     0.0      0.09 ...

Get quarterly report features from SEC data.

>>> finagg.sec.feat.quarterly.from_raw("AAPL").head(5)
                    LOG_CHANGE(Assets)  LOG_CHANGE(AssetsCurrent) ...
fy   fp filed                                                     ...
2010 Q1 2010-01-25            0.182629                  -0.023676 ...
     Q2 2010-04-21            0.000000                   0.000000 ...
     Q3 2010-07-21            0.000000                   0.000000 ...
2011 Q1 2011-01-19            0.459174                   0.278241 ...
     Q2 2011-04-21            0.000000                   0.000000 ...

Get an aggregation of quarterly and daily features for a particular ticker.

>>> finagg.fundam.feat.fundam.from_raw("AAPL").head(5)
            PriceBookRatio  PriceEarningsRatio
date
2010-01-25        0.175061            2.423509
2010-01-26        0.178035            2.464678
2010-01-27        0.178813            2.475448
2010-01-28        0.177154            2.452471
2010-01-29        0.173825            2.406396

Use installed features for exploring refined aggregations of raw data

These methods require installing refined data through the finagg install or finagg <api/subpackage> install commands.

Get a ticker's industry's averaged quarterly report features.

>>> finagg.sec.feat.quarterly.industry.from_refined(ticker="AAPL").head(5)
                                 mean                           ...
name               AssetCoverageRatio BookRatio DebtEquityRatio ...
fy   fp filed                                                   ...
2014 Q1 2014-05-15          10.731301  9.448954        0.158318 ...
     Q2 2014-08-14          10.731301  9.448954        0.158318 ...
     Q3 2014-11-14          10.731301  9.448954        0.158318 ...
2015 Q1 2015-05-15          16.738972  9.269250        0.294238 ...
     Q2 2015-08-13          16.738972  9.269250        0.294238 ...

Get a ticker's industry-averaged quarterly report features.

>>> finagg.sec.feat.quarterly.normalized.from_refined("AAPL").head(5)
                    NORM(LOG_CHANGE(Assets))  NORM(LOG_CHANGE(AssetsCurrent)) ...
fy   fp filed                                                                 ...
2010 Q2 2010-04-21                  0.000000                         0.000000 ...
     Q3 2010-07-21                  0.000000                         0.000000 ...
2011 Q1 2011-01-19                  0.978816                         0.074032 ...
     Q2 2011-04-21                  0.000000                         0.000000 ...
     Q3 2011-07-20                 -0.353553                        -0.353553 ...

Get tickers sorted by an industry-averaged quarterly report feature.

>>> finagg.sec.feat.quarterly.normalized.get_tickers_sorted_by(
...   "NORM(EarningsPerShareBasic)",
...   year=2019
... )[:5]
['XRAY', 'TSLA', 'SYY', 'WHR', 'KMB']

Get tickers sorted by an industry-averaged fundamental feature.

>>> finagg.fundam.feat.fundam.normalized.get_tickers_sorted_by(
...   "NORM(PriceEarningsRatio)",
...   date="2019-01-04"
... )[:5]
['AMD', 'TRGP', 'HPE', 'CZR', 'TSLA']

Configuration

API Keys and User Agents

API keys and user agent declarations are required for most of the APIs. You can set environment variables to expose your API keys and user agents to finagg, or you can pass your API keys and user agents to the implemented APIs programmatically. The following environment variables are used for configuring API keys and user agents:

• BEA_API_KEY is for the Bureau of Economic Analysis's API key. You can get a free API key from the BEA API site.
• FRED_API_KEY is for the Federal Reserve Economic Data API key. You can get a free API key from the FRED API site.
• INDICES_API_USER_AGENT is for scraping popular indices' compositions from Wikipedia and should be equivalent to a browser's user agent declaration. This defaults to a hardcoded value, but it may not always work.
• SEC_API_USER_AGENT is for the Securities and Exchange Commission's API. This should be of the format FIRST_NAME LAST_NAME E_MAIL.

Data Locations

finagg's root path, HTTP cache path, and database path are all configurable through environment variables. By default, all data related to finagg is put in a ./findata directory relative to a root directory. You can change these locations by modifying the respective environment variables:

• FINAGG_ROOT_PATH points to the parent directory of the ./findata directory. Defaults to your current working directory.
• FINAGG_HTTP_CACHE_PATH points to the HTTP requests cache SQLite storage. Defaults to ./findata/http_cache.sqlite.
• FINAGG_DATABASE_URL points to the finagg data storage. Defaults to ./findata/finagg.sqlite.

Other

You can change some finagg behavior with other environment variables:

• FINAGG_DISABLE_HTTP_CACHE: Set this to "1" or "True" to disable the HTTP requests cache. Instead of a cachable session, a default, uncached user session will be used for all requests.

Dependencies

• pandas for fast, flexible, and expressive representations of relational data.
• requests for HTTP requests to 3rd party APIs.
• requests-cache for caching HTTP requests to avoid getting throttled by 3rd party API servers.
• SQLAlchemy for a SQL Python interface.
• yfinance for historical stock data from Yahoo! Finance.

API References

• The BEA API and the BEA API key registration link.
• The FRED API and the FRED API key registration link.
• The SEC API.

Related Projects

• FinRL is a collection of financial reinforcement learning environments and tools.
• fredapi is an implementation of the FRED API.
• OpenBBTerminal is an open-source version of the Bloomberg Terminal.
• sec-edgar is an implementation of a file-based SEC EDGAR parser.
• sec-edgar-api is an implementation of the SEC EDGAR REST API.

Frequently Asked Questions

Where should I start?

Aggregate some data, create some analysis notebooks, or create some RL environments using the implemented data features and SQL tables. This project was originally created to make RL environments for financial applications but has since focused its purpose to just aggregating financial data and features. That being said, all the implemented features are defined in such a way to make it very easy to develop financial AI/ML, so we encourage you to do just that!

Why aren't features being installed for a specific ticker or economic data series?

Implemented APIs may be relatively new and simply may not provide data for a particular ticker or economic data series. For example, earnings per share may not be accessible for all companies through the SEC EDGAR API. In some cases, APIs may raise an HTTP error, causing installations to skip the ticker or series. Additionally, not all tickers and economic data series contain sufficient data for feature normalization. If a ticker or series only has one data point, that data point could be dropped when computing a feature (such as percent change), causing no data to be installed.

What Python versions are supported?

Python 3.10 and up are supported. We don't plan on supporting lower versions because 3.10 introduces some nice quality of life updates that are used throughout the package.

What operating systems are supported?

The package is developed and tested on both Linux and Windows, but we recommend using Linux or WSL in practice. The package performs a good amount of I/O and interprocess operations that could result in a noticeable performance degradation on Windows.