ORION Biotech Catalyst Calendar
Menu
Skip to content
ABOUT THE FOUNDER

Mission: Give humanity and its AIs a memory of medicine that can't be quietly rewritten.

Who builds ORION, why, and how fast.

I'm Jiwoo Seong (Woo). I am a student at Dong-A University in Korea, and I trade biotech stocks with my own money. I build ORION, the Biotech Catalyst Calendar, alone, directing several AI assistants, and I compare every published row with the original filing myself.

Formerly ORION Biotech Observatory.

Last updated

You asked, we changed

Who

I'm Jiwoo Seong (Woo), a student at Dong-A University in Korea. I trade biotech stocks with my own money, and since 2026-09-22 I have sealed a daily ledger for a quant paper-trading account whose rules were fixed before its first order (founder log; the ledger itself is private).

Why I am building this

On August 19, 2026, Moderna more than doubled in a day on one clinical readout (Forbes). I wanted to find the next Moderna, and that day I decided to go all in on this.

I built detectors first. They flagged 2 of 16 companies. Taking the failure apart, I found something else. When companies move a readout, they don't issue a press release. One line in a quarterly filing quietly changes from "third quarter" to "fourth quarter." And I couldn't find one public place that kept those changes together.

So I built ORION. It records the earliest statement we found from a company, every change since, and what was actually disclosed, with the original filings and dates. No scores, no forecasts. Readers judge for themselves.

My claim is simple: a company's past promises are the best evidence for how much to trust its next one. Catalyst calendars tell you the next date. ORION keeps the verified history of every promise.

As the record grows, it shows how often each company kept its word and which programs were quietly abandoned. That list is where a researcher looking to revive a drug could start. Further out, I want this verified record to be what AIs learn medicine from, a memory that people and machines can both trust.

How I build it

Alone. Several AI assistants write code and prepare sources under my direction: collecting filings, extracting candidate sentences and drafting pages. I decide what gets published, and I compare each published row with the original filing myself. The date of that comparison is shown on the record.

Unfinished review stays unfinished on the page. A candidate without my comparison is not published. When a reader reports an error, the correction is added with its date and reason; the earlier version is not deleted.

Build timeline

DateWhat happenedEvidence
Website code started.Private repository, commit e1a788f (shown on request)
Moderna more than doubled in a day on one clinical readout; that day I decided to go all in on this.Forbes, 2026-08-26 ↗
Made ORION public.Founder log · 2026-09-09
Started sealing a daily ledger for a rules-fixed paper-trading account (the ledger itself is private).Founder log · 2026-09-22
Measured the Stale Clock: 377 of 3,762 active Phase 3 trials still listed a passed estimated completion date.Stale Clock report
Published the Stale Clock data and the steps to reproduce it.GitHub orion-stale-clock ↗
Started comparing 44 promise-ledger rows with the original filings by hand.Founder log · 2026-10-07
Created the public repository for the daily hash of the watch list.GitHub orion-daily-proof ↗
Published ORION's own promises in the same format as a company record.Promises
Published the earliest founder-reviewed company record: Viking Therapeutics (VKTX)./company/VKTX
Made the earliest public correction: a status shown on an ARWR row was wrong. The row was hidden.Founder log · 2026-10-08
Published founder-reviewed records: Biohaven (BHVN) and Scholar Rock (SRRK) and Septerna (SEPN) and Corbus Pharmaceuticals (CRBP)./company/BHVN · /company/SRRK · /company/SEPN · /company/CRBP
Published founder-reviewed records: Rapport Therapeutics (RAPP)./company/RAPP

13 rows in date order, as of . Rows for records, corrections and daily hashes are generated from the site's data files.

What went wrong, in public

  1. What
    The retained ICCC source has a separate Re-Tain study and data-analysis completion objective for Q1 2026.
    Why
    The earlier ICCC rows retained financial reporting, general studies and field feedback, but omitted this literal quarter-bound analysis objective.
    Fix
    Append the source-bound Study and data analysis completion row. Keep its elapsed quarter under Window ended with no completion/results judgment; preserve all existing rows, quotations, sources and human-review labels.
    Where
    Calendar · SEC document checked

  2. What
    C0105 links to a SEC document with issuer CIK 1178670; the item company CIK is 872589.
    Why
    The source issuer was not compared with the company CIK.
    Fix
    C0105 is hidden. Its original quote and source remain in the data; C0340 uses the company SEC filing.
    Where
    Calendar · SEC document checked

  3. What
    Calendar items C0007 and C0005 have the same company, program, event type, quote and SEC source.
    Why
    The same source-bound event was listed in several items.
    Fix
    The calendar retains C0005. C0007 remains in the data with public_display false and duplicate_of C0005.
    Where
    Calendar · SEC document checked

  4. What
    Calendar items C0012 and C0010 have the same company, program, event type, quote and SEC source.
    Why
    The same source-bound event was listed in several items.
    Fix
    The calendar retains C0010. C0012 remains in the data with public_display false and duplicate_of C0010.
    Where
    Calendar · SEC document checked

  5. What
    Calendar items C0040 and C0039 have the same company, program, event type, quote and SEC source.
    Why
    The same source-bound event was listed in several items.
    Fix
    The calendar retains C0039. C0040 remains in the data with public_display false and duplicate_of C0039.
    Where
    Calendar · SEC document checked

  6. What
    Calendar items C0042 and C0041 have the same company, program, event type, quote and SEC source.
    Why
    The same source-bound event was listed in several items.
    Fix
    The calendar retains C0041. C0042 remains in the data with public_display false and duplicate_of C0041.
    Where
    Calendar · SEC document checked

  7. What
    Calendar items C0044 and C0043 have the same company, program, event type, quote and SEC source.
    Why
    The same source-bound event was listed in several items.
    Fix
    The calendar retains C0043. C0044 remains in the data with public_display false and duplicate_of C0043.
    Where
    Calendar · SEC document checked

  8. What
    Calendar items C0137 and C0136 have the same company, program, event type, quote and SEC source.
    Why
    The same source-bound event was listed in several items.
    Fix
    The calendar retains C0136. C0137 remains in the data with public_display false and duplicate_of C0136.
    Where
    Calendar · SEC document checked

  9. What
    Calendar items C0174 and C0176 have the same company, program, event type, quote and SEC source.
    Why
    The same source-bound event was listed in several items.
    Fix
    The calendar retains C0176. C0174 remains in the data with public_display false and duplicate_of C0176.
    Where
    Calendar · SEC document checked

  10. What
    Calendar items C0175 and C0176 have the same company, program, event type, quote and SEC source.
    Why
    The same source-bound event was listed in several items.
    Fix
    The calendar retains C0176. C0175 remains in the data with public_display false and duplicate_of C0176.
    Where
    Calendar · SEC document checked

  11. What
    Calendar items C0182 and C0181 have the same company, program, event type, quote and SEC source.
    Why
    The same source-bound event was listed in several items.
    Fix
    The calendar retains C0181. C0182 remains in the data with public_display false and duplicate_of C0181.
    Where
    Calendar · SEC document checked

  12. What
    Calendar item C0139 was shown as Other although its SEC quote contains PDUFA.
    Why
    The item event type was used before the literal PDUFA wording.
    Fix
    The item is classified as PDUFA; its quote, source, date and review label are unchanged.
    Where
    Calendar · SEC document checked

  13. What
    Calendar item C0195 was shown as Other although its SEC quote contains PDUFA.
    Why
    The item event type was used before the literal PDUFA wording.
    Fix
    The item is classified as PDUFA; its quote, source, date and review label are unchanged.
    Where
    Calendar · SEC document checked

  14. What
    Calendar item C0224 was shown as Other although its SEC quote contains PDUFA.
    Why
    The item event type was used before the literal PDUFA wording.
    Fix
    The item is classified as PDUFA; its quote, source, date and review label are unchanged.
    Where
    Calendar · SEC document checked

  15. What
    The SEC acceptance time shown for avexitide (AMLX) on its program page was 2026-08-06T15:04:23Z, four hours later than the time in SEC's own header for the filing (10-Q, accession 0001193125-26-336682, filed 2026-08-06). It was shown from 2026-10-12, when the program file was published.
    Why
    The time had been copied from SEC's submissions API (acceptanceDateTime). For this filing that value adds the New York offset (4 hours, EDT) a second time. The filing's SEC header reads <ACCEPTANCE-DATETIME>20260806070423, which is 2026-08-06 07:04:23 New York time, 2026-08-06T11:04:23Z UTC.
    Fix
    Replaced with the header time, 2026-08-06T11:04:23Z. Checked on 2026-10-12 by reading the filing's SEC header; the other 14 program files matched. The review label is unchanged: the acceptance time is metadata, and the quoted wording and its source are unchanged. Earlier daily hashes and records are not changed; this correction is a new record.
    Where
    Amylyx Pharmaceuticals, Inc. (AMLX) · avexitide · SEC header of the filing · Evidence for all program files

  16. What
    The SEC acceptance time shown for Intismeran autogene (MRNA) on its program page was 2026-08-01T00:17:41Z, four hours later than the time in SEC's own header for the filing (10-Q, accession 0001682852-26-000150, filed 2026-07-31). It was shown from 2026-10-12, when the program file was published.
    Why
    The time had been copied from SEC's submissions API (acceptanceDateTime). For this filing that value adds the New York offset (4 hours, EDT) a second time. The filing's SEC header reads <ACCEPTANCE-DATETIME>20260731161741, which is 2026-07-31 16:17:41 New York time, 2026-07-31T20:17:41Z UTC.
    Fix
    Replaced with the header time, 2026-07-31T20:17:41Z. Checked on 2026-10-12 by reading the filing's SEC header; the other 14 program files matched. The review label is unchanged: the acceptance time is metadata, and the quoted wording and its source are unchanged. Earlier daily hashes and records are not changed; this correction is a new record.
    Where
    Moderna, Inc. (MRNA) · Intismeran autogene · SEC header of the filing · Evidence for all program files

  17. What
    The SEC acceptance time shown for mRNA-3927 (MRNA) on its program page was 2026-08-01T00:17:41Z, four hours later than the time in SEC's own header for the filing (10-Q, accession 0001682852-26-000150, filed 2026-07-31). It was shown from 2026-10-12, when the program file was published.
    Why
    The time had been copied from SEC's submissions API (acceptanceDateTime). For this filing that value adds the New York offset (4 hours, EDT) a second time. The filing's SEC header reads <ACCEPTANCE-DATETIME>20260731161741, which is 2026-07-31 16:17:41 New York time, 2026-07-31T20:17:41Z UTC.
    Fix
    Replaced with the header time, 2026-07-31T20:17:41Z. Checked on 2026-10-12 by reading the filing's SEC header; the other 14 program files matched. The review label is unchanged: the acceptance time is metadata, and the quoted wording and its source are unchanged. Earlier daily hashes and records are not changed; this correction is a new record.
    Where
    Moderna, Inc. (MRNA) · mRNA-3927 · SEC header of the filing · Evidence for all program files

  18. What
    The SEC acceptance time shown for ALN-6400 (ALNY) on its program page was 2026-07-30T16:11:02Z, four hours later than the time in SEC's own header for the filing (10-Q, accession 0001178670-26-000062, filed 2026-07-30). It was shown from 2026-10-12, when the program file was published.
    Why
    The time had been copied from SEC's submissions API (acceptanceDateTime). For this filing that value adds the New York offset (4 hours, EDT) a second time. The filing's SEC header reads <ACCEPTANCE-DATETIME>20260730081102, which is 2026-07-30 08:11:02 New York time, 2026-07-30T12:11:02Z UTC.
    Fix
    Replaced with the header time, 2026-07-30T12:11:02Z. Checked on 2026-10-12 by reading the filing's SEC header; the other 14 program files matched. The review label is unchanged: the acceptance time is metadata, and the quoted wording and its source are unchanged. Earlier daily hashes and records are not changed; this correction is a new record.
    Where
    Alnylam Pharmaceuticals, Inc. (ALNY) · ALN-6400 · SEC header of the filing · Evidence for all program files

  19. What
    The SEC acceptance time shown for selinexor (KPTI) on its program page was 2026-08-13T15:40:35Z, four hours later than the time in SEC's own header for the filing (10-Q, accession 0001193125-26-347906, filed 2026-08-13). It was shown from 2026-10-12, when the program file was published.
    Why
    The time had been copied from SEC's submissions API (acceptanceDateTime). For this filing that value adds the New York offset (4 hours, EDT) a second time. The filing's SEC header reads <ACCEPTANCE-DATETIME>20260813074035, which is 2026-08-13 07:40:35 New York time, 2026-08-13T11:40:35Z UTC.
    Fix
    Replaced with the header time, 2026-08-13T11:40:35Z. Checked on 2026-10-12 by reading the filing's SEC header; the other 14 program files matched. The review label is unchanged: the acceptance time is metadata, and the quoted wording and its source are unchanged. Earlier daily hashes and records are not changed; this correction is a new record.
    Where
    Karyopharm Therapeutics Inc. (KPTI) · selinexor · SEC header of the filing · Evidence for all program files

  20. What
    The SEC acceptance time shown for Cobenfy (BMY) on its program page was 2026-07-30T15:05:39Z, four hours later than the time in SEC's own header for the filing (8-K, accession 0000014272-26-000018, filed 2026-07-30). It was shown from 2026-10-12, when the program file was published.
    Why
    The time had been copied from SEC's submissions API (acceptanceDateTime). For this filing that value adds the New York offset (4 hours, EDT) a second time. The filing's SEC header reads <ACCEPTANCE-DATETIME>20260730070539, which is 2026-07-30 07:05:39 New York time, 2026-07-30T11:05:39Z UTC.
    Fix
    Replaced with the header time, 2026-07-30T11:05:39Z. Checked on 2026-10-12 by reading the filing's SEC header; the other 14 program files matched. The review label is unchanged: the acceptance time is metadata, and the quoted wording and its source are unchanged. Earlier daily hashes and records are not changed; this correction is a new record.
    Where
    Bristol-Myers Squibb Company (BMY) · Cobenfy · SEC header of the filing · Evidence for all program files

  21. What
    Added source-context excerpts from 2026-09-10 SEC Form 8-K.
    Why
    Founder-requested context retained from the checked original filing.
    Fix
    Shown in the record's Corrections section with its date; the earlier version is kept.
    Where
    Biohaven (BHVN)

  22. What
    Added previously omitted 2026-09-10 SEC Form 8-K timing source.
    Why
    Selected-source supplement following founder review; prior rows retained.
    Fix
    Shown in the record's Corrections section with its date; the earlier version is kept.
    Where
    Biohaven (BHVN)

  23. What
    Added previously omitted 2026-08-06 SEC Form 10-Q timing source.
    Why
    Selected-source supplement following founder review; prior rows retained.
    Fix
    Shown in the record's Corrections section with its date; the earlier version is kept.
    Where
    Scholar Rock (SRRK)

  24. What
    Corrected the April 13 event label to Statement retained; removed that equivalent expression from Wording changes.
    Why
    The comparison treats late 2026 / early 2027 and late 2026 or early 2027 as equivalent timing alternatives. Original quotations are retained.
    Fix
    Shown in the record's Corrections section with its date; the earlier version is kept.
    Where
    Septerna (SEPN)

  25. What
    Added the August 10 10-Q sentence about extending MAD cohort dosing to 14 days as a Context disclosed row.
    Why
    Founder requested the exact company quotation and original filing link.
    Fix
    Shown in the record's Corrections section with its date; the earlier version is kept.
    Where
    Septerna (SEPN)

  26. What
    Updated the Changes explanation to: Status is set by founder review of the original filings.
    Why
    The status is the founder-selected status retained in the reviewed record.
    Fix
    Shown in the record's Corrections section with its date; the earlier version is kept.
    Where
    Septerna (SEPN)

  27. What
    ORION-01 (Daily record · Start publishing a daily hash of the full watch list): Wording clarified; the commitment and its date are unchanged. Earlier wording: “From 2026-10-11, ORION publishes a hash of its full watch list every day so that no record can be quietly rewritten later.”
    Why
    The earlier sentence claimed more than a daily hash of the watch list can show. A hash shows that the list changed; it does not by itself stop a change.
    Fix
    The promise now says any later change to the watch list can be detected. The commitment and its date are unchanged, and the earlier wording stays visible on the promise.
    Where
    Promises · ORION-01

  28. What
    Display rule corrected — equivalent timing phrases (e.g., 'H2 2026' / 'second half of 2026') are now shown as Statement retained. Source quotes and status unchanged.
    Why
    Equivalent timing notation is compared after normalization.
    Fix
    Shown in the record's Corrections section with its date; the earlier version is kept.
    Where
    Biohaven (BHVN) · Scholar Rock (SRRK)

  29. What
    A status shown on an ARWR row was wrong.
    Why
    The earlier status label was withdrawn in review; the founder log records no further detail.
    Fix
    The row was hidden. It stays unpublished until it is compared by hand.
    Where
    Founder log · 2026-10-08

  30. What
    14 quarterly reports (Form 10-Q) for the companies under review had not been read.
    Why
    They were outside the reading scope selected at the time.
    Fix
    The focus was narrowed to readout announcements, and every record now states which filings were read and through what date.
    Where
    Founder log · 2026-10-08

Corrections on records and on ORION's own promises are read from the record and promise data; mistakes the founder stated are read from the founder log. Earlier versions are kept.

What exists today (updated )

Counts use the dates beside their retained data scopes. Page-edit dates are separate. A zero is shown as zero.

What ORION does

  • Keeps, for one program and one action, the earliest statement we found about timing, every later restatement, and what was actually disclosed, each with the SEC filing it came from and its filing date.
  • Shows the exact company wording and the time window as written. Changes are shown beside the earlier wording; earlier records are kept when a correction is made.
  • States for each record which filings and company releases were read and on what date, and labels the status with one of seven defined words. Read the status definitions.
  • Publishes a record only after I have compared every displayed row with the original filing. Each page states how many rows are shown out of how many rows exist.

What ORION does not do

  • No scores, rankings, forecasts, recommendations, price targets or expected returns.
  • No judgement words. A changed window is recorded as a changed window; a missing later statement is recorded as not found in the filings read, not as a stopped program.
  • No inferred dates. A registry completion estimate is not a promised announcement date, and a filing date is not an event date. Missing values are shown as unavailable.
  • No AI comparison counted as a human check. An assistant may prepare a candidate; a row is published only after my own comparison with the original.

Contact

It's small today. If something is wrong, write to hello@orionbiotechai.com and I'll correct it in public. Questions about a company and a trial go to the same address.

Read the founder log or read ORION's own promises.

Records, Archive & Scope

6 founder-reviewed records as of 2026-10-12 · daily public hash starts 2026-10-11 · last correction 2026-10-15 · as of

For biotech newsletter writers, analysts, and investors who track clinical readout timing.

All records (6)

as of

Each published row carries one of two labels: Checked by founder — compared by hand with the SEC original by ORION's founder, Jiwoo Seong (Woo); or Quote matched to filing (not yet reviewed by founder) — the quote matches the original filing character for character, and the founder has not reviewed the row. The two are shown separately. How rows are checked.

Earlier materials — the Alnylam (ALNY) past case, the company finder and worked notes — are kept on the Archive page.

Scope

ORION shows selected registry records and dated company statements with original sources. Current collection and verification status →

Registry completion estimates and company results-announcement dates are separate facts. Coverage and method →

A source trail you can follow

Researching a new biotech company? Start with what it said before.

  1. Find the company. Search a company or drug. Filter the calendar by stage, public float and company-stated period.
  2. Read the original. Read the literal company sentence and open its SEC filing. The review label shows how the row was checked.
  3. Follow the history. Read earlier wording, later statements and any disclosure held in the record, with their source and check dates.

Explore the calendar

Company, stage, stated period.

Compare programs, then open the original sentence. These are examples from the calendar as of .

Browse the calendar
  1. RAPPRapport Therapeutics, Inc.

    RAP-219

    Stage: Phase 2SEC filing · “Phase 2”

    Public float (as of ): $275M (SEC filing) · Under $500M

    Company-stated periodOctober 2026

    Original SEC wording
    topline results are expected in October 2026.

    SEC original ↗ · Filed · Quote matched to filing (not yet reviewed by founder)

    Last checked

  2. STTKShattuck Labs, Inc.

    SL-846

    Stage: OtherSEC filing · “Preclinical”

    Public float (as of ): $22M (SEC filing) · Under $500M

    Company-stated periodOctober 17-20, 2026

    Original SEC wording
    Preclinical data for SL-846, including safety, pharmacokinetics, and receptor occupancy from the ongoing GLP toxicology study are expected to be presented in a poster at UEG Week, to be held October 17-20, 2026.

    SEC original ↗ · Filed · Quote matched to filing (not yet reviewed by founder)

    Last checked

  3. IONSIONIS PHARMACEUTICALS INC

    bepirovirsen

    Stage: PDUFASEC filing · “PDUFA target action date”

    Public float (as of ): $5.2B (SEC filing) · Over $2B

    Company-stated periodOctober 26, 2026

    Original SEC wording
    PDUFA target action date of October 26, 2026

    SEC original ↗ · Filed · Quote matched to filing (not yet reviewed by founder)

    Last checked

Public float is the value of shares held by non-affiliates on the filing’s reference date. Periods repeat company wording.

Read public corrections → Inspect the daily record hash → Read a published weekly note →