Data

Data: 30 Dec 2013 to 15 Nov 2026 (672 weeks) · Last updated: 16 Nov 2026 · next update Monday 23 Nov 2026

All series are updated every Monday with the ISO week that closed on Sunday. Each update is published only after an automated validation gate passes: data freshness, coins per day, Bitcoin present, agreement between CoinGecko and CoinMarketCap, and plausibility of the factors. Files are available here and in the Hugging Face dataset under CC BY 4.0. Please cite Stoeckl & Pukrop (2026), Liu et al. (2022) and Fieberg et al. (2024).

Pick a file

The standard factors

These are the factors to use as a benchmark for investable crypto portfolios; see The Standard for why.

Source Series CSV Parquet
CoinGecko Standard (recommended) factors_cg_investable.csv .parquet
CoinGecko Standard, 1-day lag factors_cg_investable_lag1.csv .parquet
CoinMarketCap Standard factors_cmc_investable.csv .parquet
CoinMarketCap Standard, 1-day lag factors_cmc_investable_lag1.csv .parquet
Column Description
week_start, week_end first and last day of the week (Monday, Sunday; one day later for _lag1)
CMKT market: value-weighted return of eligible coins minus RF
CSIZE size: small minus big (terciles, breakpoints from coins ≥ $100M)
CMOM momentum: winners minus losers on the return of the previous two weeks
RF 3-month T-bill over the same days (365-day accrual)
n_coins eligible coins (previous-week average cap ≥ $1M)

Returns are simple weekly returns in decimals (0.01 = 1%).

Daily and monthly

The same portfolios at daily and monthly frequency, in the spirit of Kenneth French’s data library. Portfolios are formed every ISO week exactly as above. Each coin’s weight then drifts with its price during the week, so the daily returns of a week compound exactly to the weekly series. Monthly returns compound the daily returns over complete calendar months (the portfolios are still rebalanced weekly). In both, the long-short and excess returns are formed at that frequency (long − short, market − RF).

Source Daily Monthly
CoinGecko csv · parquet csv · parquet
CoinMarketCap csv · parquet csv · parquet

Daily files have date (end-of-day UTC) and monthly files have month, month_start and month_end instead of the week columns. Daily and monthly series exist for the standard only. A month appears once all its days are available.

Other calendars

The standard on every week calendar. Use the one on which your own weekly returns are measured; lag 1 shifts the window by one day.

Week CoinGecko CoinMarketCap
Sunday close (ISO weeks) — standard csv · lag 1 csv · lag 1
Monday close csv · lag 1 csv · lag 1
Tuesday close csv · lag 1 csv · lag 1
Wednesday close csv · lag 1 csv · lag 1
Thursday close csv · lag 1 csv · lag 1
Friday close csv · lag 1 csv · lag 1
Saturday close csv · lag 1 csv · lag 1
Liu et al. calendar (Jan 1–7 = week 1) csv · lag 1 csv · lag 1

For comparison

Series CoinGecko CoinMarketCap
Standard design with all-coin breakpoints csv csv
Liu et al. (2022) exact, all-coin breakpoints csv csv
Liu et al. (2022) exact, breakpoints ≥ $100M csv csv

All have the same columns as the standard, on ISO weeks. For the Liu et al. series, CSIZE is their CSMB.

Research release

The Hugging Face dataset also holds the research release behind the standard:

Folder Contents
crosswalk/ our coin id (uid) ↔︎ CoinMarketCap id ↔︎ CoinGecko id; match_confidence high / medium / low for price-confirmed pairs, cg_only / cmc_only for coins listed by one provider; empty CSV fields are missing values
multiverse/ the 249,024 factor sets: specification, weekly factors, Sharpe ratios, non-standard errors
evaluation/ fairness, pricing and out-of-sample results per factor set; finalists, bootstrap, spanning tests
test_assets/ 18 passive investable portfolios and investable characteristic portfolios, all calendars

Use in R / Python

f <- read.csv("https://opencryptopricing.com/data/factors_cg_investable.csv")

# alpha of your weekly returns (data frame `my` with week_start and ret, same ISO weeks)
d <- merge(my, f, by = "week_start")
summary(lm(I(ret - RF) ~ CMKT + CSIZE + CMOM, data = d))
import pandas as pd
f = pd.read_csv("https://opencryptopricing.com/data/factors_cg_investable.csv")

# all files at once
from huggingface_hub import snapshot_download
snapshot_download("sstoeckl/opencryptoassetpricing", repo_type="dataset", allow_patterns="factors/*")

The underlying coin data can be retrieved with the crypto2 R package (CoinMarketCap and CoinGecko).

Coverage and licence

  • Coverage: 2014 to the last complete ISO week. Before 2018 only a handful of coins were worth $100M, so the size and momentum legs are thin; we evaluate the factors from 2018 onward.
  • Licence: our factor series, multiverse, evaluation and crosswalk are released under CC BY 4.0. The underlying market data are from CoinGecko and CoinMarketCap and subject to their terms.
@misc{stoeckl2026opencrypto,
  author      = {Stoeckl, Sebastian and Pukrop, Moritz},
  title       = {Open Crypto Pricing},
  year        = {2026},
  institution = {University of Liechtenstein},
  url         = {https://opencryptopricing.com}
}

References

Fieberg, C., Günther, S., Poddig, T., & Zaremba, A. (2024). Non-standard errors in the cryptocurrency world. International Review of Financial Analysis, 92, 103106.
Liu, Y., Tsyvinski, A., & Wu, X. (2022). Common risk factors in cryptocurrency. The Journal of Finance, 77(2), 1133–1177. https://doi.org/10.1111/jofi.13119
Stoeckl, S., & Pukrop, M. (2026). Open crypto pricing. University of Liechtenstein. https://huggingface.co/datasets/sstoeckl/opencryptoassetpricing