Alternative data datasets
Browse by provider, category, marketplace, access, delivery and license
Datamule, Teraflop AI, and Eventual jointly contributed a 590 gigabyte SEC EDGAR collection encompassing about 8 million filings and 43 billion tokens, assembled using the datamule-python library and the official datamule API.
ALEA Patent Claim Graph Reading questions about the dependency structure of real U.S. patent claim sets: which claim a dependent claim refers back to, the chain of claims it rests on, its independent claim, which and how many claims depend on a given claim, and yes / no questions about dependency. Every answer is computed deterministically from the claim text shown in the row. This dataset is part
A private collection of cleaned earnings-call transcript segments derived from a Motley Fool Kaggle release, split into 135,306 training rows tagged with ticker, exchange, date, and chunk text.
SEC-EDGAR filings, part f (copy) An unmodified copy of part of TeraflopAI/SEC-EDGAR. I keep it here as staging data for language-model pretraining. Source: TeraflopAI/SEC-EDGAR, at commit 43de32c Files: 587 files, 45.6 GB, in the source's form-type folders: 8-K (217), S-1 (216), 4 (76), 20-F (40), S-8 (25), 3 (6), 5 (4), plus the source's LICENSE file and two chart images from its top folder. Lice
The TREC Clinical Trials collections released in 2021, 2022, and 2023 are available through the TREC homepage, with corresponding papers for each year. Artur Guimarães ([email protected]) serves as the dataset curator, while inquiries regarding the original authors can be directed separately. The dataset addresses the task of linking patient profiles with appropriate clinical trials, containing entries formatted as JSON objects with fields including query identifiers, disease labels, and accompanying clinical text.
An aggregation of 1.3 million U.S. patent records each accompanied by a human-written abstractive summary, sorted into nine Cooperative Patent Classification groupings spanning human necessities, chemistry, textiles, and related domains.
Abuja Housing Prices — Groundwork Data v1 Dataset Description This is the first public, structured dataset of Abuja residential property prices — covering both rental and sale listings across 16 areas of Nigeria's Federal Capital Territory. Collected by Groundwork Data, a Nigerian civic data project building the structured, domain-specific datasets that Nigerian ML engineers, researchers, and star
A large-scale release produced jointly by Datamule, Teraflop AI, and Eventual covering roughly 590 GB of SEC EDGAR content. It comprises about eight million filings totaling 43 billion tokens, gathered using the datamule-python package and the official datamule API.
Google Patents Chemistry Terminology Benchmark This repository contains the completed google_patents_chemistry_250_luna_scratch benchmark: 4,500 segment pairs across 18 ordered translation directions, with 250 examples per direction. It is an evaluation resource, not model translation output. Generation completed on 12 October 2026. Files and Usage google-patents-18-directions-4500-pairs-manifest.
NEAR Intents On-Chain Events Decoded on-chain events from intents.near on NEAR Protocol mainnet, ingested via the FastNEAR neardata API. Part of the NEAR Hackathon 2026 Investment Thesis submission. Contents events.parquet — structured table, one row per decoded dip4 event, viewable in the Dataset Viewer above events.jsonl — same data as newline-delimited JSON Updated automatically as ingestion pr
S&P 500 earnings-call transcripts in Parquet under MIT, tens of thousands of calls. Independently compiled and comparable to other public transcript corpora for sentiment and tone work.
FDA Form 483 Clinical Trial Protocol Violation Bench High-Consequence Enterprise Evaluation Benchmark Official Maintainer: Botmartz AI Research (botmartz.com)Research Publication: botmartz.com/insightsPrimary Metric: Adverse Deviation Extraction F1 Problem Overview In high-consequence enterprise applications within BioTech & Pharma, standard out-of-the-box LLMs and naive dense vector RAG fail on m
CAR-T clinical trials registered worldwide Registered clinical trials of CAR-T cell therapies, from 1999 to 20 October 2026. The records come from the WHO International Clinical Trials Registry Platform (ICTRP) and ClinicalTrials.gov, and are grouped into unique trials. Each trial is described by its countries, World Bank income group, phase, status and sponsor type (academic, public, mixed, for-p
A sample of US patent titles, abstracts and CPC classification labels built for multi-class patent classification, tens of thousands of text records in Parquet. Useful as training material for mapping innovation activity onto technology categories over time.
China Multiple Myeloma Clinical Trials — Open Dataset Multiple myeloma clinical trials registered in China, curated from official NMPA / CDE filings by the China Myeloma Digital Network (CMDN), an independent non-profit patient advocacy organisation. This is a mirror. The citable version of record lives at doi.org/10.5281/zenodo.22690814; the source repository is chinamyeloma/china-myeloma-clinica
A collection of earnings call transcripts covering S&P 500 firms and other U.S. large-cap companies across the years 2005 through 2025, intended for financial analysis, NLP modeling, and sentiment research.
A small text-classification dataset by ClarusC64 that labels whether borrow-rate, short-interest and price signals cohere into a real short squeeze or diverge into a false signal. It is published on a model hub with an MIT license tag and a 10-row train split.
Mechanical Utility Patents — 100 HTML Records A curated collection of 100 distinct granted U.S. utility patents from 100 patent families, saved as HTML for mechanical-engineering document analysis. Published by cmuchancel. Source disclosure: the patent text, bibliographic metadata, and drawings were obtained from Google Patents. Original PDFs and drawing assets are hosted on patentimages.storage.g
Records of initial public offerings launched on Indian markets between 2006 and 2025, with fields covering open and close dates, listing date, face value, issue price and size, lot size, first-day price, total shares offered and their allocation across anchor, NII, QIB and retail categories, minimum investment, and subscription figures for each investor class.
CryptoKeyhole on-chain daily metrics for Bitcoin and Ethereum Daily on-chain series for Bitcoin and Ethereum, calculated by CryptoKeyhole from full nodes it runs itself, plus the daily history of CryptoKeyhole's own Fear & Greed Index for the crypto market, Bitcoin and Ethereum. This is a snapshot taken on 30 October 2026 from CryptoKeyhole's free public API. The same series, updated every hour, a
A tabular extract of clinical study descriptions drawn from ClinicalTrials.gov on 5 February 2025, emphasizing eligibility, design, and objective fields for healthcare and machine learning research.
Vector embeddings of clinical-trial records, hundreds of thousands of entries in Parquet (Apache 2.0). Enables similarity and retrieval analysis of trial design, indication crowding and competitive positioning.
SEC Filings Structured Signals A structured dataset derived from real SEC EDGAR filings, containing information extracted from corporate disclosures submitted to the U.S. Securities and Exchange Commission. The dataset transforms filing text into structured signals covering financial risks, hedging strategies, litigation, supply-chain dependencies, causal relationships, entity relations, and manag
DAPFAM patent documents, a curated sample of US patents in Parquet with text for retrieval and classification tasks (CC BY-NC-SA 4.0). Scopus of claim text for building patent-similarity and technology-mapping models.
View source information and access options.
edgar_xbrl_companyfacts is a Hugging Face dataset published by DenyTranDFW that aggregates U.S. SEC EDGAR XBRL company-facts filings into a single Parquet file. The training split contains roughly 125 million rows of structured financial facts tagged by reporting period and unit.
Dataset Factory · v2026.10 Dataset website · GitHub release · Hugging Face · Catalogue Artifact: 2c1dade2913d0a8b643c89e628095a18e03223a6167beb9363316389486a544d Suitability and evaluation requirements Retained for retrospective model-development research. Each row is one distinct trade mark application whose features are its own filing-time attributes as recorded in the single frozen 2018-02-13 I
Prepared by Electric Sheep Africa, this MDPA-sourced dataset supplies 10 entries of hotel room occupancy rates for Mauritius over 2019-2023, delivered as ML-ready Parquet files accompanied by uniform Hugging Face metadata and source provenance.
A small Parquet-format collection from Electric Sheep Africa reporting room occupancy rates for large hotels in Mauritius from 2019 to 2023, containing ten records drawn from MDPA and packaged with Hugging Face metadata for machine learning workflows.
Assembled by Electric Sheep Africa from MDPA, this dataset provides 60 rows of monthly hotel room occupancy rates for Mauritius spanning 2019-2023, formatted as ML-ready Parquet with consistent Hugging Face metadata and source attribution.
Compiled by Electric Sheep Africa, this MDPA-derived dataset offers 20 records of quarterly hotel room occupancy rates for Mauritius across the 2019-2023 period, distributed in ML-ready Parquet format with standardized Hugging Face metadata and source traceability.
Electric Sheep Africa releases 60 quarterly observations from MDPA on hotel room occupancy for large establishments in Mauritius from 2019 through 2023, formatted as Parquet with consistent metadata.
Structured numeric records parsed from corporate filings such as 10-K, 10-Q, and 8-K submissions to the U.S. SEC. The figures are packaged inside a compressed DuckDB instance made available through the open-source Datapond registry for fast querying.
A collection of 49,023 patent filings focused on distributed ledger technology is available, comprising roughly 1.296 billion tokens. The corpus is a component of the broader DLT-Corpus initiative and is intended to facilitate natural language processing research, innovation analysis, and patent examination within the distributed ledger field.
Indeed Job Postings (2026) is a 1,000-row sample of US Indeed listings joined to company firmographics, published by fact-den on Apify. License terms are not stated, and the publisher directs users to a commercial source actor for fresh or filtered data.
FINRA Short Interest, last 12 months Shares sold short for every exchange-listed and OTC security, twice a month, for the last 12 months (502,924 rows): short shares, change from the prior report, average daily volume and days to cover. import pandas as pd df = pd.read_parquet("hf://datasets/FinzData/finra-short-interest/data/short_interest.parquet") df[df.ticker == "GME"] In the full FinzData dat
US Earnings Release Dates from SEC 8-K Filings (2004 onward) Every earnings release filed on SEC Form 8-K Item 2.02 since 2004 (229,038 releases): filing time in New York time and whether it came before the open, during the session or after the close. Useful for event studies. import pandas as pd df = pd.read_parquet("hf://datasets/FinzData/sec-earnings-release-dates/data/earnings_dates.parquet")
Company Earnings Results 2019–2026: findings, method and data card Fliar Research · released October 2026 · licence CC BY-NC 4.0 How this dataset was acquired. Every row comes from the Radar API, Fliar's company-data API and MCP server. Radar reads business news continuously and extracts each reported earnings result into structured fields: company, period, fiscal year, currency, revenue, net inco
A sampled extraction of IPO-related tables from SEC filings between 1994 and 2026, delivering raw table HTML with provenance fields and targeting roughly 100 tables per year, with edge years possibly containing fewer valid extractions.
Text of IPO filings, including S-1 and related prospectuses, in JSON: hundreds of thousands of records in English (CC BY 4.0). Pre-IPO disclosure text is one of the few information sources available ahead of a listing.
HoldingsIntel Smart Money Open Data What hedge funds disclose to the SEC, as clean CSV files. A new 13F edition every quarter. Every row links to its SEC filing or to the HoldingsIntel page that shows it. Latest 13F edition: Q2 2026. The Q2 2026 Smart Money Report tells its story. What's inside Folder Edition Files data/13f/<quarter>/ Every 13F quarter, once the funds' filings are in stocks.csv, f
SEC Form S-1 IPO Filing Dataset | Free 50-Row Preview Curated by IT Altitudes Inc. Get the Complete 5,000-Record Dataset Full Excel & CSV Buyer Kit — $159 USD Purchase the Full Dataset on Gumroad About This Dataset Explore a curated reference dataset of 5,000 U.S. SEC Form S-1 registration filings covering 2010–October 2026. This repository provides a free 50-row preview to help researchers evalua
A consolidated, ready-to-use repository of 13F institutional holdings drawn from SEC EDGAR filings, capturing the most recent quarterly disclosures of more than 13,000 investment managers including hedge funds, mutual fund complexes, pension funds, banks, and family offices, with each record summarizing total assets under management and individual positions.
A joint release by Datamule, Teraflop AI, and Eventual provides the SEC-EDGAR dataset, comprising 590 GB of content drawn from all major filings in the SEC EDGAR database and totaling roughly 8 million samples and 43 billion tokens. The data was assembled using the datamule-python library and the official datamule API developed by John Friedman, with datamule serving as a Python package for collecting and manipulating such records.
A collection of 2,000 Malaysian property transaction records spanning every state offers a detailed look at the national housing market in 2025. Compiled from Brickz, a recognized source for real estate transaction data, it covers location, tenure, property type, median pricing, and transaction volumes.
sec-earnings-bench sec-earnings-bench measures how accurately language models extract GAAP figures from US earnings press releases: revenue, net income and diluted EPS, for the quarter and the fiscal year to date (the full year in a fourth-quarter release). Each release is the press-release exhibit of an 8-K filed on SEC EDGAR, and each figure is graded automatically against XBRL in the company's
Earnings-call transcripts from an IBM research project, tens of thousands of records with finance tags, in CC0. A filings-adjacent corpus for call-tone and question-and-answer analysis.
CAIA On-Chain Investigation Benchmark CAIA evaluates whether tool-using agents can investigate Ethereum activity from raw JSON-RPC evidence. The public viewer separates objectively scored questions from open-ended investigation questions. Configurations Configuration Public input Output style Primary capability objective_questions Transaction or block anchor A–E choice plus evidence Classification
Datamule partnered with Teraflop AI and Eventual to publish the SEC-EDGAR dataset, which covers all major filings from the SEC EDGAR database and amounts to 590 GB across about 8 million samples and 43 billion tokens. Collection relied on the datamule-python library and John Friedman's official datamule API, tools designed in Python for gathering and handling SEC filing data.
Metadata for every EDGAR filing submitted to the U.S. Securities and Exchange Commission from 1994 through December 14, 2024, derived from the quarterly master index files and including CIKs, issuer names, and form types.
Google patents 2024–2026: an index of 4,651 patent families An AI model rated 339 Google and DeepMind inventions from 2024–2026 as directly about search. Only 93 of them have a write-up in any of the three sources that track Google patents for SEO. Roughly three in four are not covered there. This dataset lists all 4,651 patent families with a publication between January 1, 2024 and September 17,
The Future of Patent Corpus Management One Document, Fully Decoded - Without Anyone Reading It The signal you cannot get from a search interface The highest-value competitor signal in a patent portfolio is not the invention. It is the amount of money and urgency a company committed to it. That signal is never in an abstract, and no search interface surfaces it. It comes out of this corpus for ever
Earnings Call LLM Insights 📚 Read the Full Story: For a deep dive into the methodology, the wildest moments we found, and key takeaways, check out the blog post:KnowTrend.ai: Auto-Grading Ten Years of Earnings Calls for Prescience and Delusion This dataset contains LLM-generated analysis of ~70,000+ earnings call transcripts. The analysis was performed using Kimi k2-0905-preview, focusing on extra
Earnings-call transcripts for S&P 500 companies, tens of thousands of calls in English with timestamps, in Parquet (MIT). Direct input for call-tone and question-and-answer sentiment analysis across the large-cap earnings cycle.
Poland job postings — daily counts and advertised pay by voivodeship and category Publisher: listingworks · Licence: CC BY 4.0 · Source: public job listings on Praca.pl, read once a day from the public list pages only · Personal data: none — the files contain counts and medians only. Period: 2026-10-02 to 2026-11-12 (40 days). Updated automatically once a day (in the 13:00 hour, Japan time, after
This is the frozen October sample (2–13 Oct 2026). The continuing daily series is at listingworks/poland-job-postings-daily. Poland job postings — daily counts and advertised pay by voivodeship and category (October 2026) Publisher: listingworks (Yusuke Suda, sole proprietor, Yokohama, Japan) · Licence: CC BY 4.0 · Source: public job listings on Praca.pl, read once a day from the public list pages
Metadata for hundreds of thousands of clinical trials in English, French and Spanish, structured in Parquet with study-level fields such as phase, condition, sponsor and status (Apache 2.0). Trial starts and phase transitions lead clinical development, so this works as an early indicator of biotech pipeline activity.
A corpus of around 1.3 million U.S. patent filings paired with human-authored abstractive summaries, organized into nine Cooperative Patent Classification categories ranging from human necessities to textiles and paper.
SEC Filing Search Queries Realistic search queries over SEC 10-K, 10-Q and 8-K filings, each with the one filing (and passage) that answers it. Built to measure filing search the way it is used: terse keyword searches, analyst questions with fiscal or relative periods, and follow-ups that leave the company implicit. Each query is a rewrite of a question from nvidia/Nemotron-SFT-Finance-Document-Gr
Patent All Claims EP granted claim sets (indepdendent and dependent claims). One claim per line. Claims from 20210915-20250806 Splits Train: 96% Validation: 2% Test: 2% Usage from datasets import load_dataset ds = load_dataset("mhurhangee/ep-patent-all-claims")