A reverse stock split is a corporate action, and every listed company that does one has to say so in an SEC filing: the board fixes the ratio, the company announces the effective date, and the charter amendment is filed. That makes reverse splits a good dataset to work with in Python, because every row can be traced back to a document on EDGAR.
This tutorial uses the free reverse-splits CSV that Signal8 publishes at signal8.ai/datasets/reverse-splits. It is licensed CC BY 4.0 and built only from SEC filings, so you can use it in your own projects as long as you credit the source.
What is in the file
One row per SEC filing in which a Nasdaq- or NYSE-listed company fixed a reverse split ratio, announced its effective date, or filed the charter amendment. The columns you will use most:
| column | what it is |
|---|---|
filing_date |
date EDGAR accepted the filing |
ticker, company_name, cik
|
the issuer, from its SEC EDGAR record |
event_types |
ratio_fixed, date_announced, charter_filed (semicolon-separated) |
ratio_1_for |
N in a 1-for-N split, as stated in the filing |
effective_date |
the effective date the filing states (blank if none, never estimated) |
exchange_named_in_filing |
Nasdaq, NYSE or NYSE American, read from the filing text |
filing_url |
the EDGAR index page for the filing |
The full data dictionary is on the dataset page. When I wrote this (November 20, 2026) the file had 964 filings from 535 issuers, covering December 1, 2025 to November 19, 2026. It is rebuilt at most every 6 hours, so your numbers will differ.
Load it
You only need pandas:
pip install pandas
import pandas as pd
URL = "https://signal8.ai/datasets/reverse-splits.csv"
df = pd.read_csv(
URL,
dtype={"cik": str}, # keep the leading zeros
parse_dates=["filing_date", "effective_date", "market_effective_date"],
storage_options={"User-Agent": "Mozilla/5.0 (reverse-split-screen tutorial)"},
)
print(len(df), "filings,", df["cik"].nunique(), "issuers")
Read cik as a string, or pandas turns 0001009891 into 1009891.
Screen 1: splits taking effect in the next two weeks
A company often files more than once about the same split, so keep the latest filing per issuer:
today = pd.Timestamp.today().normalize()
upcoming = df[df["effective_date"].between(today, today + pd.Timedelta(days=14))]
upcoming = (upcoming.sort_values("filing_date")
.drop_duplicates("cik", keep="last")
.sort_values("effective_date"))
print(upcoming[["effective_date", "ticker", "company_name", "ratio_1_for",
"exchange_named_in_filing"]].to_string(index=False))
On November 20, 2026 this printed 12 issuers with a stated effective date between November 20 and December 4. One of them had a blank ratio_1_for: the filing stated a date but not a fixed ratio. The file leaves that cell blank rather than guessing, so check ratio_range_min and ratio_range_max or open filing_url.
Screen 2: large ratios
recent = df[df["filing_date"] >= today - pd.Timedelta(days=90)]
big = recent[recent["ratio_1_for"] >= 50].sort_values("ratio_1_for", ascending=False)
print(len(big))
print(big[["filing_date", "ticker", "ratio_1_for", "form_type"]].head(10).to_string(index=False))
30 filings in the 90 days to November 20, 2026 stated a ratio of 1-for-50 or larger. Most of the largest were 6-Ks, the form foreign private issuers use. Across the whole file the median stated ratio was 1-for-12.
Screen 3: issuers that fixed a ratio more than once
fixed = df[df["event_types"].str.contains("ratio_fixed", na=False)]
span = fixed.groupby(["cik", "company_name"])["filing_date"].agg(["min", "max", "count"])
repeat = span[(span["max"] - span["min"]).dt.days > 90]
print(len(repeat))
This returned 16 issuers. Treat it as a list to read, not a finding: one CIK can be a fund trust that covers many funds, and a later filing can be a further step in the same split. Open the filings before you call anything a second split.
Screen 4: filings per month
monthly = df.set_index("filing_date").resample("MS").size()
print(monthly.to_string())
Counts ran from 50 filings in January 2026 to 102 in July 2026. The file starts in December 2025, so there is no year-over-year comparison yet.
Things to keep in mind
- The file covers filings whose text names Nasdaq, the NYSE or NYSE American. OTC-only issuers, and splits announced only in a press release, are not in it.
- A blank cell means the filing did not state the value. Nothing is estimated.
- There are no prices, volumes or market caps in the file. Join your own market data on
tickerif you need them, and remember a ticker can change after a rename. -
filing_urlis the source of every value in the row. When a number matters, read the filing.
Credit
If you publish anything built on the file, cite it as:
Signal8, "Nasdaq and NYSE Reverse Stock Splits from SEC Filings," https://signal8.ai/datasets/reverse-splits, CC BY 4.0. Source data: SEC EDGAR.
There is also a companion file of Nasdaq and NYSE listing deficiency notices (8-K Item 3.01 and the 6-K equivalent), with the same licence. Joining the two on cik lets you look at how often a bid price notice is followed by a reverse split.
The complete script is below. Nothing in this post is investment advice.
import pandas as pd
URL = "https://signal8.ai/datasets/reverse-splits.csv"
df = pd.read_csv(URL, dtype={"cik": str},
parse_dates=["filing_date", "effective_date", "market_effective_date"],
storage_options={"User-Agent": "Mozilla/5.0 (reverse-split-screen tutorial)"})
today = pd.Timestamp.today().normalize()
upcoming = df[df["effective_date"].between(today, today + pd.Timedelta(days=14))]
upcoming = upcoming.sort_values("filing_date").drop_duplicates("cik", keep="last").sort_values("effective_date")
print(upcoming[["effective_date", "ticker", "company_name", "ratio_1_for"]].to_string(index=False))
recent = df[df["filing_date"] >= today - pd.Timedelta(days=90)]
print(recent[recent["ratio_1_for"] >= 50].sort_values("ratio_1_for", ascending=False)
[["filing_date", "ticker", "ratio_1_for", "form_type"]].head(10).to_string(index=False))
fixed = df[df["event_types"].str.contains("ratio_fixed", na=False)]
span = fixed.groupby(["cik", "company_name"])["filing_date"].agg(["min", "max", "count"])
print(span[(span["max"] - span["min"]).dt.days > 90])
print(df.set_index("filing_date").resample("MS").size().to_string())
Top comments (0)