巴菲特／波克夏 Berkshire 持股查詢｜13F 跨季比較｜Filingo

Example: what one report looks like

Send this as your first message (copy it as is):
Compare Berkshire's last two 13F quarters and attach the report.

Real output from 2026-10-06 (Berkshire Hathaway Inc, CIK 0001067983, 2026-03-31 → 2026-06-30):

• Sources: SEC EDGAR 13F-HR 0001193125-26-226661 and 0001193125-26-352200 (filed 2026-05-15 and 2026-08-14), with links.
• Check: 30 rows reconciled row by row and against the filing's cover-page total.
• Summary: 28 comparable holdings (filed quantity up 7, down 6, unchanged 15); 1 newly listed, 1 no longer listed. Newly listed is not the same as bought; no longer listed is not the same as sold out.

The first three rows of the difference table in alphabetical order (format only, not a selection):

The full report lists all 30 rows with values, weights, same-period corrections and a source locator for every row, as .md, .html and .json.

範例：一份報告長什麼樣子

第一句直接複製貼上：
比較波克夏最近兩季的 13F，附上報告。

2026-10-06 實跑結果（Berkshire Hathaway Inc，2026-03-31 → 2026-06-30）：可比較 28 項（申報數量增加 7、減少 6、不變 15），本期新列示 1 項、未再列示 1 項（新列示不等於買入，未再列示不等於全部賣出）；30 列逐列核對、與申報總額核對通過。上表是差異表依證券名稱字母順序的前三列，只示範格式，不代表任何挑選。

What you get

• Pick a filer by alias (a hand-checked list of well-known managers; filers prints it) or any CIK among the roughly 8,900 institutions that file 13F. Hand-checked filers carry the label "人工核對": we reviewed their last eight quarters' amendment chains by hand. Any other filer still runs with the same rules and carries "自動核對" plus what the automatic checks found: how many of the last eight quarters resolve, which ones are paused, whether the two periods include amendments, and the cover-page total check.
• Choose two periods from the verified list inside the most recent eight report periods (periods shows which are sellable). Periods whose amendment chain cannot be verified are listed as paused and are never substituted.
• Each report: nine-field header (subject, scope, periods, filing dates, accession numbers, version chain, fetch time, completeness check, usage), summary counts, full difference table with comparability flags, same-period correction table, and a copyable next-call sentence that records the last target period you saw.
• Three files per report: .md, .html (self-contained), .json (fixed schema, the source of truth for the other two).

How the hour works

• One purchase = the number of hours you choose of cloud access; nothing is charged after the time ends, and you can buy again any time.
• Up to 10 reports per hour of access: the tool counts reports from the first report of each 60-minute window and stops at the limit, telling you when the count restarts. Requests that fail (source unavailable, period not verifiable) are not counted. Listing periods or filers is free.
• The assistant returns a short brief and attaches the files. Each purchase starts fresh: the count and the "compared with your last query" line apply within that purchase only.

Buying, cancelling, refunds

• Hourly access does not auto-renew. Refund requests follow Capafy's 7-day dispute window from successful payment; after that the transaction is final. Ask through the platform or email the support address on this page.

Data freshness and delays (please read before buying)

• Institutional managers file 13F within 45 days after each quarter end (SEC Rule 13f-1); filings can be late or amended. A report is only as fresh as the latest filing available at the time you call. The end of every report states the next statutory filing date; it is a legal deadline, not a promise that new data will be there.
• If a filing cannot be verified (amendment chain unresolved, missing base period), the tool says so and does not substitute another period.

What it does not do

• Filed data only: no investment advice, no price targets, no rankings, no inference of anyone's intent. Differences are reported as filed share counts and values.
• 13F is a quarter-end snapshot of US long positions; it is not a live portfolio and does not equal any individual person's holdings. A question like "who holds X" is answered only as "which of the covered filers listed X in which period".
• No scheduling or push: data is fetched only when you call.

How it is produced

• Reports are produced by deterministic code with no model in the loop. The assistant that relays them to you runs on Anthropic Claude through the platform's relay; your messages to it are processed by that model.

Sources

• SEC EDGAR 13F-HR / 13F-HR/A (public domain). Every row carries the accession number so every number can be checked at the source.

中文說明

• 輸入機構別名或 CIK、選兩個季度，得到繁體中文白話報告（Markdown／網頁／JSON 三版數字一致）。人工核對過的名人機構清單直接用代號；其他約 8,900 家申報機構給 CIK 也能查，報告會寫出自動核對的結果。
• 只寫申報顯示的股數與市值變化，不寫買賣、不寫目標價、不推測意圖；13F 是機構申報，不等於任何個人的持股。
• 每小時最多 10 份報告（從每 60 分鐘的第一份起算），工具會計數並在上限停下；失敗的查詢不計。報告以檔案附件交付。
• 按小時計費＝一次買妳選的時數，不自動續約；退款依平台 7 天爭議期規則。
• 13F 是季末後最晚 45 天申報的資料，可能延遲或修正；查不到可驗證的申報就明講，不拿別的期間代替。
• 報告由固定程式產生，過程中沒有模型參與判斷；把報告轉述給你的助理跑在平台轉送的 Anthropic Claude 上。

Disclaimer

• This service only organizes public holdings and filing data for information and research. It does not provide investment analysis opinion, personalized or other investment advice, recommendations to buy or sell securities, solicitations, price targets, or guaranteed returns. No affiliation, sponsorship, or endorsement by any named person, institution, fund, exchange, or data source is implied. Data may be delayed, incomplete, or amended, and does not represent anyone's complete current portfolio, transaction history, cost basis, or investment intent. Form 13F primarily describes publicly reported quarter-end positions and excludes short positions; changes in reported value or quantity alone do not establish purchases or sales. Check the original sources, dates, and limitations shown in each report; missing or incomparable data is identified and must not be read as "no change". Assess your own needs and consult an appropriately qualified professional as appropriate based on your individual circumstances. This report must not be relied upon as a basis for promoting, endorsing, or acting as a spokesperson for any investment decision. Any user who publicly expresses an investment opinion based on the contents of this report does so at their own responsibility, and this service shall not, by reason thereof, be deemed to have jointly made, endorsed, or promoted such opinion or recommendation. Any increase or decrease in reported holdings or market value constitutes solely a statement of objective fact and does not constitute, and should not be interpreted as, any prediction, recommendation, or implication regarding the future price movement of the relevant security.