Seven Business Days: What FINRA’s 2026 Short Interest Schedule Puts Between Settlement and Publication

On September 6, 2026, the newest short interest figure available to the public describes positions as they stood on August 14. It has been public since August 25. It will not be replaced until September 10, so September 9 is the last full day it stands as the newest public number, and on that day it is 26 calendar days old.

That is not a backlog. It is the schedule working as written. FINRA’s 2026 short interest reporting calendar has 24 cycles, and every one of them places publication exactly seven business days after the settlement date the data describes. What moves is the calendar gap, which runs from 9 to 12 days.

Short interest is usually read as a state variable — how crowded is this position right now. The published series is a sampled variable with a fixed business-day lag and a staleness profile that can be worked out in advance from the dates FINRA already publishes. What follows does that arithmetic. FINRA publishes the schedule and the definitions; the lags, ages and session counts below are this article’s calculation.

Seven Business Days, Repeated Twenty-Four Times

The reporting deadline is written in business days. FINRA Rule 4560(a) states: “Reports shall be received by FINRA no later than the second business day after the reporting settlement date designated by FINRA.” The schedule page puts a clock on the same deadline: “All short interest positions must be reported by 6 p.m. Eastern Time on the second business day after the reporting settlement date designated by FINRA.”

Publication then follows five business days after the due date. That second leg is not stated as a rule anywhere on the schedule page, but it holds in all 24 rows of the 2026 table. January 15 settlement, January 20 due, January 27 publication. August 31, September 2, September 10. December 31, January 5, January 12. Two business days, then five, twenty-four times.

The arithmetic is tight enough that the market holiday calendar falls out of it. The June 30 cycle publishes on July 10, which is five business days after the July 2 due date only if July 3, 2026 is not a business day. The March 31 cycle publishes on April 10, which requires April 3 to be excluded as well.

In calendar days the same seven business days stretch and compress. The gap is 9 days in six of the 2026 cycles — April 15 to April 24, July 15 to July 24, September 15 to September 24, September 30 to October 9, November 30 to December 9, and December 15 to December 24. It is 12 days in four of them, including January 15 to January 27 and May 15 to May 27. The mean across the 24 cycles is 10.5 calendar days. A trader who budgets “about a week and a half” is right on average and wrong by up to a day and a half in either direction on any single cycle.

The Figure Keeps Ageing Until the Next One Lands

Publication lag is only half the picture, and it is the smaller half. Once a figure is published it stays the newest public number until the next publication date. Its age on the last day of that run is not 9 to 12 days but 23 to 27.

Plotting the age of the newest public figure across every calendar day from January 27, 2026 to January 11, 2027 produces a sawtooth with 23 teeth. Each tooth starts at the publication lag and climbs one day per day until the next publication resets it. Over those 350 calendar days the mean age is 17.6 days.

The practical form of that number: on a randomly chosen day in 2026, the short interest on the screen describes positions from roughly two and a half weeks earlier. Across those 350 days it sits closer to three weeks than to one week on 245 of them and closer to one week on 82, with the remaining 23 days landing exactly halfway.

The sawtooth also explains a common misreading of change. Two consecutive published figures are 14 to 18 calendar days apart at their settlement dates, not at their publication dates. A change of a few percent from one cycle to the next describes what happened between two settlement dates that may straddle an earnings release, an index rebalance or an expiration — events that sit inside the window and are invisible in the endpoints.

Days to Cover Divides by a Window That Changes Length

The second arithmetic problem is the denominator. FINRA’s glossary defines days to cover as “The number of days of average share volume it would require to buy all of the shares that were sold short during the reporting cycle.” The average in that sentence is defined immediately above it: “Total Volume or Adjusted Volume in case of splits / Total trade days between (previous settlement date + 1) to (current settlement date).”

That window is one reporting cycle long, and reporting cycles are not a fixed number of sessions. Counting trading sessions between consecutive 2026 settlement dates gives 9 in the shortest cycles and 12 in the longest. The February 14 to February 27 window holds 9 sessions, cut short by the February 16 closure. The March 14 to March 31 window holds 12. The denominator behind days to cover is therefore computed over a window whose longest 2026 instance is a third longer than its shortest.

Window length matters most when volume is uneven. Take a cycle in which one session prints five times the volume of the others. In a 9-session window the average becomes 13 normal sessions divided by 9, or 44.4 percent above normal; in a 12-session window it becomes 16 divided by 12, or 33.3 percent above normal. Days to cover falls by 30.8 percent in the first case and 25.0 percent in the second, from the same short interest number and the same single volume spike.

A vendor that computes days to cover from a 30-session or three-month average volume will not reproduce FINRA’s published ratio, because the windows differ. And a days-to-cover series read across cycles is a ratio whose denominator definition is stable but whose measurement window is not, so small moves in the ratio are not reliably moves in crowding.

The As-Of Date Is a Settlement Date, Not a Trade Date

One more session hides in the definition. FINRA’s glossary describes the reported figure as “The total number of shares in the issue that are reflected on the books and records of the reporting firms as short as defined by Rule 200 of Regulation SHO as of the current cycle’s designated settlement date.” The reference point is a settlement date.

Under SEC Rule 15c6-1 the standard settlement cycle is one business day after the trade date; the Commission’s announcement of the final rules states that “The compliance date for the final rules is May 28, 2024.” A short sale executed on the session before the designated settlement date settles on it and is captured. A short sale executed on the settlement date itself settles the following session and belongs to the next cycle. Execution to publication is therefore eight business days at minimum, not seven.

What Would Invalidate This

• Daily short sale volume is different data. FINRA publishes daily short sale volume files that are not short interest. Volume executed short on a given day says nothing directly about positions carried overnight, and the two series move independently.
• The cadence is under review. In SR-FINRA-2026-012, filed May 1, 2026 and published in the Federal Register on May 18, FINRA proposed to increase reporting frequency by “(1) requiring members to submit short interest reports on a weekly rather than a bi-monthly basis, and (2) reducing the two business-day reporting turnaround period to one business day to allow for a more streamlined and timely publication process for short interest reports.” The filing states that these modifications “would allow short interest data to be published weekly, five business days after the reporting settlement date” and argues they would give regulators and market participants a more current view of short interest. The Commission’s July 6, 2026 notice of designation “designates August 14, 2026, as the date by which the Commission shall either approve or disapprove, or institute proceedings to determine whether to disapprove, the proposed rule change (File No. SR-FINRA-2026-012).” That date has passed. Checked on September 6, 2026: a Federal Register search on the file number returns two documents, the May 18 notice of filing and the July 6 notice of designation, and the Commission’s rulemaking page for the filing lists two notices, the most recent dated June 30, 2026. No approval order, disapproval order or order instituting proceedings appears in either place as of that check. If the proposal is approved and implemented, the seven-business-day lag and the 17.6-day mean age that follows from it both change.
• The schedule is annual. Every date above comes from the 2026 table. The 2 + 5 business-day structure has to be re-checked against each year’s published schedule rather than assumed.
• Coverage is not universal. Rule 4560(a) applies to short positions carried on the books and records of FINRA members, and it carves out Restricted Equity Securities as defined in Rule 6420. Positions outside that reporting perimeter are not in the number.
• Staleness is not a signal. Nothing here says a stale figure is a wrong figure. In a name whose float and borrow are stable, a 20-day-old short interest reading may be perfectly serviceable. The argument is that the age is knowable and should be carried explicitly.

Concrete Framework

A checklist for reading a short interest figure with its age attached.

1. Write down the settlement date, not the publication date. The figure describes the settlement date. Publication tells you when you learned it.
2. Compute the age before using the number. Age equals today’s date minus the settlement date. In 2026 that runs from 9 days to 27 days depending on where in the cycle you are standing.
3. Check where you sit in the cycle. A figure read the day before the next publication is at its oldest; a figure read the day it lands is at its youngest. The same data source is roughly three times fresher at one end than the other.
4. Look at what happened inside the window. List the earnings dates, expirations and index events between the two settlement dates before treating a cycle-over-cycle change as a position change.
5. Match the days-to-cover denominator to the source. FINRA’s ratio uses the trade days between the previous settlement date and the current one. If a vendor’s ratio differs, the volume window is the first thing to check.
6. Size the position to the age, not to the ratio. A crowding estimate that is 17.6 days old on average is a slow input. Rules that require it to be current — entering purely because days to cover crossed a threshold, for instance — are asking more of the data than the publication schedule can deliver.
7. Re-run the arithmetic each January. The schedule is republished annually, and a proposed rule change is pending. Both the lag and the staleness profile are inputs that expire.

This article is for information only and is not investment advice, and it does not recommend buying or selling any security. Figures are drawn from FINRA’s published 2026 short interest reporting schedule, FINRA rule text and glossary definitions, and SEC filings, all as available on September 6, 2026.