FINRA Short Interest Data, Wey E Dey Come From

Almost every short-interest number wey you don read before, e go come from FINRA — di US broker-dealer regulator. FINRA dey publish two different short-selling datasets, and people dey mix dem together all di time.

One of dem na di short interest file: na di open short positions wey brokers report, dem dey compile am two times every month.

Di oda one na di daily short volume file: e come from trade reporting, dem dey publish am every day, but e dey measure something else.

Dis page go explain both of dem from di source — di schedule, wetin dey inside, how to read one real print line by line, where you fit find di data yourself, and di failure modes wey we don catch inside di files. (Short interest itself get im own explainer if you never sabi di concept before.)

The two-times-a-month short interest file

Broker-dealers dey report their customers' and their own open short positions to FINRA as of two settlement dates each month — mid-month and month-end — and the compiled totals per security dey disseminated for one published schedule. The requirement na regulation, no be convention: FINRA Rule 4560 dey make member firms report their short positions for all equity securities two times every month. That cadence dey verifiable for the data:

SELECT toYear(settlement_date) AS year,
       uniqExact(settlement_date) AS settlement_dates
FROM global_markets.stocks_short_interest
WHERE settlement_date >= '2018-01-01'
GROUP BY year
ORDER BY year

Every full year from 2018 through 2025 show exactly 24 settlement dates on file; 2026 stand for 12 so far. Each record carry three fields per security: the open short position (shares short), one average-daily-volume figure, and the days-to-cover ratio wey dem build from them. Two properties matter pass anything for the file: e be snapshot of open positions (the thing wey people really mean when dem talk "short interest"), and e be slow — positions as of the settlement date, dem publish am about two weeks later. The lag get im own page, with the publication delays wey dem measure for each print.

The 2026 settlement calendar

"Two times a month" resolve to specific dates: the 15th of each month (dem move am to the business day before am when the 15th fall on weekend or holiday) and the last business day of the month. Here na the calendar wey the file don produce for 2026:

SELECT toString(settlement_date) AS settlement,
       formatDateTime(settlement_date, '%W') AS weekday,
       uniqExact(ticker) AS securities_on_file
FROM global_markets.stocks_short_interest
WHERE settlement_date >= '2026-01-01'
GROUP BY settlement_date
ORDER BY settlement_date

12 settlements so far this year, from 2026-01-15 through 2026-06-30 — and the weekday column show the business-day adjustment dey work (one mid-February settlement on Friday the 13th, with the 15th on Sunday). The coverage column too dey worth to check: the file don grow from 21262 securities on the first print of the year to 22207 on the latest — every equity security with reportable position, no be only index names.

How to read one real short-interest print

See wetin the file really dey talk about one popular name wey everybody sabi: Apple. The five latest settlement records dey here, with the change from one print to the next one calculated alongside am:

SELECT settlement, shares_short_mn, adv_mn, days_to_cover, change_pct
FROM (
    SELECT settlement_date,
           toString(settlement_date) AS settlement,
           round(short_interest / 1e6, 1) AS shares_short_mn,
           round(avg_daily_volume / 1e6, 1) AS adv_mn,
           days_to_cover,
           round((short_interest - lagInFrame(short_interest) OVER (ORDER BY settlement_date))
                 / lagInFrame(short_interest) OVER (ORDER BY settlement_date) * 100, 1) AS change_pct
    FROM global_markets.stocks_short_interest
    WHERE ticker = 'AAPL'
      AND settlement_date >= '2026-04-15' AND settlement_date <= '2026-06-30'
)
WHERE settlement_date > '2026-04-15'
ORDER BY settlement_date

Read the latest row the way person wey don master the work go read am: 140.5 million shares short as at the 2026-06-30 settlement, one -2.6% change from the print wey come before, against one average daily volume of 81.1 million shares — 1.73 days to cover. Across the five prints, the position build through May (e go up 12.3% go reach the 2026-05-29 settlement, na the biggest swing for inside the table), then e cool down for two prints wey follow one after the other.

The column wey get sense na days-to-cover. E print 1.73 for the latest settlement — na the lowest among the five — while the position itself just drop small by -2.6%. Na the denominator move instead: the average-daily-volume figure jump from 52.3 million shares for the mid-June print go reach 81.1 million by the time month end. Days-to-cover dey divide one position wey no dey move fast by one volume average wey dey move fast, so if trading burst come, e fit squeeze the ratio sharp sharp without anybody covering at all — read all three columns, no be the ratio alone.

The daily short volume file — the one wey no be short interest

Every day, FINRA dey publish per-ticker "short volume": the part of that day reported trades wey the sell side mark as short. E dey DAILY and fresh — and na flow e be, no be position. Market makers dey short all the time as normal inventory management, so the marked-short percentages dey high — often well over one-third to pass half of the file volume — for ordinary stocks for ordinary days. That claim need receipt, so here na five household names for July 10, 2026, one normal Friday wey get complete file:

WITH covered AS (
    SELECT ticker,
           max(short_volume) AS short_vol,
           max(total_volume) AS covered_vol
    FROM global_markets.stocks_short_volume
    WHERE date = '2026-07-10'
      AND ticker IN ('AAPL', 'MSFT', 'NVDA', 'TSLA', 'KO')
    GROUP BY ticker
),
tape AS (
    SELECT ticker, sum(volume) AS tape_vol
    FROM global_markets.delayed_stocks_minute_aggs
    WHERE ticker IN ('AAPL', 'MSFT', 'NVDA', 'TSLA', 'KO')
      AND window_start >= toDateTime('2026-07-10 04:00:00', 'America/New_York')
      AND window_start < toDateTime('2026-07-10 20:00:00', 'America/New_York')
    GROUP BY ticker
)
SELECT covered.ticker AS ticker,
       round(covered.short_vol / 1e6, 1) AS marked_short_mn,
       round(covered.covered_vol / 1e6, 1) AS file_total_mn,
       round(covered.short_vol / covered.covered_vol * 100, 1) AS pct_marked_short,
       round(tape.tape_vol / 1e6, 1) AS consolidated_tape_mn,
       round(covered.covered_vol / tape.tape_vol * 100, 1) AS file_share_of_tape_pct
FROM covered
JOIN tape ON covered.ticker = tape.ticker
ORDER BY indexOf(['AAPL', 'MSFT', 'NVDA', 'TSLA', 'KO'], covered.ticker)

The five range from 38.8% marked short (Microsoft) to 57% (Coca-Cola) — and Coca-Cola, wey almost no be short-seller battleground as the market fit give, print the highest share of the group. Nothing special happen for these names that day — na so the file dey normally. Look the denominator too: the file total-volume column dey count only the reported volume wey the dataset cover, no be the whole market. Apple 11.3 million file-total shares compare with 27.9 million for the consolidated tape that day — the file see 40.5% of Apple tape, and the other four names dey the same range. To treat the daily file as fast short interest na the single most common short-data mistake for internet; the two-datasets comparison take am apart with worked example.

One vocabulary note. The file "exempt" columns dey count short-exempt volume — trades wey dem execute under exception to SEC Regulation SHO, the rule wey dey govern how short sales fit execute (the locate requirement and the circuit-breaker price test). Reg SHO na separate regime from position reporting, and e get im own daily artifact: the threshold list of securities wey get persistent settlement fails. FINRA dey host the daily short-volume files for im Reg SHO data pages, wey no help the two-datasets confusion at all.

How far back the data dey reach?

Two different answer for the two files — you gatz pin am down before you trend anytin:

SELECT toString(min(settlement_date)) AS settlement_file_start,
       uniqExact(settlement_date) AS settlement_prints,
       (SELECT toString(min(date)) FROM global_markets.stocks_short_volume) AS daily_file_start,
       (SELECT uniqExact(date) FROM global_markets.stocks_short_volume) AS daily_files
FROM global_markets.stocks_short_interest

For this warehouse, the settlement file don reach back to 2017-12-29 — na 205 settlement prints dey inside — while the daily short-volume history start from 2024-02-06 (we get 608 trading days on file). FINRA own archive still dey go further back; the depth wey dey up there na wetin every panel for this site dey use.

How e dey fail — proof from our own store

To read any data source with sense, you go first know how e fit break. Inside the last one month, our pipeline don catch the daily short-volume file commot cut-short two times — na only alphabetical part of the normal ticker world:

SELECT toString(date) AS d,
       uniqExact(ticker) AS tickers_on_file
FROM global_markets.stocks_short_volume
WHERE date IN ('2026-06-26', '2026-06-29', '2026-06-30', '2026-07-06', '2026-07-07', '2026-07-08')
GROUP BY date
ORDER BY date

The file wey June 29 carry get 5489 tickers, but normal day suppose be around 15052. The one for July 7 carry 4333 — both of dem na alphabetical cut-short, both happen on days wey news break, and our recap don flag am when e happen. The simple rule be: before you quote any single day short volume for any ticker, first check say that day file complete.

The twice-monthly file get her own wahala: dem fit revise am after. The firm wey dey report fit correct wetin dem submit, and a settlement print fit change after dem first publish am. Our warehouse dey keep the latest value for each print — once dem correct am, e go overwrite the original row — na why every page here dey re-run her queries when e regenerate: one revised print go update every number wey quote am, instead of am to just stand like stone.

How you fit check short interest by yourself

You no need to pay any product before you see all dis one — di main sources dey public:

• FINRA dey publish both data sets for dia website: consolidated equity short interest (you fit search am per security and download am per settlement date under FINRA Data) and di Daily Short Sale Volume files, dem dey post am every evening for dia Reg SHO data pages.
• The exchanges — Nasdaq website get per-ticker short interest lookup wey get print-by-print history for di companies dem list, and NYSE-listed totals dey flow through di same FINRA consolidation.
• Broker platforms and screeners — di "short interest" and "short % of float" fields wey most platforms dey show na repackaged versions of di same prints, sometimes dem put estimates inside between settlements. If number look like say e old, e fit just mean say e get settlement date wey don pass two weeks by design.

Which file you suppose use for wetin?

• "How shorted dis stock be?" — di twice-monthly short interest file, read am with im date. Di most-shorted leaderboards dem build from im latest print.
• "Shorts dey press today?" — no public tin clean answer dis one. Di daily file na di closest thing wey dey, and e mostly dey measure market-making plumbing; use am for patterns across days, never as same-day position estimate.
• "When new data go come?" — settlements dey twice monthly for di published calendar wey dey up; daily files dey every evening for di session wey just pass.

FINRA short interest FAQ

How often does FINRA publish short interest?

Open short positions dey reported under FINRA Rule 4560 as of two settlement dates per month — the 15th (or the business day before am) and the last business day for that month, twenty-four per year — and dem dey publish am roughly two weeks after each settlement. The daily short volume file na different dataset, dem dey publish am every trading day.

Where can I look up short interest for a specific stock?

FINRA website dey publish the consolidated short interest data per security, Nasdaq site get per-ticker lookup wey get print history, and most broker platforms dey show the latest print as "short interest" or "short % of float." No matter the source, check the settlement date wey dey attached to the number — every one of them trace to that same twice-monthly file.

Is FINRA short interest data free?

Yes — FINRA and the exchanges dey share both files publicly. Most paid "short interest" products na repackaging (or estimating between) these same prints.

How far back does short-interest data go?

This site warehouse get every settlement print since 2017-12-29 — 205 of them — and daily short-volume files since 2024-02-06. FINRA own archives reach back further still.

Why does short volume show 50% short when short interest small?

The daily file dey mark the sell side of reported trades, and market makers dey sell short constantly as inventory plumbing — e be flow statistic wey intermediation dey dominate, no be position statistic. For one ordinary Friday, Coca-Cola print 57% of im reported volume marked short. The two numbers dey measure different things and dem dey routinely diverge by order of magnitude.

Is short interest the same as the Reg SHO threshold list?

No. Short interest na twice-monthly count of open short positions wey dem report under FINRA Rule 4560. The Reg SHO threshold list na daily list of securities wey get persistent failures to deliver at settlement, wey dem produce under one SEC rule about short-sale execution. One stock fit heavy short and never show for threshold list.

Can FINRA short interest data wrong?

E fit late, truncated, or revised — all of them wey we don document above from our own archive. Treat every print as "as reported, as of im date," and prefer sources wey show which file and date one number come from.

Every panel na one stored, versioned query over the files themselves — expand the SQL, or audit any date coverage for the Strasmore terminal.