This InsightMeter guide defines institutional “new position” and “add” in ways that resist marketing language. Vendor screens and social posts love badges: NEW, ADDED, INCREASED. Those badges are only as honest as the definition behind them. Your reader job is to state a share-count rule, apply it across comparable report periods, watch for corporate actions that create fake news, and place any single-quarter label in multi-quarter context. Education first: a clean “new” label is still a lagged long-only residue, not a live trade blotter and not a recommendation.
Before you say a manager “started” or “added,” write your rule. Example: “New means prior-quarter share count was zero (or absent) for this manager–issuer–share-class key, and current-quarter share count is greater than zero on the amended table if an amendment exists.” Example: “Add means prior and current share counts are both positive and current shares > prior shares by more than a trivial threshold you chose in advance.”
If you skip the rule, you will inherit someone else’s. Inherited rules often mix market-value increases, percentage-of-portfolio optics, and share increases without saying which one drove the badge. Arguments about badges without definitions are cosplay debates.
Keep “trim” and “exit” as the mirrors of add and new: exit means current shares are zero after a positive prior; trim means both positive and current < prior. Symmetric language keeps your notebook honest when the narrative wants only bullish verbs.
In 13F research education, the least lying definition of new is appearance from absence on a share-count basis: the manager’s information table shows a positive share quantity for a security that had zero shares (or no row) in the prior comparable report for the same manager entity you are tracking.
Honest caveats belong in the same sentence. The position could have been opened on the first day after prior quarter-end or the last day of the current quarter—13F will not tell you which. The manager might have held the name in a non-13F sleeve, short book, or derivative that never appears. “New in the 13F long table” is the precise claim; “new to the firm’s entire economic exposure” is usually stronger than the filing supports.
Entity hygiene matters. Comparing Manager North LP’s table to Manager North Advisors Ltd without documenting a mapping can create false news and false exits. Write which CIK and which filing entity you used.
Add should mean a share-count increase between two comparable snapshots: same manager entity, same issuer security key, positive shares in both periods, current shares higher than prior. Prefer absolute share deltas first; percentage increases can exaggerate tiny bases.
Decide in advance whether a 1% share increase counts as an “add” worth noting. Without a materiality threshold, every rounding wiggle becomes content. Thresholds are research design choices—state them—not truths of nature.
Dollar-value increases with flat shares are not adds under a share-count definition. Label them “value up on price” if you must mention them. Calling price-driven value increases “adding” is the most common definitional lie in retail 13F media.
Splits, reverse splits, spinoffs, ticker/CUSIP changes, mergers, and share-class consolidations can make tables look like news when the economic book did not “start” or “add” in the vernacular sense. If shares double on a 2-for-1 split and your naive code sees a 100% add, your definition lied to you.
Practical habit: when a name flips from absent to present or jumps sharply, ask “did the security identifier change?” before asking “what does management know?” Check issuer events around the quarter. Adjust share history to a common basis when you can; when you cannot, mark the label provisional.
Amendments (13F-HR/A) can turn a vendor’s “new” into “was already there after correction,” or the reverse. Always prefer the latest amendment for the report period before you publish a badge. Stale screenshots of originals are how false news hardens.
Late filings and irregular reporting calendars also distort quarter-to-quarter joins. If prior period is missing, absence is not proof of zero—it may be a data gap. Distinguish “prior showed zero” from “prior unavailable.”
A single new or add label is a two-point comparison. Managers churn. A new line next quarter may be gone. An add may be a temporary rebalance. Educational notebooks show at least three to four quarters when arguing that a name is a “core build,” or they explicitly say the sample is only one step.
Position size relative to the manager’s reported 13F book can matter for narrative humility. A brand-new 0.05% sleeve is not the same research object as a top-ten line. Size does not make a label true or false; it changes how loudly you should talk.
Compare peers only with shared definitions. If Peer A’s “new” uses shares and Peer B’s feed uses value-up flags, your peer table is a mess. Harmonize rules before ranking managers by “aggressiveness.”
Remember structural lag: by the time you read the badge, weeks or months may have passed since quarter-end. Pair this guide with the 13F lag article so “new” never silently means “new this morning.”
Fiction A — True new under share rule. Prior quarter: no WidgetCo row. Current: 400,000 shares. No split or CUSIP change found. Amendment none. Label: new (13F long table). Notebook also says: timing inside quarter unknown; not a live blotter.
Fiction B — Fake add from price. Prior: 1,000,000 shares value $50M. Current: 1,000,000 shares value $65M. Label under share rule: flat, not add. Marketing headline “boosted stake” is rejected.
Fiction C — Fake new from identifier change. Prior: OldCo shares 800,000. Issuer merges into NewCo with new CUSIP; current shows NewCo 800,000 and OldCo absent. Naive join says exit OldCo + new NewCo. Adjusted view: continuous economic line pending ratio terms. Label: corporate-action continuity, not vernacular “new.”
Fiction D — Multi-quarter trim after one add. Q1 200k, Q2 350k (add), Q3 100k (trim), Q4 0 (exit). A thread that only screenshots Q2 “ADD” misrepresents the biography. Your note shows the path.
Use this checklist before you repeat a NEW/ADD badge in writing.
Checklists are educational process tools. They do not create profitable trades.
Most “new vs add” errors are definition laundering.
Fixing them makes 13F conversation slower and much clearer.
Even perfect new/add labels do not reveal shorts, many derivatives, intra-quarter round trips, or intent. They classify public table differences. Classification is not alpha.
Thresholds, entity maps, and corporate-action adjustments require judgment. Two careful researchers can disagree at the margin. Document disagreements; do not pretend the badge is objective metaphysics.
Nothing in this guide is a recommendation to buy or sell any security, or to copy any manager. Educational definitions reduce lying labels; they do not guarantee profitable outcomes.
Keep a one-page card: your new/add/trim/exit rules, materiality threshold, how you handle missing priors, and how you mark corporate-action continuity. Attach the card to any table you share so readers inherit the rules.
When a vendor badge disagrees with your card, your card wins for your research—or you explicitly adopt the vendor rule and quote it. Silent disagreement is how teams talk past each other.
Revisit the card after you get burned by a split or amendment. Burns are curriculum.
Give juniors Fiction B and Fiction C above and ask for labels under a stated share rule. If they return “smart money piled in” without share math, the assignment fails.
Require as-of date, filing date, accession, and definition citation in every new/add note. Adjectives without those fields are incomplete.
Pair with share-count vs dollar-value, 13F lag, amendments, and comparing-managers guides so badges connect to the rest of 13F literacy.
Prefer: “Under a share-count rule, North’s WidgetCo line is new as of [as-of]; prior zero; current N shares; accession …; lag reminder …” Avoid: “North just discovered WidgetCo.”
When corporate action confounds the join, say so in the first sentence. Burying the confound is how false news travels.
If you only have value data, say “value change only—shares not verified” rather than borrowing an “add” verb.
Why share-count changes beat dollar headlines · 13F lag: quarter-end ≠ trade timing · 13F amendments and late filings · Comparing two 13F managers without survivorship bias
Define new and add with share counts, check corporate actions and amendments, and read badges in multi-quarter context with lag honesty. Definitions that don’t lie beat marketing verbs. Continue with related guides, methodology, and glossary. Nothing here recommends buying or selling any security.