datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Stocks-Daily-Price
Stocks Daily Price
This dataset includes daily price data for various stocks.
25,986,919 rows over 7,764 symbols, 8 columns, covering 1962-01-02 to 2026-08-05. Refreshed monthly.
Strategies Built on This Data
2,401 papers in the Papers With Backtest catalogue declare this dataset as an input. 2,238 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.35, and 45% clear a t-statistic of 1.96 on their own sample… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Stocks-Daily-Price.Universe-Daily-Price
Universe Daily Price
The index of every symbol carried by the public daily price datasets, with the repository that holds it.
21,693 rows over 20,895 symbols, 4 columns. Updated by Papers With Backtest.
Why It Matters
This is the lookup table the other price datasets need:
Routing: A symbol on its own does not say which file holds it. repo_id answers that in one join, so a strategy that mixes equities, futures and rates loads from the right place without… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Universe-Daily-Price.ML-ArXiv-PapersThis dataset contains the subset of ArXiv papers with the "cs.LG" tag to indicate the paper is about Machine Learning.
The core dataset is filtered from the full ArXiv dataset hosted on Kaggle: https://www.kaggle.com/datasets/Cornell-University/arxiv. The original dataset contains roughly 2 million papers. This dataset contains roughly 100,000 papers following the category filtering.
The dataset is maintained by with requests to the ArXiv API.
The current iteration of the dataset only contains… See the full description on the dataset page: https://huggingface.co/datasets/CShorten/ML-ArXiv-Papers.Forex-Daily-Price
Forex Daily Price
This dataset includes daily price data for various FX pairs.
437,430 rows over 166 symbols, 7 columns, covering 1970-01-04 to 2026-07-31. Refreshed monthly.
Strategies Built on This Data
292 papers in the Papers With Backtest catalogue declare this dataset as an input. 260 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.21, and 34% clear a t-statistic of 1.96 on their own sample, against 48%… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Forex-Daily-Price.Indices-Daily-Price
Indices Daily Price
This dataset includes daily price data for various indices.
815,441 rows over 113 symbols, 8 columns, covering 1927-12-30 to 2026-08-03. Refreshed monthly.
Strategies Built on This Data
1,324 papers in the Papers With Backtest catalogue declare this dataset as an input. 1,226 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.49, and 62% clear a t-statistic of 1.96 on their own sample, against… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Indices-Daily-Price.daily-papers-statsETFs-Daily-Price
ETFs Daily Price
This dataset includes daily price data for various ETFs.
10,813,800 rows over 5,802 symbols, 8 columns, covering 1978-01-03 to 2026-08-03. Refreshed monthly.
Strategies Built on This Data
502 papers in the Papers With Backtest catalogue declare this dataset as an input. 476 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.55, and 70% clear a t-statistic of 1.96 on their own sample, against 48%… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/ETFs-Daily-Price.Bonds-Daily-Price
Bonds Daily Price
This dataset includes daily rate data for various bonds.
1,415,096 rows over 209 symbols, 7 columns, covering 1969-04-30 to 2026-07-31. Refreshed monthly.
Strategies Built on This Data
590 papers in the Papers With Backtest catalogue declare this dataset as an input. 560 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.48, and 59% clear a t-statistic of 1.96 on their own sample, against 48%… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Bonds-Daily-Price.researchscope-papers
ResearchScope Papers
Open CS research paper dataset maintained by ResearchScope.
Updated automatically via GitHub Actions.
Quick start
from datasets import load_dataset
ds = load_dataset("kishormorol/researchscope-papers", "papers", split="train")
print(ds[0])
See Usage below for per-source splits, instruction-tuning, and the per-section fine-tuning data.
Stats
34,912 papers (raw metadata) — 9,912 arXiv · 20,000 conference · 5,000 journal
174,112… See the full description on the dataset page: https://huggingface.co/datasets/kishormorol/researchscope-papers.writing-model-papers-2016-2021
writing-model-papers-2016-2021
Private snapshot of papers from 2016 through 2021 (2022 excluded), filtered to the venue catalog under venues/ in the writing_model project.
PDFs are open-access only (arXiv, CVF, NeurIPS, PMLR, ACL Anthology, USENIX, JMLR). Paywalled publisher copies were not collected. The PDF tree stopped at a 48 GB disk budget.
Layout
path
contents
metadata/*.jsonl
one file per venue: title, year, authors, abstract, doi, arxiv_id… See the full description on the dataset page: https://huggingface.co/datasets/JonesLin/writing-model-papers-2016-2021.Stocks-Quarterly-Earnings
Stocks Quarterly Earnings
This dataset includes quarterly earnings report data for various stocks.
355,371 rows over 6,406 symbols, 8 columns, covering 1996-01-31 to 2026-07-31. Refreshed monthly.
Strategies Built on This Data
490 papers in the Papers With Backtest catalogue declare this dataset as an input. 456 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.20, and 32% clear a t-statistic of 1.96 on their own… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Stocks-Quarterly-Earnings.Commodities-Daily-Price
Commodities Daily Price
This dataset includes daily price data for various commodities.
508,406 rows over 63 symbols, 7 columns, covering 1927-12-30 to 2026-07-31. Refreshed monthly.
Strategies Built on This Data
259 papers in the Papers With Backtest catalogue declare this dataset as an input. 231 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.39, and 53% clear a t-statistic of 1.96 on their own sample… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Commodities-Daily-Price.Stocks-1Min-Price
Stocks 1min Price
This dataset includes 1-minute price data for various stocks.
5,710,703,934 rows, 7 columns, covering 2010-01-04 to 2026-07-31. Refreshed monthly.
Why It Matters
This dataset enables intraday equity strategy research and execution modeling by:
Intraday signal research: 1-minute bars enable microstructure-aware signals, VWAP tactics, and short-horizon alphas.
Realistic execution modeling: High-frequency OHLCV supports slippage and impact studies… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Stocks-1Min-Price.Cryptocurrencies-Daily-Price
Cryptocurrencies Daily Price
This dataset includes daily price data for various cryptocurrencies.
263,889 rows over 88 symbols, 7 columns, covering 2010-07-14 to 2026-08-14. Refreshed monthly.
Strategies Built on This Data
84 papers in the Papers With Backtest catalogue declare this dataset as an input. 73 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.47, and 42% clear a t-statistic of 1.96 on their own… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Cryptocurrencies-Daily-Price.ICLR_2025_Accepted_PapersPaper Decision Results for ICLR 2025
ICLR 2025 Accepted Paper List
https://openreview.net/group?id=ICLR.cc/2025/Conference#tab-accept
All-Monthly-CoreEconomicData
All Monthly Core Economic Data
Monthly macro indicators covering labor, credit, housing, volatility, policy uncertainty, and more.
786 rows, 276 columns, covering 1959-02-28 to 2024-07-31. The most recent observation is 2024-07-31.
Strategies Built on This Data
212 papers in the Papers With Backtest catalogue declare this dataset as an input. 204 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.50, and 60% clear… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/All-Monthly-CoreEconomicData.top-ml-conference-papers-2024
Top ML Conference Papers 2024
A dataset of 9,817 papers from ACL 2024, CVPR 2024, ICLR 2024, and NeurIPS 2024, processed with KNOWHERE.
Series
This collection of KNOWHERE-processed paper datasets is continuously expanding. New datasets and editions will be released — stay tuned!
Top ML Conference Papers
2023
https://huggingface.co/datasets/JensCS/top-ml-conference-papers-2023… See the full description on the dataset page: https://huggingface.co/datasets/JensCS/top-ml-conference-papers-2024.Stocks-Quarterly-FactorSignals
Stocks Quarterly FactorSignals
Quarterly factor exposures and scores for US equities.
763,836 rows over 22,791 symbols, 35 columns, covering 1998-03-31 to 2026-06-30. Refreshed monthly.
Strategies Built on This Data
53 papers in the Papers With Backtest catalogue declare this dataset as an input. 52 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.52, and 58% clear a t-statistic of 1.96 on their own sample… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Stocks-Quarterly-FactorSignals.ECCV2024-papersStocks-Weekly-ShortOverShorted
Dataset Information
Weekly Over-Shorted Stocks metrics for publicly traded companies, provided by SOV.AI.
Each record captures short interest positioning, changes, and context for individual tickers.
Coverage: Major US-listed equities with available short interest data
Update cadence: Weekly snapshots sourced from short interest models
Fields:
over_shorted: Relative measure of how over-shorted a ticker is
over_shorted_chg: Week-over-week change in the over-shorted measure… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Stocks-Weekly-ShortOverShorted.top-ml-conference-papers-2023
Top ML Conference Papers 2023
A dataset of 7,035 papers from ACL 2023, CVPR 2023, ICLR 2023, and NeurIPS 2023, processed with KNOWHERE.
Series
This collection of KNOWHERE-processed paper datasets is continuously expanding. New datasets and editions will be released — stay tuned!
Top ML Conference Papers
2023 (this)
https://huggingface.co/datasets/JensCS/top-ml-conference-papers-2023… See the full description on the dataset page: https://huggingface.co/datasets/JensCS/top-ml-conference-papers-2023.astrorag_papersiclr-rejected-papers-with-code-1k
Rejected ICLR Papers with Reviews and Code
This dataset contains 1,000 rejected ICLR submissions from 2018–2026. Each row
has the OpenReview submission metadata and reviews, the rejected submission PDF,
and a commit-pinned archive of a matched public GitHub repository.
This collection was built directly from OpenReview. It does not use a
third-party ICLR review dataset.
Project repository: TheAppliedScientist
Contents
1,000 unique rejected OpenReview submissions… See the full description on the dataset page: https://huggingface.co/datasets/Vidushee/iclr-rejected-papers-with-code-1k.1m_papers_text
Generated from Research Library:
https://github.com/peytontolbert/Research_Library
Deduped Paper Text Dataset
This dataset merges an existing paper-text parquet export with one or more
backfill parquet shards, then keeps exactly one row per canonical_paper_id.
One row corresponds to one canonical arXiv paper. Where multiple versions are
available, the merge keeps the strongest text row according to the repo's
selection logic: prefer non-partial text, then prefer… See the full description on the dataset page: https://huggingface.co/datasets/PeytonT/1m_papers_text.tsiolkovsky-papers
Tsiolkovsky Papers: the complete personal archive as a machine-readable corpus
Machine transcriptions of all 51,008 sheets of fond 555 of the Archive of the
Russian Academy of Sciences — the personal archive of Konstantin Tsiolkovsky
(1857–1935), who derived the rocket equation and described the multistage rocket
decades before anyone could test either.
The archive had been scanned and put online, but without a catalogue you could
query, full-text search, or a dataset. This is… See the full description on the dataset page: https://huggingface.co/datasets/vladimirbesk/tsiolkovsky-papers.Stocks-Quarterly-InstitutionalTrading
Stocks Quarterly InstitutionalTrading
Quarterly institutional ownership and trading flows for US equities.
186,938 rows over 5,290 symbols, 112 columns, covering 2016-12-31 to 2026-06-30. Refreshed monthly.
Strategies Built on This Data
28 papers in the Papers With Backtest catalogue declare this dataset as an input. 27 of them have been coded and run over their own full history. The median replicated Sharpe ratio is +0.25, and 33% clear a t-statistic of 1.96 on… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Stocks-Quarterly-InstitutionalTrading.CVPR2024-papersStocks-Weekly-ShortVolume
Stocks Weekly ShortVolume
Weekly short-selling volume for US equities, split by the type of participant behind the trade.
2,569,339 rows over 6,402 symbols, 8 columns, covering 2011-01-07 to 2026-07-03. Refreshed monthly.
Why It Matters
Short volume is the flow side of short interest, and it arrives weekly rather than twice a month:
Pressure, not positioning: short_volume_ratio_exchange measures how much of the week's trading was sold short. It moves before short… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Stocks-Weekly-ShortVolume.CVPR2023-papersStocks-Weekly-EarningSurprise
Stocks Weekly Earnings Surprise
Weekly earnings surprise probabilities and outcomes for publicly traded companies.
2,204,032 rows over 6,067 symbols, 8 columns, covering 2016-12-30 to 2026-07-03. Refreshed monthly.
Why It Matters
This dataset supplies high-frequency earnings-surprise context for equity strategies by:
Pre-event positioning: Surprise probabilities guide sizing and hedging ahead of earnings announcements.
Post-event drift: Actual vs. estimated EPS… See the full description on the dataset page: https://huggingface.co/datasets/paperswithbacktest/Stocks-Weekly-EarningSurprise.