# BeliefState vs EODHD

> BeliefState and EODHD score within a point of each other for market fundamentals, 61.1 and 61.1 out of 100. Prices, MCP, x402, uptime and agent notes side by side.

- Canonical: https://www.anchorterminal.com/compare/beliefstate-vs-eodhd
- Markdown: https://www.anchorterminal.com/compare/beliefstate-vs-eodhd.md (~4,350 tokens)
- Slim: https://www.anchorterminal.com/compare/beliefstate-vs-eodhd.min.md (~1,080 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/compare/beliefstate-vs-eodhd.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-15

BeliefState and EODHD score within a point of each other on agent readiness, 61.1 (C) and 61.1 (C). EODHD leads on reliability and agent ergonomics. Both do market fundamentals. BeliefState accepts x402. EODHD doesn't.

- BeliefState: grade C, 61.1/100, rank #726 of 1526. Markdown https://www.anchorterminal.com/tools/beliefstate.md · JSON https://www.anchorterminal.com/api/v1/tools/beliefstate.json
- EODHD: grade C, 61.1/100, rank #727 of 1526. Markdown https://www.anchorterminal.com/tools/eodhd.md · JSON https://www.anchorterminal.com/api/v1/tools/eodhd.json
- Best financial market data APIs for AI agents: https://www.anchorterminal.com/best/market-data/index.md
- All 89 markets comparisons: https://www.anchorterminal.com/compare/market-data/index.md

## Which one, for what

### BeliefState (C)

Good for: Agents doing equity research that want who said what about a US stock, why, and what changed, with citations.

Ahead on:
- Schema & documentation, 86 against 70
- Security & auth, 59 against 39
- Payments & pricing, 84 against 45

Also in its favour:
- An agent can pay per call over x402, with no account

Watch for: Which sources the theses come from, and under what rights, isn't on any public page

### EODHD (C)

Good for: Agents that need long end-of-day history, dividends and splits and fundamentals across 60 exchanges at a low monthly price, with weighted calls to budget.

Ahead on:
- Reliability, 65 against 20
- Agent ergonomics, 76 against 67

Also in its favour:
- Runs on your own machine
- Free to start without a card

Watch for: Every docs page shows the token as `api_token=` in the query string, and the MCP v1 address carries it in the URL


## Why the edge

### Reliability: EODHD +45

- BeliefState (20): Read with the hosted lines. No status page found; /status returns 404 and `/v1/status` is an account utility (0). No incident history to read (5). No rate limits with numbers found in the docs, agents.md or the OpenAPI file, though the x402 routes list a 429 response (0). The quickstart documents Retry-After on a 503 `payment_pending` and retrying the same saved request and signature, which returns the original response instead of charging twice, and says what a 402 or 410 means (10 of 15). No SLA (0). The wallet routes carry no preview label, but some API references are marked Preview, the MCP guide says open registration stays closed until a public admission gate passes, and A2A is gated (5 of 10). 0+5+0+10+0+5 = 20.
- EODHD (65): Read with the hosted lines. A public status page exists at uptime.eodhistoricaldata.com, an UptimeRobot page announced on the blog on 27 March 2020 (20), but its figures load by script and our reader got an error, so no incident history could be read (5). Limits are published with numbers, 20 calls a day free, 100,000 a day on paid plans, 1,000 requests a minute, with call weights per endpoint (15). 429 carries Retry-After, X-RateLimit headers report the ceiling, and 402 marks an exhausted allowance; reads only (15). The terms promise commercially reasonable efforts and no SLA, with SLAs sold by sales (0). The core API is generally available, with several newer APIs marked beta (10). 20 + 5 + 15 + 15 + 0 + 10 = 65.

### Payments & pricing: BeliefState +39

- BeliefState (84): x402 v2 on the vendor's own domain for all 20 paid research tools, over HTTP and the wallet MCP server; `/v1/beliefs` and `/v1/sources` need a key (9 of 10). Per-call prices (0.02, 0.50 and 1.50 USDC) and plan prices published without a login (30). Free ticker, company and SEC-filing discovery with no account or wallet, and five signup credits for eligible new accounts; no free research call without paying or signing up was found (15 of 30). An agent with a funded Base wallet buys research with no account or key (30). 9+30+15+30 = 84.
- EODHD (45): No x402, MPP or L402 (0). Plan prices are public without login, personal and commercial, sold by the month with daily call allowances and per-endpoint call weights rather than per call (10). A free plan at 20 calls a day with no card (20). The token comes from a browser signup and no key API was found (0). 0 + 10 + 20 + 0 = 30. Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 30 is 45.

### Security & auth: BeliefState +20

- BeliefState (59): OAuth browser sign-in for account MCP with one revocable grant per client, and revocable read-only workspace keys made through device authorisation and stored as hashes. The pricing page calls keys scoped, but no scope list or rotation was found (24 of 30). Every tool reads, and the device-flow key is read-only (18 of 20). Results include text drawn from third-party research, news and filings. The docs tell agents to check sources and flag partial or unverified source text, with no prompt-injection guidance (6 of 15). A usage and credit metering endpoint and x402 receipts, with no per-call log found (6 of 15). security.txt is valid but its contact is a personal Gmail address, and no disclosure policy, bug bounty or certification was found (5 of 20). 24+18+6+6+5 = 59.
- EODHD (39): One token per account with no scopes, shown as `api_token=` in the query string on every docs page, with a Bearer header allowed per llms.txt; the MCP v1 address embeds the token and v2 uses OAuth 2.1 with one scope, full-access (20 less 10 for the query string, so 10). All REST calls read; three MCP report tools trigger provider-side report generation and email (14 of 20). News and sentiment are third-party text and no guidance was found (5 of 15). A usage endpoint and dashboard show calls spent, not a per-call log (6 of 15). No security.txt, disclosure policy, bug bounty or certification found; the MCP server's CI runs Bandit, Semgrep and pip-audit (4 of 20). 10 + 14 + 5 + 6 + 4 = 39.

### Schema & documentation: BeliefState +16

- BeliefState (86): OpenAPI 3.1.0 with 48 operations, request and response schemas, and an x402 security scheme. The MCP tool schemas weren't read because robots.txt closes `/mcp` (23 of 25). llms.txt, agents.md, skill.md, index.md and Markdown tool pages (10). Descriptions say when to use each tool and when to use another, such as starting company overviews with `get_brief` and paging `get_cases` only when the user wants more (18 of 20). Typed parameters and bounds on the operations read, not checked for every tool (11 of 15). Examples on tool pages, a sample Broadcom record, and error guidance for 402, 409, 410 and 503 (12 of 15). Routes under `/v1`, dated OpenAPI publications and a dated changelog, but pages disagree on counts in the same week: the quickstart says fourteen data tools at 0.02 USDC, the FAQ and agents.md eighteen, and the account server is 25 tools in the changelog and 26 on the tools page (12 of 15). 23+10+18+11+12+12 = 86.
- EODHD (70): An OpenAPI 3.1 document in a public repository covers 74 endpoints, last updated 18 February 2026, and the MCP server's 88 tools carry typed inputs and annotations in the source (22 of 25). llms.txt of 32 KB and an llms-full.txt, plus documentation resources inside the MCP server (10). Each API page states its purpose and plan, and tool titles name the dataset, with little on when not to use one (14 of 20). Parameters are documented with formats and the symbol convention is spelled out, but allowed values sit in prose (8 of 15). Examples on every API page and the demo token, with 401, 402, 403, 404, 422 and 429 documented in llms.txt and the limits article (11 of 15). Exchange Details v2 shipped in May 2026 with v1 kept, MCP has v1 and v2, and new APIs are announced on a dated blog, but there is no changelog page (5 of 15). 22 + 10 + 14 + 8 + 11 + 5 = 70.

### Agent ergonomics: EODHD +9

- BeliefState (67): 23 wallet tools and 26 account tools, in the 11 to 30 band (15). Briefs come in overview, current and full views, cases page with a cursor, and filters cover investor, direction, time, `changed_since` and `as_of` cut-offs (18 of 20). Documented status meanings and recovery steps for payment states, with no full error-code table found (15 of 20). An unfinished purchase is retried with the same saved URL, body and signature, and the docs say not to pay twice. MCP annotations couldn't be read (12 of 20). Few required arguments on the tools read. No official SDK; a Postman collection, plugins and an MIT examples repository (7 of 15). 15+18+15+12+7 = 67.
- EODHD (76): The REST API is the surface graded. `from`, `to`, `period`, `fmt`, `filter` on fundamentals and `limit` on news size responses, and the hosted MCP server lists 72 or 88 tools (15 of 25). `limit` and `offset` on news and the screener, date ranges on series, bulk per exchange (15 of 20). Error codes an agent can branch on, 402 and 403 with meanings, and the MCP server answers 403 with what the plan covers (15 of 20). Every REST call is a read; 85 of 88 MCP tools carry readOnlyHint true and the three Praams report tools are marked otherwise (16 of 20). Few required parameters and official Python and Node.js libraries, with R, C# and PHP clients (15). 15 + 15 + 15 + 16 + 15 = 76.

## Score by category

| Category | Weight | BeliefState | EODHD | Edge |
| --- | --- | --- | --- | --- |
| Reliability | 16% (20 this run) | 20 | 65 | EODHD +45 |
| Performance | 10%, pending | pending | pending | not scored in this run |
| Schema & documentation | 13% (16.2 this run) | 86 | 70 | BeliefState +16 |
| Agent ergonomics | 13% (16.2 this run) | 67 | 76 | EODHD +9 |
| Security & auth | 14% (17.5 this run) | 59 | 39 | BeliefState +20 |
| Payments & pricing | 10% (12.5 this run) | 84 | 45 | BeliefState +39 |
| Task success | 10%, pending | pending | pending | not scored in this run |
| Maintenance & community | 7% (8.8 this run) | 78 | 82 | EODHD +4 |
| Transparency & trust | 7% (8.8 this run) | 52 | 54 | EODHD +2 |
| Negative events | ≤15 | 0 | 0 | |
| **Total** | | **61.1 · C** | **61.1 · C** | |

## Facts side by side

| Fact | BeliefState | EODHD |
| --- | --- | --- |
| Kind | HTTP API | HTTP API |
| Vendor | BeliefState | Unicorn Data Services |
| Hosted endpoint | `https://beliefstate.ai/v1/research/mcp` | `https://eodhd.com/api` |
| Transports | Streamable HTTP | HTTP, Streamable HTTP, SSE (legacy), stdio, websocket |
| Auth | API key | OAuth or key |
| Pricing | Pay per use | Freemium |
| Price for market fundamentals | not published | $59.99 per month (plan) |
| x402 | yes, from $0.02/call | no |
| Tools exposed | 23 | 88 |
| Read-only variant documented | yes | no |
| llms.txt | yes | yes |
| MCP registry | `ai.beliefstate/investor-research` | not listed |
| Last release | 2026-11-10 | 2026-10-15 |
| Terms last updated | not read yet | no date given |
| Privacy policy last updated | not read yet | no date given |
| Customer content may train models | not read yet | not found in the text |
| Terms restrict automated access | not read yet | not found in the text |
| Terms restrict benchmarking | not read yet | not found in the text |
| Terms or service can change without notice | not read yet | not found in the text |
| Arbitration or class-action waiver | not read yet | not found in the text |
| Popularity | none | 24 stars |

## How the scores have moved

### BeliefState

No score change since it was listed.

### EODHD

- +1.9 on 13 November 2026: checklist
- +0.2 on 12 November 2026: time

## Verdicts

**BeliefState.** Agents can buy cited investor theses, price targets, risks and revisions per call by x402 with no account, at published USDC prices, or connect by OAuth or a read-only key. Sources and their rights aren't public, there is no status page or rate-limit figure, and docs pages disagree on tool counts.

**EODHD.** Limits, call weights, rate-limit headers, 429 with Retry-After and 402 on an exhausted allowance are documented, with an OpenAPI document and an open-source MCP server of 88 read-only tools. The token is documented in the query string, the MCP v1 address embeds it, the status page could not be read, and the terms carry no date, SLA or governing law.

## Before you call either

### BeliefState

1. Check coverage first with the free `get_tickers` or `get_companies`; an absent ticker means unreleased, not unknown
2. Show the x402 quote to the user and keep the exact URL, body and signature; on 503 `payment_pending` retry those, never pay again
3. Start a company with `get_brief`, then open cases with `get_case`; `get_search_results` costs 1.50 USDC a call
4. Keep each record's publication date and first-seen date apart, and read `source_text_coverage` before treating a source as complete
5. Don't use the device-flow API key inside a hosted chat; use OAuth there

### EODHD

1. Send the token as `Authorization: Bearer`, which llms.txt allows, rather than `api_token=` in the URL, and prefer the v2 OAuth MCP address over v1
2. Budget calls, not requests; fundamentals cost 10, intraday and news 5, bulk 100 or more, and a 404 for a bad ticker is still charged, so resolve tickers with the search endpoint first
3. Read `X-RateLimit-Remaining` and honour Retry-After on 429; a 402 means the daily allowance is gone until midnight GMT and no retry will help
4. Use `{TICKER}.{EXCHANGE}` symbols in upper case, such as AAPL.US or EURUSD.FOREX, and `BTC-USD.CC` for crypto
5. Expect 20 calls a day and one year of history on the free plan; a 403 means the plan lacks the endpoint, not a bad key

## Questions

### Which is better for AI agents, BeliefState or EODHD?

BeliefState and EODHD score within a point of each other on agent readiness, 61.1 (C) and 61.1 (C). EODHD leads on reliability and agent ergonomics.

### Do BeliefState and EODHD need an API key?

BeliefState needs an API key. EODHD takes an API key or an OAuth sign-in. An agent can also pay BeliefState per call over x402, with no account.

### Can an agent call BeliefState and EODHD without installing anything?

Yes. BeliefState has a hosted endpoint at https://beliefstate.ai/v1/research/mcp and EODHD at https://eodhd.com/api.


## For agents

- This comparison as JSON: https://www.anchorterminal.com/compare/beliefstate-vs-eodhd.json, and with the fewest tokens: https://www.anchorterminal.com/compare/beliefstate-vs-eodhd.min.md
- Over MCP at https://www.anchorterminal.com/mcp (no key): `compare_tools {"a": "beliefstate", "b": "eodhd"}`. From a terminal: `anchor compare beliefstate eodhd`
- Each listing in full: https://www.anchorterminal.com/api/v1/tools/beliefstate.json and https://www.anchorterminal.com/api/v1/tools/eodhd.json

## Other comparisons with BeliefState or EODHD

- [Alpha Vantage vs EODHD](https://www.anchorterminal.com/compare/alpha-vantage-vs-eodhd.md)
- [Databento vs EODHD](https://www.anchorterminal.com/compare/databento-vs-eodhd.md)
- [Alpha Vantage vs BeliefState](https://www.anchorterminal.com/compare/alpha-vantage-vs-beliefstate.md)
- [BeliefState vs Benzinga](https://www.anchorterminal.com/compare/beliefstate-vs-benzinga.md)
- [BeliefState vs Eulerpool Financial Data API](https://www.anchorterminal.com/compare/beliefstate-vs-eulerpool.md)
- [BeliefState vs Financial Datasets](https://www.anchorterminal.com/compare/beliefstate-vs-financial-datasets.md)
- [BeliefState vs Financial Modeling Prep](https://www.anchorterminal.com/compare/beliefstate-vs-financial-modeling-prep.md)
- [BeliefState vs Finnhub](https://www.anchorterminal.com/compare/beliefstate-vs-finnhub.md)
- [BeliefState vs Intrinio](https://www.anchorterminal.com/compare/beliefstate-vs-intrinio.md)
- [BeliefState vs TickerLayer](https://www.anchorterminal.com/compare/beliefstate-vs-tickerlayer.md)
- [BeliefState vs Tiingo](https://www.anchorterminal.com/compare/beliefstate-vs-tiingo.md)
- [BeliefState vs Twelve Data](https://www.anchorterminal.com/compare/beliefstate-vs-twelve-data.md)
- [EODHD vs Eulerpool Financial Data API](https://www.anchorterminal.com/compare/eodhd-vs-eulerpool.md)
- [EODHD vs Financial Datasets](https://www.anchorterminal.com/compare/eodhd-vs-financial-datasets.md)
- [EODHD vs Financial Modeling Prep](https://www.anchorterminal.com/compare/eodhd-vs-financial-modeling-prep.md)
- [EODHD vs Finnhub](https://www.anchorterminal.com/compare/eodhd-vs-finnhub.md)
- [EODHD vs Intrinio](https://www.anchorterminal.com/compare/eodhd-vs-intrinio.md)
- [EODHD vs Massive (formerly Polygon.io)](https://www.anchorterminal.com/compare/eodhd-vs-massive.md)
- [EODHD vs TickerLayer](https://www.anchorterminal.com/compare/eodhd-vs-tickerlayer.md)
- [EODHD vs Tiingo](https://www.anchorterminal.com/compare/eodhd-vs-tiingo.md)
- [EODHD vs Twelve Data](https://www.anchorterminal.com/compare/eodhd-vs-twelve-data.md)
- [Benzinga vs EODHD](https://www.anchorterminal.com/compare/benzinga-vs-eodhd.md)

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
