Financial Data APIs for an AI‑Augmented Valuation Platform

API limits and pricing change frequently; verify on the vendor site before committing to an integration (data current as of 14 Jan 2025).

Project Context and API Requirements

Intrinsic Edge is a 12-week MVP for a value-investing platform. It will use a modern stack (Next.js frontend, FastAPI backend, DuckDB+Parquet for local time-series, PGVector for embeddings). To empower AI-driven stock valuation, the chosen financial data API(s) must provide:

• Comprehensive Fundamentals: Deep company financials (income statements, balance sheets, cash flows) with at least 10+ years of history. Ideally also SEC filings or earnings call transcripts for qualitative analysis.
• Market Data: Historical stock price time series (daily OHLCV minimum, ideally decades back) and possibly real-time or delayed quotes for current pricing.
• Developer-Friendly: Clear documentation, simple REST or SDK access, and generous free/low-cost tiers suitable for prototyping (no mandatory hefty subscription). The API should be easy to integrate with Python or JavaScript.
• Usage Terms: Permissive free-tier or trial usage for early development, ideally without requiring upfront fees or long-term contracts. Licensing should allow using the data in a prototype application (at least for internal or non-commercial use).

Keeping these needs in mind, below is a comparison of several candidate financial data APIs, followed by a recommendation for the MVP.

Comparison of Financial Data APIs

The following table compares 7 popular APIs often used for stock data and fundamentals. It highlights their coverage, free tier limits, and notable pros/cons (including documentation quality and any licensing caveats):

API	Coverage & Features	Free Tier & Pricing	Key Limits / Notes
Alpha Vantage<br>(alphavantage.co)	• Fundamentals: Basic financial statements (annual and quarterly) for U.S. and global companies (official NASDAQ data vendor). Fundamentals observed in internal tests limited to approx. 5 years via API examples; AV docs do not promise depth.<br>• Market Data: Realtime (15-min delay) and historical prices for stocks (20+ global exchanges, decades of data); also FX, crypto, technical indicators.<br>• Other: Company overviews, earnings calendar, economic indicators, etc.	• Free: 25 API calls per day (as of Jan 2025, sufficient for light prototyping). No credit card required – instant API key signup.<br>• Premium: Higher call limits and real-time data starting around $50/month (optional, only if scaling up is needed).	• Fundamentals API library is limited in endpoints (covers the basics but not extensive data). <br>• Rate limiting on free tier (5 calls/minute and 500/day historically, now effectively 25/day). Must cache data locally (DuckDB) to avoid hitting limits.<br>• Docs: Comprehensive and straightforward with coding examples; official Python/JS libraries available. <br>• Usage: Data is reliably sourced (exchange-licensed). Free tier is meant for non-commercial or dev use – higher tiers needed for production scale or licensed redistribution.

| Financial Modeling Prep (FMP)<br>(financialmodelingprep.com) | • Fundamentals: Wide range of fundamental data – 10-K/10-Q financial statements, ratios, profiles. Free plan: ~5 years of historical fundamentals; Premium plans: up to 30 years. Also covers IPOs, analyst estimates, ESG, etc. <br>• Market Data: End-of-day stock prices (U.S. and some international) included; free plan provides ~5 years history. <br>• Other: SEC filings (parsed), insider trades, earnings calendars, and even earnings call transcripts (premium Ultimate plan) – 150+ API endpoints in total. | • Free (Basic): 250 calls/day limit. Access to most endpoints (fundamentals, EOD prices, profiles, etc.) but only annual data and ~5Y history for fundamentals. No monthly fee for basic use.<br>• Paid: Starter $22/mo, Premium $59/mo, Ultimate $149/mo (current as of Jan 2025). | • Free tier is robust for prototyping but historical depth is limited to 5 years (may not meet the 10+ year ideal without upgrading). <br>• API ease: JSON responses are well-structured; an API "Explorer" and documentation are provided (fairly developer-friendly). Community SDKs exist for Python/JS.<br>• Licensing: Data usage is intended for personal/individual use. Redistribution or display of data requires a special license agreement – so in a public app, showing raw financial figures from FMP may violate terms unless you upgrade and arrange licensing. <br>• Coverage: Primarily U.S. stocks on free plan (paid plans add UK/Canada/global). No transcripts in free tier (those are premium only). |

| Yahoo Finance (Unofficial via yfinance) | • Fundamentals: Basic financials (income, balance sheet, cash flow) for last ~4 years, key metrics and ratios, dividend history. Data scraped from Yahoo's site (not an official API).<br>• Market Data: Extensive historical price data for equities worldwide (decades of daily OHLCV), plus dividends and splits. Also covers indices, forex, crypto, etc. via Yahoo's datasets.<br>• Other: Analyst estimates, sector info, and news headlines available through Yahoo's pages (accessible via unofficial libraries). | • Free: 100% free, no account needed. yfinance Python library makes it easy to download data directly into pandas DataFrames. No formal rate limits, but subject to undisclosed throttling; heavy scraping can trigger blocks.<br>• No official paid tier – this is community-supported. (There are paid alternatives wrapping Yahoo data on RapidAPI, but not necessary for MVP). | • Not an official API: Data quality is generally good (Yahoo aggregates from various sources), but there are no uptime/accuracy guarantees. Unofficial use of Yahoo may break at any time if Yahoo changes its site. Not recommended for heavy production use.<br>• Limited fundamentals depth: Yahoo typically provides only a few years of financials on its site, not the 10+ years desired. You might get key stats and TTM figures, but for decade-long history you'd need another source.<br>• Docs & Integration: No official docs, but yfinance usage is well-documented by the community. Very quick for prototyping (one-liner to fetch years of price data).<br>• Licensing: Using Yahoo data in a commercial product is technically against Yahoo's terms. For a private MVP or research, it's commonly done, but one should avoid openly redistributing Yahoo's data. |

| Tiingo<br>(tiingo.com) | • Fundamentals: U.S. stocks coverage (approx. 5,500 companies) with up to 20+ years of history and ~80 fundamental indicators (both quarterly and annual). However, this requires a paid add-on (fundamentals are not included in the free tier).<br>• Market Data: End-of-day prices for 65,000+ global tickers (stocks, ETFs, mutual funds) with 30 years history. Real-time IEX data for U.S. equities (with subscription). Also supports crypto, FX, and news feeds (premium).| • Free: No monthly fee. 1,000 API calls/day (50/hour) for personal use. Includes historical price data (EOD) for up to 30+ years. Does not include fundamental data on the free plan.<br>• Premium: "Professional" or "Power User" plan at $10/month gives higher call limits (50k/day) and access to news; fundamentals require an add-on subscription (they launched fundamentals as a paid add-on to the $10 plan). | • Free tier limitation: Fundamental API is unavailable on the free plan. Tiingo is great for free price data, but to get 10+ years of financials you'd need to pay ($10 + add-on).<br>• Coverage: Fundamentals are U.S.-only. For global stocks or SEC filings, Tiingo won't help. Price data has wide coverage though.<br>• Documentation: Solid and includes a Python client (tiingo-python). Getting started is straightforward, and JSON/CSV responses are offered. <br>• Usage terms: Free and $10 plans are for internal/personal use only. To use Tiingo data in a product for others, one would need a commercial license. Keep this in mind if the platform will be used by outside users. |

| Polygon.io<br>(polygon.io) | • Fundamentals: Basic financials and reference data via REST API. U.S. stocks only, with fundamentals observed in practice covering 2–15 years depending on company; Polygon doesn't state a guarantee. (Coverage can be inconsistent – newer companies have only a couple years, older ones up to ~15). Lacks some advanced metrics; only core fundamentals are provided.<br>• Market Data: Real-time and historical market data for U.S. equities (tick-by-tick, intraday, daily), options, forex, and crypto. Great for live price feeds (WebSockets available). Corporate actions (splits, dividends) and news are also offered. | • Free: 5 calls/minute rate limit on the free plan. No hard monthly cap mentioned (effectively ~7,200 calls/day if spread out). All fundamental and historical endpoints are accessible on free tier, making Polygon's free plan quite generous. No credit card needed.<br>• Premium: "Developer" plan from $29/month raises the rate limits and gives more real-time streaming data. | • Fundamental data limitations: Polygon's fundamental API is minimalistic – only a few endpoints and not comprehensive. It may not satisfy a deep 10-year analysis (and only covers U.S. companies). Consider Polygon mainly for price data and basic ratios, not detailed financial statements. <br>• Strength: Excellent for real-time stock quotes and trades, if the platform needs live pricing or intraday data. Free tier allows basic real-time usage for prototyping. <br>• Documentation: Very developer-friendly, with interactive docs and Postman collections. Easy integration in Python (e.g., via requests or community SDKs). <br>• Licensing: Data is provided for development; for commercial use or higher volumes, one must upgrade. Polygon's terms allow using their data in applications, but redistributing large raw datasets might be restricted. |

| Finnhub<br>(finnhub.io) | • Fundamentals: Global coverage of company fundamentals – up to 30+ years of financial statements (annual/quarterly) for many markets. Also provides financial metrics, analyst estimates, and even alternative data (e.g. ESG, sentiment).<br>• Transcripts & Filings: Provides 15+ years of earnings call transcripts (with audio downloads) for many companies, plus a global filings search (SEC, SEDAR, etc.) – very valuable for deep fundamental analysis. <br>• Market Data: Real-time quotes and historical prices for global stocks, forex, crypto. Economic indicators and corporate news feeds (with sentiment analysis) are included as well. | • Free: 60 calls/minute rate limit, with up to 30/day for certain endpoints (e.g., /transcript endpoint capped at 30/day). In free mode, fundamental data is available (e.g. 10+ years statements) but some advanced features (like unlimited history or all transcripts) might be restricted. No monthly cost for the free tier, just sign up for an API key.<br>• Premium: $49–$99/month for individual premium (higher limits, more data), and higher tiers for enterprise. The free plan is generous enough for a prototype; premium needed if you require extensive history of alternative data or higher call volumes. | • Pros: Very feature-rich API – one of the few offering earnings call transcripts via API, plus international coverage. Developer-friendly (REST JSON API with clear docs, similar in style to Alpha Vantage). Integration is smooth; wrapper libraries exist (e.g., finnhub-python). <br>• Cons: Data accuracy is good but not perfect – some users report minor inconsistencies (Finnhub uses multiple sources and NBBO pricing, which is generally fine). Also, the sheer breadth means some endpoints (e.g. certain international filings) can be complex. <br>• Community: Smaller community support compared to Yahoo or Alpha Vantage, but active documentation and support from the company. <br>• Licensing: Free tier is intended for non-commercial or developmental use. For a commercial app or broad redistribution, Finnhub offers a separate enterprise plan (or startup license). Ensure compliance if the MVP's data will be shown to many end-users. |

| EOD Historical Data (EODHD)<br>(eodhistoricaldata.com) | • Fundamentals: Extremely comprehensive – 30 years of financial statements for stocks worldwide (70+ exchanges). Includes not just raw statements but ratios, dividends, earnings, and macro data. Often regarded as the one-stop shop for fundamentals + market data.<br>• Market Data: End-of-day and intraday stock prices globally, real-time data (with premium plan), plus economic indicators, mutual funds, ETFs, etc. Covers corporate actions, splits, and even logos and news. <br>• Other: Provides bulk download options, and even no-code solutions (Excel/Google Sheets add-ins). | • Free: There is a free plan, but fundamental data is not included on free. The free tier mainly lets you try a few endpoints or get limited history for a couple of symbols (sufficient to test the API format). <br>• Paid: To get full fundamentals, $59.99/month (Fundamentals plan) gives US + global fundamentals, 30Y history, etc.. A higher $99/mo plan includes real-time and all data (current as of Jan 2025). | • Complete solution (at a cost): EODHD is often recommended for all-in-one data (prices + fundamentals + news). If budget allows, it can power the entire platform with a single integration. <br>• Quality & docs: Data quality is considered high, and documentation is very thorough with examples and field explanations. The API is straightforward (CSV or JSON output). <br>• Prototype vs Production: For the MVP phase, the cost might be high if you're avoiding subscriptions. But EODHD could be a future upgrade once the project matures (to replace patchwork free sources with a unified, license-compliant source). <br>• Licensing: Paid plans come with data usage rights for your applications (still subject to terms, but intended for developers to use in apps). No display licensing issues as long as you are a subscriber. |

Discussion of Options

From the comparison, we note the following key points:

• Breadth vs. Depth: No completely free API offers unlimited 10+ year fundamentals. Financial Modeling Prep and Alpha Vantage have free tiers but cap historical fundamentals around 5 years. Finnhub and Polygon free plans offer deeper data (Finnhub claims decades, Polygon up to 15 years), but Finnhub's richest data (transcripts, global filings) may require a premium, and Polygon is U.S.-only. EODHD has the full 30-year global coverage but essentially requires a paid plan.

• Stock Price Data: This is easier to obtain freely. Yahoo Finance (yfinance) is a convenient go-to for historical price series (decades of data, all for free). If avoiding unofficial sources, Alpha Vantage or Polygon can provide daily prices; Alpha Vantage's free tier allows full-history daily data ticker by ticker (just abide by the call limits) and Polygon's free tier gives real-time U.S. quotes. Tiingo is another excellent source of clean EOD prices (free 30-year history). For the MVP, you can mix and match: e.g. use yfinance to quickly pull price history for many stocks into your local DuckDB, and use another API for fundamentals.

• Fundamentals and Financial Statements: Financial Modeling Prep stands out for ease of use – their JSON responses for financial statements are structured and they have many convenience endpoints (ratios, profiles, etc.). The free plan's 250 calls/day is workable for a prototype, and you get a lot of data breadth (150+ endpoints) even if depth is limited to 5 years. Finnhub offers a wider time horizon (potentially 10+ years statements) and extras like international stocks and transcripts, but integrating all those endpoints can be more complex. If focusing on U.S. equities, Polygon or Tiingo (paid) could fill the gap, but Polygon's fundamentals are sparse and Tiingo's require a subscription.

• Transcripts and Filings: This is a specialized need. Among the APIs, Finnhub and FMP (premium) provide earnings call transcripts through an API. Finnhub's transcripts cover many companies back ~15 years, but free tier may only include recent transcripts (premium needed for full history). FMP's Ultimate plan ($149/mo) includes transcripts, which is likely overkill for an MVP. If transcripts are a core feature, a practical approach is to use Finnhub's free plan to get the most recent year of transcripts (to experiment with NLP on earnings calls), and not worry about older transcripts for now. Alternatively, one could use SEC EDGAR filings (which are free via SEC's API or data downloads) for textual analysis – though that requires more custom parsing and doesn't cover the Q&A like transcripts do.

Recommendation and Next Steps

For a 12-week solo-built MVP, a combination of a primary API plus a couple of supplemental sources is recommended to balance coverage and cost:

• Primary API for Fundamentals: Financial Modeling Prep (FMP) is a strong choice to start. Its free tier gives you quick access to financial statements, ratios, profiles, etc., sufficient for a prototype. You can pull ~5 years of financial history for each company, which covers most use cases in a demo. The JSON data is easy to ingest into DuckDB. Caveat: 5 years is below the ideal 10+ year goal – but you can use this to build out the app's features now, and later either upgrade to FMP Premium (to unlock 30-year data) or switch to a more extensive API if needed. FMP's breadth (many endpoints) will let you experiment with additional data (analyst ratings, SEC filings, etc.) easily in the MVP phase.

• Secondary Source for Historical Prices: Rather than using up API calls for price history, leverage Yahoo Finance via the yfinance library for bulk price downloads. It's free and provides decades of daily data for U.S. and international stocks. You can quickly fetch the entire price history of a company and store it in Parquet/DuckDB locally. This "local-first" approach aligns with your architecture – once you have the data in Parquet, your app can query it without hitting an API repeatedly.

• Augmenting Missing Pieces: For any data not covered by FMP or Yahoo:

  • Longer Financial History – If 10+ year fundamentals are a must-have for certain companies in the MVP, consider using Finnhub's free API for those specific cases. Finnhub's standard financials endpoint can provide up to 10 years or more for U.S. stocks. You could query Finnhub for older data (e.g., years 6–15) while using FMP for the recent years, merging the results.
  • Transcripts or SEC Filings – To incorporate NLP on earnings calls, try Finnhub's transcripts on the free tier for a couple of recent transcripts. If the free tier doesn't allow it, you might use Finnhub's $50/mo plan for one month to download a batch of transcripts of interest (then store them locally for your vector DB).
  • Real-Time Quotes – If your app needs live pricing, Polygon.io could be used in free mode for U.S. stocks (5 calls/min of quote updates) or Finnhub (free tier provides real-time as well).

Recommendation Summary

For the Intrinsic Edge MVP, use Financial Modeling Prep as the primary data API for fundamentals and basic stock info, complemented by Yahoo Finance data (via yfinance) for historical price coverage. This combination covers most needs without upfront cost. As needed, tap into Finnhub's free API for any gaps (such as extended financial history or earnings transcripts).

This approach minimizes costs while providing robust data to your AI models. You'll be able to populate your DuckDB with 5+ years of financials and decades of prices, implement valuation algorithms, and experiment with NLP on filings – all within the 12-week timeline. Just be mindful of each API's usage limits and terms; design your data ingestion to cache results (avoiding redundant API calls) and credit the sources appropriately.

Overall, FMP + yfinance (with a dash of Finnhub) should accelerate your development now. As the project evolves, you can evaluate a move to a paid plan or a more comprehensive API like EODHD for long-term sustainability. This phased strategy gives you the data you need today and flexibility for tomorrow, enabling you to build Intrinsic Edge with confidence in your data foundation.