Lido & stETH Weekly Data Report — Week of September 21–27, 2026

1. Executive summary

The sixth snapshot is the second in a row with a fully fresh sector board, and the first whose headline is a fall: the network staking APR dropped 39bp between weekly prints. Six findings, none of them requiring an invented figure:

• Lido TVL rose 2.9% to $26.47B (DERIVED: 25.72 → 26.47, both direct reads from the same series). Mostly the price leg again: ETH rose 5.1% while the dollar TVL rose less.
• The price feed reprinted our Sep 20 baseline lower — ETH −1.65%, stETH −1.83%, LDO −4.54% versus what edition five printed for the same date. Every cross-week number in this snapshot uses the re-read; the details and the rules are ledger item 1 (section 6).
• Staking reached 36.19% on the same circulating-supply denominator family as the last three weekly prints — +0.42pp against the prior print of 35.77%, with the Sep 21 print (35.91%) in between.
• The network staking APR fell to 2.18% — a 39bp single-week step after three stable prints at 2.57–2.62%, the largest move in our records. The stETH 7-day average fell 8bp to 2.52%. This snapshot prints both; the deep dive examines what the drop can and cannot be.
• The ETF bid returned: +$689.8M across five positive sessions (DERIVED from the Farside daily table), an $830.4M swing from last week's −$140.6M.
• The supply cross-check gap narrowed to 1.2% (from 1.6%) — but only because the price rose against a supply print that is now 20 days old. Section 3 shows why that is weaker news than it looks.

2. Week-over-week table

Source: all figures as cited in sections 3–7 and the sources list. “As published” priors are the numbers edition five printed on Sep 20, kept for lineage; where the re-read differs, both appear.

3. The cross-check: a smaller gap on an older input

The standing check multiplies stETH supply by the same-source stETH price and compares the product with Lido's TVL:

The gap narrowed from 1.6% to 1.2%, and the improvement is mostly an illusion of the price leg: a higher stETH price lifts both sides of the comparison, but it lifts the TVL side more because the TVL rides the full price while the dated supply print cannot grow at all. The flow side of the ledger says the same thing from underneath: net issuance ran +19,815.93 ETH over the week (Ebunker, Sep 28), about 2,830 stETH a day (DERIVED). Eighteen days of issuance at that rate adds roughly 51,000 stETH — under half of the 116,520 gap. The gap is still, first and foremost, a supply print that has not moved in twenty days while the market around it repriced 9%. Aggregators continue to disagree: BingX prints 9,657,537 (Sep 7, on our basis); LBank prints 9.43M (Sep 18, which would widen the gap).

Source: supply print per CryptoCompare via TickerReport (Sep 9); TVL and price as tabled; net issuance per the Ebunker weekly (Sep 28); aggregator prints.

4. The queues: two series, two sources, both dated

This snapshot changes how the queue rows are sourced, and the change is disclosed rather than smoothed. The ETH-count series (validatorqueue.com) could not be re-read this week; the wait-time series (Ebunker's weekly) could, and it is the one that refreshed. The two measures are not the same thing — on overlapping weeks they have differed by more than twelve days — so we print each with its own dates and compute no ratio across them:

Carried from the last obtainable validatorqueue read (Sep 20): entry 2,480,000 ETH at a 43.1-day wait; exit 182,352 ETH at 3.2 days. Against Ethereum's churn limit of roughly 57,600 ETH a day, the entry backlog is on the order of six weeks of full-capacity throughput — consistent in direction with the Ebunker wait series, though the levels differ by the twelve days described above. Two shape facts survive the source change. The entry wait has now fallen in four consecutive weekly prints (34.6 → 31.0 → 30.3 → 27.7 days). And the exit wait's jump from under an hour to over two days happened at the Sep 21 print — the week before this window — and has since held near 2.8 days. CMv2 consolidations bypass both queues entirely, so the migration itself neither drains nor fills them; the deposit reserve decision (Sep 25) is the piece of the week's governance that touches this plumbing directly.

Source: Ebunker Ethereum staking weeklies (the last four editions); validatorqueue.com via Tremplin (Sep 20); churn limit per Ethereum protocol parameters as previously cited.

5. The yield corridor against a 2.18% network print

The gap arithmetic, on one price and one set of rates. Position: 10 ETH = $27,092.58 at the Sep 27 print of $2,709.26. Every line below is DERIVED:

Cash out-pays the carried user-net corridor by $461 a year against bills and $575 against the two-year — 168–171bp and 210–213bp across the 2.21–2.24% band. On the fresher network APR the shortfalls widen: 174bp against the carried bill (DERIVED: 3.92 − 2.18), 172bp against the rate the Fed pays on reserves, 182bp against the policy rate and 192bp against the SEP's 4.1% end-2026 median. Every one of those readings is a record for this column, and every one was set by a single weekly print rather than a Fed decision.

Source: rates as tabled above; position value from the DefiLlama coingecko feed, Sep 27 print; all conversions are our own arithmetic on those inputs.

6. Data-quality ledger

Four items this week, all of them ours to explain.

1. The price feed reprinted our Sep 20 baseline — lower. Re-reading the same daily series on Sep 29, the Sep 20 points print ETH $2,577.37 against the $2,620.59 edition five published (−1.65%), stETH $2,573.90 against $2,621.87 (−1.83%), and LDO $0.39973 against $0.41873 (−4.54%); BTC re-printed within 0.02%. The same re-read left every TVL point untouched — the Sep 20 Lido TVL re-printed within 0.01% of $25.72B — so the revision is confined to the price series. Our standing rule does the work: every cross-week comparison in this snapshot is computed from one same-session read (the Sep 20 and Sep 27 prints taken together on Sep 29), never by mixing reads from different days. One visible consequence: the Sep 20 peg, which edition five printed as a 5bp premium (1.00049), prints a 13.5bp discount (0.99865) on the re-read basis — and this week's discount stretch is measured against that. Edition five's pages stand as published; this ledger is the bridge between the two bases.

2. The queue wait series changed source. validatorqueue.com, the source of the ETH-count queue figures we have printed since edition one, was not reachable this week through any endpoint we hold. The Ebunker weekly's wait-time series refreshed instead, and this snapshot prints the two side by side, dated, without blending them: on overlapping weeks the measures differ materially (43.1 days on validatorqueue's Sep 20 read against Ebunker's 30.3 days at the Sep 21 print). If validatorqueue becomes readable again, both series resume; until then the wait column is Ebunker's and the ETH counts stay carried at Sep 20.

3. The supply print is now 20 days old, and the gap is 1.2%. The cross-check's supply leg has not moved since September 9. Eighteen days of net issuance at this week's rate adds about 51,000 stETH against a 116,520 gap — issuance alone does not close it, and neither aggregator's alternative figure is a dated read we can defend. The gap stays open, printed, and smaller this week for the mechanical reason given in section 3.

4. The validator count is carried — and the hero card says so. Both aggregator endpoints that publish an active-validator count returned access errors this week, so the 906,000 figure is carried from the last obtainable read (Sep 20/21). The homepage card that displays it now labels it as carried rather than dressing the staleness up as a trend. The count matters more than usual right now: CMv2's one-third validator reduction is the migration's central projection, and the weekly count is the number that tests it.

Source: DefiLlama coingecko feed and protocol-history endpoints (same-session reads); Ebunker weeklies (Sep 21 and Sep 28); validatorqueue.com via Tremplin (Sep 20); supply print as cited in section 3.

7. Macro and flows

No new policy print. The FOMC's Sep 16 hike to 3.75–4.00% stands, with the implementation rates (IORB 3.90% effective Sep 17) unchanged through the window; the committee next meets in October. The carried 3.92% bill still predates the hike and is labeled accordingly wherever it enters an arithmetic.

The ETF bid returned in size. Five sessions, all positive: +$270.0M (Sep 21), +$162.2M (Sep 22), +$104.5M (Sep 23), +$66.1M (Sep 24), +$87.0M (Sep 25). Window net +$689.8M (DERIVED) against −$140.6M the week before — an $830.4M swing. The largest day, Sep 21, alone out-raised all five sessions of the prior window.

The sector board, second consecutive full refresh. Every row below is a DefiLlama protocol-history read from the same session, pointing at the Sep 27 print:

One honest observation about the board's shape: the chain's dollar TVL rose 1.8% in a week when ETH rose 5.1%, which means the board is slightly smaller in ETH terms than it was. Lido's 2.9% outgrew the chain it lives on; the four liquid-staking protocols on this board hold $43.11B between them, and Lido plus Binance Staked ETH account for 85% of that (DERIVED). Lido is 2.6× Binance Staked ETH and 1.4× Aave (both DERIVED).

Source: Farside Investors daily table; DefiLlama protocol-history reads (api.llama.fi/protocol/<slug> and /v2/historicalChainTvl/Ethereum), same session; FOMC decision and implementation note, Sep 16–17.

8. Conclusion

The sixth snapshot's story is a divergence. The market leg of the page moved the way a bull week moves: price up 5.1%, TVL up 2.9%, ETFs taking in $689.8M, the board higher in every row. The yield leg moved the other way: the network APR printed 2.18% after three weeks at 2.57–2.62%, the stETH 7-day average followed for 8bp, and every cash-versus-staking gap this column tracks set a new record — not because the Fed moved, but because the yield did. Between the two legs sit the structural facts, which barely moved at all: the staking ratio rose 0.42pp to 36.19%, the entry queue's wait shortened for a fourth straight print, and 8.4M ETH began the long re-keying into CMv2 that will define Lido's operating profile into next year. The honest summary of the week is that capital is buying Ethereum while Ethereum pays less for its capital — and the deep dive this week is about whether that 2.18% print is real.

9. Daily prints, this window

The window's daily path, all on the same re-read basis (the Sep 20 row is the re-read of the date edition five printed as a 5bp premium):

Two readings survive the daily granularity. The ratio's worst prints came early in the week (Sep 20 and Sep 22, both near −13.4bp) and narrowed into the close, so the week's discount is better described as steady than as widening. And ETH's own path peaked Sep 22 and round-tripped — the same session structure LDO showed one day offset — which is the background against which the re-read basis matters: every cross-week comparison in this snapshot is taken on the Sep 29 read, never by mixing days.

Source: DefiLlama coingecko feed, ETH and stETH daily prints at the same fixed UTC hour, same-session re-read.

FAQ

Related reading

• Lido & stETH Weekly News Review — Week of September 21–27, 2026
• The 39-Basis-Point Week: What Just Happened to Staking Yield
• Lido & stETH Weekly Data Report — Week of September 14–20, 2026