Finance Database

[FinanceDatabase]

[GitHub Sponsors] [Buy Me a Coffee] [LinkedIn] [MCP Server] [Download MCP Bundle] [Documentation] [Supported Python Versions] [PYPI Version] [PYPI Downloads]

As a private investor, the sheer amount of information that can be found on the internet is rather daunting. Trying to understand what types of companies or ETFs are available is incredibly challenging, with millions of companies and derivatives available on the market. Sure, the most traded companies and ETFs can quickly be found simply because they are known to the public (for example, Microsoft, Tesla, S&P 500 ETF, or an All-World ETF). However, what else is out there is often unknown.

This is why I created the FinanceDatabase, a database featuring 300,000+ symbols containing Equities, ETFs, Funds, Indices, Currencies, Cryptocurrencies and Money Markets. It allows you to obtain a broad overview of sectors, industries, types of investments and much more, entirely for free. Everything is stored in CSV files that anyone can read and edit, so the database grows and improves through the community.

The aim of this database is explicitly not to provide up-to-date fundamentals or stock data, as those can be obtained with ease (with the help of this database) by using the Finance Toolkit 🛠️. Instead, it gives insights into the products that exist in each country, industry and sector and provides the most essential information about each product. With this information, you can analyze specific areas of the financial world and/or find a product that is hard to find. By utilising both, it is possible to do a fully-fledged competitive analysis with the tickers found from the FinanceDatabase inputted into the FinanceToolkit.

Some key statistics of the database:

🔌 The Finance Database is also available as an MCP Server

Explore all 300,000+ symbols from Claude, Copilot, Cursor, Windsurf or any MCP-compatible client without writing code. No API key needed.

• Hosted: connect to https://financedatabase.jeroenbouma.com/mcp — nothing to install and no API key.
• Local: uvx --from "financedatabase[mcp]" financedatabase-mcp-setup — sets up your client config automatically. See MCP Server Documentation for manual setup.

Table of Contents

1. Installation
2. Functionality
3. MCP Server
4. Questions & Answers
5. Contributing
6. Contact

Installation

Before installation, consider starring the project on GitHub which helps others find the project as well.

[image]

To install the FinanceDatabase it simply requires the following:

pip install financedatabase -U

Then within Python use:

import financedatabase as fd
equities = fd.Equities()

No API key is needed. Each asset class is downloaded once and cached locally, see Questions & Answers for how the cache works.

Functionality

This section is an introduction to the FinanceDatabase. Find with the link below a Jupyter Notebook in which you can see many more examples, including the full output of each query.

A basic example of how to use the FinanceDatabase is shown below. Every code snippet in the sections that follow builds on this same equities instance. Initialization of each asset class is only required once, so save it to a variable to query the database much more quickly.

import financedatabase as fd
# Initialize the Equities databaseequities = fd.Equities()
# Select all equitiesequities.select()

A portion of the output is shown below. The tables in this section are cut off to five entries and a selection of the columns due to the sheer size of the database.

And below the actively listed equities are counted per sector.

Each asset class has the same three functions: select to filter on the database's categories, search to look for any text in any column and show_options to see which values a category can take. The asset classes are fd.Equities(), fd.ETFs(), fd.Indices(), fd.Currencies() and fd.Cryptos(), which are updated weekly from official exchange and market sources.

Three capabilities cut across all of them:

• Lists of values. Every filter accepts a single value or a list, e.g. country=['Netherlands', 'Belgium'], returning the entries that match any of them.
• only_primary_listing and exclude_delisted. A company is often listed on many exchanges. only_primary_listing=True keeps only its primary listing, and delisted symbols are left out by default (exclude_delisted=False to include them).
• pandas or Polars. Queries run lazily with Polars and return a pandas DataFrame by default, or a Polars DataFrame with as_pandas=False.

Exploring the Options

With show_options, all possible options are given per column. This is useful as it doesn't require loading the larger data files.

# Show the options of every column of the equitiesoptions = fd.show_options("equities")
# Select the sectorsoptions["sector"]

This returns the eleven sectors used for equities:

['Communication Services', 'Consumer Discretionary', 'Consumer Staples', 'Energy', 'Financials', 'Health Care', 'Industrials', 'Information Technology', 'Materials', 'Real Estate', 'Utilities']

Once an asset class is loaded, show_options is also available on the class itself, where it shows the options that remain after filtering. For example, the industries of the financial companies in the Netherlands:

# Show the industries of financial companies in the Netherlandsequities.show_options(    selection="industry",    sector="Financials",    country="Netherlands",)

Which returns:

['Banks', 'Capital Markets', 'Consumer Lending', 'Insurance', 'Payments & Specialty Finance']

And below the number of companies in each of these industries is shown. Each company is counted once, however many exchanges it is listed on: listings that share an ISIN, share class FIGI, website or the first word of their name (such as Aegon and its perpetual bonds) are combined.

The options of every column can be shown this way, including currency, exchange, market, country and market_cap (Mega, Large, Mid, Small, Micro and Nano Cap). Find the Notebook here and the full documentation here.

Selecting Equities

Given these options, it becomes possible to filter the database on the categories you are interested in. For example, the 'Insurance' companies in the 'United States', one row per company. The sector can be omitted here since the industry already implies 'Financials'.

# Select the primary listings of insurance companies in the United Statesequities.select(    country="United States",    industry="Insurance",    only_primary_listing=True,)

This returns 194 companies, of which a few of the larger ones are shown below.

Without only_primary_listing=True, the same query returns 455 listings on 23 exchanges, because every exchange a company trades on is shown by default. Progressive, for example, also appears as PGV.F (Frankfurt), PGV.SG (Stuttgart), PGR.MX (Mexico) and P1GR34.SA (B3). The primary listing of a company is its listing on the main exchange of its home market, or without one its listing on the largest main exchange abroad. It is worked out from the data itself, so it works for companies anywhere: for ASML it is ASML.AS on Euronext Amsterdam, not ASML on NASDAQ or ASMLF over the counter. To get the listings in a particular country instead, such as the US listings of ASML, filter on the exchange, market or currency.

For the Netherlands, it makes sense to select the market "Euronext Amsterdam" (exchange "AMS"), here together with the market cap:

# Select the large insurance companies on Euronext Amsterdamequities.select(    country="Netherlands",    industry="Insurance",    market="Euronext Amsterdam",    market_cap="Large Cap",)

This gives the following three companies:

Every filter also accepts a list, so both queries can be combined into one with country=["Netherlands", "United States"] and market=["Euronext Amsterdam", "New York Stock Exchange", "NASDAQ Global Select"]. Equities can be selected on country, sector, industry_group, industry, currency, exchange, mic, market and market_cap. Find the Notebook here and the full documentation here.

Searching the Database

If the categorization doesn't lead to the results you are looking for, search filters on any column via a custom string. If the text is found anywhere in the column, the entry is returned. Searches are not case sensitive unless case_sensitive=True is set, index searches the symbol and, just like select, every argument accepts a list.

# Search for robotics or education companies with equipment on the Frankfurt Stock Exchangeequities.search(    summary=["Robotics", "Education"],    industry="Equipment",    market="Frankfurt",    index=".F",)

This returns the instruments listed on the Frankfurt Stock Exchange in an industry containing "Equipment" and with "Robotics" or "Education" in their summary. Filtering on index=".F" is an alternative way to find the exchange or market you are looking for.

And below the results are counted per industry.

The search function works on every column of every asset class, including name, isin, cusip and figi. Find the Notebook here and the full documentation here.

Exploring other Asset Classes

All of these functions are also available for the other asset classes. The only difference is the class name and the columns. For example, ETFs use fd.ETFs() and are selected on category_group, category and family instead.

# Initialize the ETFs databaseetfs = fd.ETFs()
# Select the Fixed Income ETFs of Vanguardetfs.select(    category_group="Fixed Income",    family="Vanguard Asset Management",)

For example, see some of the Vanguard bond ETFs listed in Berlin below:

The same applies to search, for example to find semiconductor indices across providers and countries:

# Initialize the Indices databaseindices = fd.Indices()
# Search for indices with "Semiconductor" in their nameindices.search(name="Semiconductor")

Which returns semiconductor indices from the US, South Korea and Taiwan, including their price, net and total return variants, of which a few are shown below:

And below all actively listed ETFs are divided over their category groups.

The categories of each asset class can again be found with show_options, e.g. fd.Indices().show_options(selection="category") returns the 42 index categories, from 'Corporate Bonds' and 'Emerging Markets' to 'Small Cap' and 'Value'. Cryptocurrencies are selected on cryptocurrency and currency and currencies on base_currency and quote_currency. Find the Notebook here and the full documentation here.

Combining with the Finance Toolkit

The FinanceDatabase has a direct integration with the Finance Toolkit, making it possible to do financial analysis on the instruments you've found. Any selection can be loaded into the Finance Toolkit with to_toolkit.

To be able to get started, you need to obtain an API Key from FinancialModelingPrep. This is used to gain access to 30+ years of financial statements, both annually and quarterly. Note that the Free plan is limited to 250 requests each day, 5 years of data, and only features companies listed on US exchanges.

Returning to the three large insurance companies in the Netherlands:

# Select the large insurance companies on Euronext Amsterdamdutch_insurance_companies = equities.select(    country="Netherlands",    industry="Insurance",    market="Euronext Amsterdam",    market_cap="Large Cap",)
# Load them into the Finance Toolkittoolkit = dutch_insurance_companies.to_toolkit(api_key="FINANCIAL_MODELING_PREP_KEY")
# Obtain historical market data for all tickershistorical_data = toolkit.get_historical_data()
# Select the results for ASR Nederlandhistorical_data.xs("ASRNL.AS", axis=1, level=1)

For example, a portion of the historical data for ASR Nederland is shown below.

And below the cumulative returns of Aegon, ASR Nederland and NN Group are plotted, including the S&P 500 as benchmark.

Now let's make it more advanced by calculating the profitability ratios for each company:

# Collect all Profitability Ratios for all tickersprofitability_ratios = toolkit.ratios.collect_profitability_ratios()
# Select the results for ASR Nederlandprofitability_ratios.loc["ASRNL.AS"]

For example, see some of the profitability ratios of ASR Nederland below.

And below these and other profitability ratios of ASR Nederland, each with its latest value, the change over the period and its trend.

This works for the other asset classes too. For example, Ethereum quoted in BTC, CAD, EUR, GBP and USD can be loaded with fd.Cryptos().select(cryptocurrency="ETH").to_toolkit(api_key=...), after which get_historical_data(period="quarterly") returns its quarterly returns in each currency. This is just a small snippet of what is available within the Finance Toolkit, see the GitHub page of the Finance Toolkit here or the example Notebook here for more information.

MCP Server

The Finance Database MCP Server gives any AI assistant that supports the Model Context Protocol (MCP) direct access to the database. Ask in plain English for, say, every mid cap semiconductor company in Taiwan or the bond ETFs of a given issuer, and the assistant queries the database on your behalf. No API key is needed. Find the tools, example prompts and how to host the server yourself in the MCP Server Documentation.

Remote server

Connect directly to the hosted server at https://financedatabase.jeroenbouma.com/mcp. Nothing needs to be installed locally and no API key or sign-in is required. See the privacy policy for what the hosted server processes.

Local installation

Run the setup wizard — it locates your client's config file and writes the MCP entry automatically:

uvx --from "financedatabase[mcp]" financedatabase-mcp-setup

For manual config, add the following to your client's MCP config file (e.g. claude_desktop_config.json or .cursor/mcp.json; VS Code's .vscode/mcp.json uses servers instead of mcpServers):

{  "mcpServers": {    "finance-database": {      "command": "uvx",      "args": ["--from", "financedatabase[mcp]", "financedatabase-mcp"]    }  }}

Alternatively, download the Finance Database MCPB bundle and open it with Claude Desktop.

Questions & Answers

This section includes frequently asked questions. If you have any questions that are not answered here, consider creating an Issue or reach out to me via the contact details below.

How is the data obtained?

The data is an aggregation of publicly available sources: the official listing directories that exchanges publish (such as Nasdaq Trader, HKEX, NSE, the Japan Exchange Group, ASX, TSX, Deutsche Börse Xetra, Euronext and SIX Swiss Exchange), open reference data (OpenFIGI, SEC EDGAR, the ISO 4217 currency lists, CoinGecko) and contributions from the community. I strictly maintain the rule that all data in this database must be freely accessible to everyone. Data requiring paid subscriptions is never included. Information that companies charge for is typically owned and maintained by those companies, making public sharing of such data a violation of their Terms of Service (ToS). However, publicly available data can be freely shared (read more about the legality of web scraping here). This database will always remain completely free.

What categorization method is used?

Equities are classified in this database's own three levels: 11 sectors, which are divided into industry groups and then industries. Every valid combination is listed in scripts/categories.json. The categories are curated by hand and from public sources, and the weekly update classifies new companies the same way. All categorizations in the database are independently developed and can be freely modified.

How can I find out which countries, sectors and/or industries exist within the database without needing to check the database manually?

For this you can use the show_options function, either for an asset class as a whole (fd.show_options("equities")), which doesn't require any data to be loaded, or on a loaded asset class to see the options that remain after filtering. See Exploring the Options for more information.

When I try to collect data I notice that not all tickers return output, why is that?

Some tickers are merely holdings of companies and therefore do not really have any data attached to them. Therefore, it makes sense that not all tickers return data. If you are still in doubt, search the ticker on Google to see if there is really no data available. If you can't find anything about the ticker, consider updating the database by visiting the Contributing Guidelines.

How is the database kept up to date?

Every Sunday (and after every change to the database) an automated update runs on GitHub Actions:

Each source is applied on its own, so a source that is unavailable or changes its format is skipped for that week without affecting the others, and implausible changes (such as a large share of market cap tiers changing at once) are not applied. Before anything is committed, the update is checked against the previous version of the database and undone if it would introduce duplicate symbols, symbols in two asset classes, changed columns or a large loss of rows. Delisted tickers are kept on purpose, marked as delisted, for historical research. The scripts and their rules are described in scripts/README.md.

Not everything can be automated. Classifications of existing companies, corporate events outside the automated sources (mergers, renames, migrations) and exchanges without a public listing directory still rely on the community. If you notice outdated information, please consider contributing through the Contributing Guidelines.

What happened to Funds and Money Markets?

Since October 2026, funds and money markets are no longer maintained. Unlike exchange-listed instruments, mutual and money market funds have no public source that keeps them current worldwide, and the data had become outdated (many funds closed or merged, many new ones missing). fd.Funds() and fd.MoneyMarkets() remain available in the Python package as they are, with a warning when loaded, but they are no longer part of the MCP server. Funds that trade on a stock exchange under a regular ticker, such as listed open-ended funds in China or listed funds on Euronext Amsterdam, were moved to the ETFs so they stay maintained and searchable.

Is the data downloaded every time I use the package?

No. Each dataset is downloaded once and cached in your user cache folder (or the folder set in FINANCEDATABASE_CACHE_DIR). Once a day the package checks for a newer version and only downloads it when it changed; offline, the cached copy is used. Queries run lazily with Polars and return pandas by default, or Polars with as_pandas=False, e.g. equities.select(country="Canada", as_pandas=False).

Contributing

First of all, thank you for taking the time to contribute (or at least read the Contributing Guidelines)! 🚀

The FinanceDatabase serves the role of providing anyone with any type of financial product categorization entirely for free. To achieve this, it relies on community involvement to add, edit and remove tickers over time. This is made easy enough that anyone, even those with a lack of coding experience, can contribute because of the use of CSV files that can be manually edited with ease.

Below are those that made significant contributions to the project. Thank you!

Contact

If you have any questions about the FinanceDatabase or would like to share with me what you have been working on, feel free to reach out to me via:

• Website: https://jeroenbouma.com/
• LinkedIn: https://www.linkedin.com/in/boumajeroen/
• Email: [email protected]

If you'd like to support my efforts, either help me out via the Contributing Guidelines or Sponsor Me.

[Star History Chart]