Point-in-time positions

One gzipped CSV per fund at /pit/TICKER.csv.gz, columns:

asOf, filedDate, cusip, isin, name, assetCat, issuerCat, country, fairValLevel, valueUSD, pctOfNetAssets, isCashCollateral, isLoanByFund

• pctOfNetAssets is the filing's pctVal: percent of net assets, as filed, collateral lines included and flagged.
• isCashCollateral = Y marks securities-lending cash collateral; the site's headline figures exclude these rows and renormalise (see /data/dictionary/).
• Rows with pctVal <= 0 are omitted. Names are truncated to 120 characters. asOf is the reporting period end.
• Files up to 1000 KB are hosted at /pit/{TICKER}.csv.gz; larger ones are listed here with hosted=false until a bulk host is in place.
• Index of every file with row counts and date ranges: /pit/_manifest.json. Field meanings: data dictionary.

import pandas as pd
df = pd.read_csv("https://themulden.com/pit/VOO.csv.gz")
own = df[df.isCashCollateral != "Y"]            # the fund's own investments
w = own.groupby("asOf").pctOfNetAssets.transform(lambda s: s / s.sum())
eff = (w**2).groupby(own.asOf).sum().rpow(-1)   # effective holdings per quarter