Corrections

The study left out NVDB and TSLI as trading as a different fund, on the price feed's cached name for each ticker, which was out of date: each is the fund filed. It printed a slope for AMDD, MARU, PLTD and TIPD that belonged to the fund now trading under each ticker, because an out-of-date SEC record still gave each ticker to the fund filed. As published: 347 funds measured, 287 behind their own promise and 60 ahead, a median gap of -2.78 percentage points (long funds -3.45, inverse +1.75), and 231 behind the naive reading of their multiple, by a median of -2.34.

NVDB and TSLI are measured over the same window. The rows for AMDD, MARU, PLTD and TIPD now say what each ticker trades as, with no slope. Now: 349 funds measured, 289 behind their own promise and 60 ahead, a median gap of -2.80 percentage points (long funds -3.45, inverse +1.75), and 233 behind the naive reading of their multiple, by a median of -2.41. What a ticker trades as is now one test everywhere ETFIQ asks it: the exchange's own listing, held against the fund's own prices.

The study left out AMDD, NVDB, PLTD and TSLI as trading as a different fund, on the price feed's cached name for each ticker, which was out of date: each is the fund filed. It printed a slope for MARU, ONG and TIPD that belonged to the fund now trading under each ticker, because an out-of-date SEC record still gave each ticker to the fund filed. As published: 595 funds measured, 490 behind their own promise and 104 ahead, a median gap of -2.27 percentage points (long funds -2.65, inverse +1.58), and 460 behind the naive reading of their multiple, by a median of -4.57.

AMDD, NVDB, PLTD and TSLI are measured over the same window. The rows for MARU, ONG and TIPD now say what each ticker trades as, with no slope. Now: 599 funds measured, 492 behind their own promise and 106 ahead, a median gap of -2.25 percentage points (long funds -2.65, inverse +1.55), and 463 behind the naive reading of their multiple, by a median of -4.57. What a ticker trades as is now one test everywhere ETFIQ asks it: the exchange's own listing, held against the fund's own prices.

The study left out NVDB and TSLI as trading as a different fund, on the price feed's cached name for each ticker, which was out of date: each is the fund filed. It printed a slope for AMDD, MARU, PLTD and TIPD that belonged to the fund now trading under each ticker, because an out-of-date SEC record still gave each ticker to the fund filed. As published: 404 funds measured, 332 behind their own promise and 72 ahead, a median gap of -0.82 percentage points (long funds -1.00, inverse +0.52), and 300 behind the naive reading of their multiple, by a median of -1.15.

NVDB and TSLI are measured over the same window. The rows for AMDD, MARU, PLTD and TIPD now say what each ticker trades as, with no slope. Now: 406 funds measured, 334 behind their own promise and 72 ahead, a median gap of -0.82 percentage points (long funds -0.99, inverse +0.52), and 302 behind the naive reading of their multiple, by a median of -1.15. What a ticker trades as is now one test everywhere ETFIQ asks it: the exchange's own listing, held against the fund's own prices.

Each answered "Does this fund stake?" with "XRP is a proof of stake chain, so staking is possible". XRP is neither proof of work nor proof of stake: its ledger settles by a consensus its validators are not paid for, so there is nothing to stake and no fund holding XRP can pay a staking yield. The claim was in the visible answer and in the FAQPage structured data on all ten pages.

Both answers now read the chain from the trust's own filings through one function, which is where the rest of the site already read it. The wrong answer came from a second copy of the rule written as a list of three coin names, so anything not on that list was called proof of stake.

On 966 fund pages the category score was declared to machine readers as running from 0 to 100. No fund can reach either end: a percentile of a rank tops out at 100 times (n minus a half) over n, and ties and blending press it lower still, so the thematic set actually runs 0.2 to 92.4 and income's single-company set 12.2 to 76.7. The claim had already been taken off every visible surface and was left in the copy a search engine reads.

The ends now come from the set's observed lowest and highest, through the same call the drawing beside them uses, so the picture and the structured data cannot state different ends for one set.

138 funds showed “not published” for their expense ratio. Every US ETF states one in its prospectus and most issuers print it on the fund's own page: Innovator gives 0.79% for JANH on a page this site already links to. The field was absent from ETFIQ's reading, not from the world, and the page said the opposite.

The cell now reads “not read by ETFIQ”, which is what is true, and the fund's own page on the issuer's site is in the page's links block for anyone who wants the figure today.

Two rankings numbered their rows 1 to 15 where all fifteen funds were level on the figure they were ranked by: every fund on “Buffer ETFs with the most protection left” at 100.0%, and every fund on “The thematic ETFs least like the S&P 500” at 100.0% active share. The numbering stated an order the data did not have.

Both pages now say how many are level at the top and that the order among them is not a ranking. The figures and the sort are unchanged.

The buffer depth was blank on all 245 buffer fund pages. The page read a field the buffer data had never carried, so the row rendered empty rather than showing, for example, 15%.

The depth is now derived from the published start and end of the buffer, which are the figures the band itself is drawn from.

On 304 of 337 income charts the benchmark label was drawn outside the chart and did not appear, so the bar comparing the fund with its benchmark was unlabelled.

The label is placed inside the chart when there is no room beside the bar.

A positive one-year return was drawn in the color this site uses for a fund behind its benchmark, so a gain could read as a loss.

Return figures are drawn in ink. Color on this site carries one meaning per kind of fund and is not used to decorate a number.

Net assets were carried from an SEC filing that could be months old, without saying so. AIS was published at about $697mm while the issuer's own page showed roughly $972m.

Net assets now come from the freshest source that carries its own date, and every page names which source and which date.