How fast do 351 and 368 funds shed their starting holdings?

By group

Measured from

Earliest available book (mixed): the launch book where one exists, else the first SEC report. Fund pages and the release file use it.

Every view

By launch quarter, median Seed removed

Funds without a number

How this is measured

1. Ordinary launches: launches with no 351 or 368 transaction found in their SEC filings. The label comes from the absence of evidence; it does not prove a cash launch.
2. Seed removed: the share of the starting portfolio, by value, no longer held a year later.
3. First SEC report: the start is the fund's first N-PORT filing on or after its launch, or its 351 or 368 event. End: the N-PORT closest to 365 days later, within 330 to 400 days. Every fund starts from the same kind of public filing.
4. Launch day: the start is the fund's holdings on launch day, or at its 351 or 368 event, from the ETF Action / FactSet launch book; the end is the first SEC filing on or after day 365. Only funds with a launch book count, so every fund in this view starts the same way.
5. Earliest available book (mixed, Details only): the launch book where there is one, else an SEC filing within 120 days; same end as Launch day.
6. Launches: 2023Q1 to 2025Q2 by default, the funds old enough for a full year of filings. "All" adds 2025Q3 launches.
7. Coverage rule (First SEC report and Launch day): a fund's year counts only if the starting holdings matched to a security inside the retention basket make up at least 50% of the starting portfolio's noncash value. Otherwise it is listed under funds without a number. The mixed series keeps the earlier rule for unexplained zeros.
8. 351 and 368: a filing cites that section. Every other launch: ordinary launches.
9. Seeded with ETF shares: more than 25% of the starting portfolio's value, in the earliest available book, was fund shares. Seeded with stocks: 25% or less. A fund without that record counts only under all launches.
10. Gaps compare medians. The 95% interval resamples launch events 5,000 times: a launch event is one sponsor's funds launched on the same day, drawn together because they tend to move together. A gap needs at least 3 funds in each group.
11. Weighted by seed size: each fund counts in proportion to its seed dollars, its net assets in its first SEC filing at the earliest available book (N-PORT, or the seed value in its N-CSR). A fund without that value carries no weight. The bar is the level with half the seed dollars above it and half below. Share of all seed dollars removed: Seed removed averaged over seed dollars.
12. Trimmed 5%: the lowest and highest 5% of each group dropped before the statistic. It changes only the weighted statistic: an unweighted median does not move when the same number of funds is dropped from each end.
13. One sponsor's funds as one: funds whose names lead with the same brand count as one, at their median; their seed dollars add up.
14. With any of these on, each draw resamples launch events, then applies the sponsor grouping, trim and weights.
15. Benchmark: the 90th percentile of ordinary launches over the same window, compared like for like by seeding type a year from the first SEC report (pooled at day 90). Missing funds: the plan's lower bound places blocked 351 or 368 funds at their group's 25th percentile and blocked ordinary launches at their 75th, liquidated funds at 100%; the worst case scores every missing 351 or 368 fund at 0% and every missing ordinary launch at 100%.
16. Gain delivered in redemption: built-in gain that left the fund in redemptions in its first year, not recognized by the fund, in form, per $100 of average N-PORT net assets, annualized over 270 to 400 day windows. Diagnostic: the plan asks for at least 30 351 funds before reading it as a result.
17. Source: SEC N-PORT filings.