Home › Datasets

Every fresh 10-K and 10-Q risk section as clean text

For: Equity and credit analysts, financial NLP and RAG builders, risk and compliance researchers, and AI agents reading filings.

Riskroll cuts Risk Factors (Item 1A), MD&A, market risk, Item 1C cybersecurity and Business sections out of recent 10-K and 10-Q filings and cleans them to plain text: one row per filing and section, with ticker, form and filing date. Filings it could not parse get a status row, so you know what is missing.

Riskroll
10-K risk factors dataset, by CyberMax

The dataset is free (CC0) on Hugging Face and Kaggle. Paid next step for the numbers behind the text: Figurewell API, point-in-time SEC financials, 100 free calls a day, a $5 pack of 1,500 calls or $19/month. Paid option for work use: the CyberMax Dataset Commercial License, $49/year for a company of up to 10 people ($199/year up to 250, $499/year any size). It covers every CyberMax dataset and adds an invoice, a PDF certificate in your company's name, named email support and a freshness commitment with service credits.

Financials API: $5 for 1,500 calls Licence for work: $49/year (up to 10 people)Browse the free 10-K sections All plans

14-day money-back promise: if it doesn't work as described, email us within 14 days and we fix it or refund you in full. Refund terms.

Figurewell API live demo: SEC financials as they were known, on any date (15 s, no sound)

sections filed 12 Aug to 23 Sep 2026, word counts of the extracted text

Ticker	Company	Form	Filed	Section	Words
CRWD	CrowdStrike Holdings	10-Q	2026-08-27	risk_factors	26,225
AVGO	Broadcom Inc.	10-Q	2026-09-10	risk_factors	14,003
LULU	lululemon athletica	10-Q	2026-09-03	risk_factors	10,644
ADBE	Adobe Inc.	10-Q	2026-09-22	risk_factors	10,294
PANW	Palo Alto Networks	10-K	2026-09-10	cybersecurity	1,548

Source: Riskroll sections table, build covering filings of 12 Aug-23 Sep 2026 (products/riskroll/store/listing.json; HF dataset CyberMax-tools/riskroll-sec-10k-10q-sections). Source: SEC EDGAR. Information only, not investment advice.

Riskroll vs the alternatives

Published prices, each checked on the date shown; prices change, so confirm on each site.

Product	Price	Free	Checked
Riskroll dataset (+ Figurewell API)	Dataset free; Figurewell API $5 for 1,500 calls or $19/month; work licence from $49/year	the full section text, CC0	live
sec-api.io (Extractor API)	Personal & Startups $49/month ($588/year); Business $199/month	first 100 API calls	2026-10-10
Apify: SEC EDGAR Filings Intelligence (benthepythondev)	from $23.20 per 1,000 results (filing metadata and links, no section text)	Apify free monthly credit	2026-10-10

Why teams pick Riskroll

Section text already cut and cleaned, free; sec-api.io's extractor sits in plans from $49/month.
Includes 10-Qs, not only annual reports.
Item 1C cybersecurity disclosures in their own rows.
A status row for every filing it could not parse: honest coverage you can measure.

Try it in one call

import pandas as pd
df = pd.read_parquet('https://huggingface.co/datasets/CyberMax-tools/riskroll-sec-10k-10q-sections/resolve/main/data/sections-2026-09.parquet')
print(df.head())

FAQ

Is there a free dataset of 10-K risk factors?

Yes. Riskroll has Item 1A and other sections as clean text, CC0. The build covering 12 Aug-23 Sep 2026 has 2,098 filings cut into 5,245 sections, including 1,560 Risk Factors and 173 Item 1C cybersecurity sections.

Why not sec-api.io?

sec-api.io extracts any section on demand, in real time, inside plans from $49/month. Riskroll is free and already extracted for recent filings; use sec-api.io when you need any filing on demand.

Why not EDGAR-CORPUS on Hugging Face?

It is a large historical corpus of annual reports last updated in 2023. Riskroll covers 2026 filings and includes 10-Qs.

What is not covered?

Older filings and any filing the parser could not split; those appear as status rows.

Do I have to pay to use it?

No. The data keeps its open licence for everyone, at work too. The $49/year licence is for companies that want an invoice, a certificate in their name, named support and service levels; it covers every CyberMax dataset.