MULDEN DATASETS
============================================================
Assembled 2026-10-10 from https://themulden.com

WHAT THIS IS
Measurements computed from SEC Form N-PORT filings -- the funds' own
quarterly position-level reports -- for US-registered funds. Every figure
here is derived from a filing, not from a vendor feed and not estimated.

LICENCE
Creative Commons Attribution 4.0 (CC BY 4.0). Use it commercially, modify
it, redistribute it. The one condition is attribution: credit
themulden.com and say if you changed anything.

If attribution does not work for your use -- embedded in a product, in a
client deliverable, anywhere a credit line cannot go -- there is a
commercial licence: https://themulden.com/licence/

METHOD, IN SHORT
  Effective holdings = 1 / sum of squared portfolio weights. It says how
  many EQUAL holdings would give the same spread as the fund's actual
  weights, which is usually far fewer than its holding count.
  Overlap = sum over shared securities of the smaller of the two weights.
  Positions are matched on CUSIP, then ISIN, then name.
  Securities-lending cash collateral is excluded and weights renormalised
  (definition ex-collateral-v2), because it is not an investment the fund
  chose -- counting it made IWM's 2,017 stocks read as 33.
  Full field definitions and every dated definition change: dictionary.json

DATES -- read this before quoting a number
  Every fund files on its own schedule. A figure carries the filing period
  it describes (asOf) and the date the SEC received it (filedDate), and any
  comparison between two funds compares two filings that may be weeks
  apart. Both dates are in the data. Nothing here is real-time; it is as
  current as the last filing.

LIMITS
  Leveraged, inverse and heavily derivative funds are marked not comparable
  on the holdings measures; their notional would make the arithmetic
  meaningless. Funds that file no position-level N-PORT -- unit investment
  trusts such as SPY and DIA -- are absent entirely.
  Country is the issuer country as the filer reported it, usually the place
  of incorporation rather than where the business is.

VERIFICATION
  Before each release every dataset is recomputed and cross-checked: the
  overlap matrix against an independent implementation, the filed returns
  against a price-derived series, the lending figures re-read with a second
  parser. A disagreement stops the release. The log is public at
  https://themulden.com/verification/

IF A FIGURE IS WRONG
  Say so at https://themulden.com/contact/ and it will be corrected in
  place with a dated note. That has happened and is recorded.

NOT IN THIS FILE
  Raw N-PORT filings (public domain, from EDGAR) and the per-fund
  point-in-time position files, which are published separately at
  https://themulden.com/data/positions/

CONTENTS
  concentration.json  (1.6 MB)
      Current snapshot per fund: effective holdings, top-ten weight, HHI, country and asset splits, collateral.
  history.json  (6.8 MB)
      The same measures every quarter since 2019, per fund. 24,300 observations.
  overlap.json  (0.6 MB)
      The 275 hand-measured fund pairs with their shared positions.
  overlap_matrix.csv.gz  (1.5 MB)
      Every pair of every comparable fund over $1bn: 235,641 rows of overlap, shared count, combined effective holdings.
  overlap_matrix.json  (0.1 MB)
      Index to the matrix: each fund's closest match, median overlap, and the funds-of-funds flags.
  lending.json  (0.5 MB)
      Securities lending from Form N-PORT Item B.4.a: borrowers by name and LEI, amount on loan, cash collateral cover.
  returns_filed.json  (0.3 MB)
      Monthly total returns as the funds filed them (Item B.5), cross-checked against a price-derived series.
  country_world.json  (0.1 MB)
      Every comparable fund pooled by filed issuer country, per quarter since 2019.
  fund_changes.json  (1.3 MB)
      Filing-to-filing changes per fund: positions added and dropped, largest weight moves, turnover proxy.
  leverage.json  (1.7 MB)
      Derivative exposure: swap and futures notional as % of net assets, named counterparties, the financing leg.
  dictionary.json  (0.0 MB)
      What every field means, the shared conventions, and the dated log of definition changes.
