stETH Market Intelligence — September 2026 Monthly Report

1. The month in one number

September was the month Ethereum's price and Ethereum's staking yield moved in opposite directions at the fastest pace this site has recorded. ETH closed the month at $2,674.18, up 10.7% from the Aug 31 close of $2,415.61 — a range of $2,385.90 (Sep 3) to $2,749.04 (Sep 23). In the same month, on the same weekly series this site has tracked since edition three, the network staking APR went 2.62% → 2.57% → 2.57% → 2.18% — ending with a 39bp single-week step, the largest in our records, examined in the deep dive of September 29. LDO gained +35.9%. Lido TVL grew +13.3%. A Fed hike, a 5.11% ten-year Treasury, a record ETF inflow week and a validator-exit security event all fit inside the same four weeks. This report consolidates the month from the five editions that covered it, on one stated basis — every monthly change is computed on a single read of the DefiLlama coingecko feed and DefiLlama protocol history, with the Aug 31 anchor drawn from the same read session as the Sep 30 close, so no figure mixes prints.

Source: all figures from this site's editions 3–6 and the single September read session of the DefiLlama coingecko feed and DefiLlama protocol history; basis and coverage notes in sections 9–10.

2. The September price board

All four rows come from one read session, so the monthly changes are computed on a single basis rather than mixed prints. LDO's month: a 16 Sep low, a +22.3% final week (its best in our records, on two double-digit single-session drops along the way), and a close that still sat 2.6% below the Aug 27 launch-week high it printed in between. ETH's month broke cleanly into thirds: the low on Sep 3, the recovery through the Fed hike, the Sep 23 high, and a choppy, roughly flat final week that closed with the Sep 30 MetaMask disclosure. The ratio of LDO's month to ETH's — roughly 3.4× — is the single number that captures where the beta lived: not in the asset, but in its governance token.

Source: DefiLlama coingecko feed, one read session, Aug 31 and Sep 30 points; extremes computed from the same read.

3. The four weekly arcs

Week of Aug 31–Sep 6 (edition 3). The founding-month's close: a record 35.55% staking ratio, the entry queue at 2.07M ETH waiting 36 days, the exit queue exactly empty, and staking paying 2.61% network APR against a 3.92% T-bill. ETH +4.9% on Sep 4 was the week's tape.

Week of Sep 7–13 (edition 4). The staking-ratio print stepped to 35.77%; Bitmine's staked position printed 5.07M ETH (Sep 8); Stakely's two Lido V3 vaults arrived (Sep 9); and the week carried the worst data error of the site's short life — a queue story from September 2025 misread into the current record, caught and corrected with banners across five archived pages on Sep 29.

Week of Sep 14–20 (edition 5). The Fed hiked 25bp on September 16 (12–0 vote, to 3.75–4.00%), ETH fell −4.5% that session and ETFs printed −$224.1M the same day — the month's worst outflow. The CLARITY Act's cloture vote failed (Sep 15); the SEC's tokenized-securities exemption landed (Sep 17); Lido Core's first CMv2 consolidation executed (Sep 17); Simple DVT retired (Sep 18). The exit queue stayed near-empty at 182,352 ETH; the entry queue stood at 2.48M.

Week of Sep 21–27 (edition 6). The ETF bid returned for five straight days worth +$689.8M; LDO printed its best week in our records at +22.3%; the network APR stepped −39bp to 2.18% — the month's defining yield print — while the staking ratio reached 36.19% on the Ebunker denominator; governance moved twice (Vote #205 on-chain Sep 21; the committee's reserve dial Sep 25); and Bitmine's treasury reached 5.98M ETH (Sep 21).

Source: editions 3–6 as published; each figure carries its own source in the edition of origin.

4. ETF flows: a $831M month with a violent middle

September's US spot ETH ETF net flow: +$831.3M on the Farside aggregation (SoSoValue prints $832.4M on its slightly different universe — both stated), across 13 positive and 8 negative sessions. The shape mattered more than the total: the month's largest inflow (+$270.0M, Sep 21) and largest outflow (−$224.1M, Sep 16) sat five sessions apart on either side of the Fed hike, and the final three sessions of the month (Sep 28–30) ran −$45.3M — a closing outflow streak that left September with momentum to the downside. August's comparison: +$1.85B. September was a deceleration, not a reversal — cumulative net inflows ended the month near $13.9 billion, and fund net assets printed $16.72B on the Sep 18 dated read. Reading the month as a single +$831M number hides its shape: a market that bought ETH aggressively into the Fed hike, then sold it aggressively out of the same week, and closed the month two sessions into an outflow streak.

Source: Farside Investors daily table (monthly aggregation per SuperEx monthly report, Oct 2026; SoSoValue figure per Particle News and Guavy, Oct 2026); Aug comparison per Guavy (Oct 2026).

5. TVL: the month the board re-rated

Lido TVL opened the month at $23.29B (Aug 31 point) and closed at $26.38B (Sep 30 point) — +13.3% on one read basis. Ethereum chain TVL: $48.77B → $53.29B, +9.3%. The weekly board's movement was broad: Aave finished September at $19.24B, Binance Staked ETH at $10.04B, Morpho at $11.04B — every major row fresh by month-end after editions one through four had run on carried figures. Lido ended the month holding roughly 22.6% of staked ETH (DERIVED on the Sep 30 TVL read against the 43,681,037 ETH Sep 28 board — the month-end board is the consistent denominator) and roughly a fifth of the chain's DeFi TVL. The TVL board's re-rating is the quiet headline of the month: a +13.3% Lido print driven as much by ETH's +10.7% as by net deposit inflows, which is exactly what a TVL denominated in dollars should do when its underlying asset re-rates.

Source: api.llama.fi/protocol/lido and /v2/historicalChainTvl/Ethereum, one read session; weekly board rows per editions 5–6.

6. Staking and yield: the 39bp month

The staking ratio's September path on the Ebunker series: 35.55% (Sep 7) → 35.77% (Sep 14) → 36.19% (Sep 28) — +0.64pp across the month on a consistent denominator family, driven by treasuries and a queue that never emptied for long. The yield path is the month's headline: 2.62 → 2.57 → 2.57 → 2.18 on the network APR, with the stETH 7-day average following 2.60 → 2.52. The queues told the same institutional story in volumes: entry 2.07M ETH / 36 days (Sep 2), a mid-month trough, then 2.48M / 27.7 days by the Sep 28 print, while the exit queue ran from exactly empty (Sep 2) to 182,352 ETH (Sep 20) to a 2.8-day wait (Sep 28) — the last September print before the Sep 30 MetaMask disclosure, which is expected to lift the queue into October. The validator count fell through the month on carried prints: 911,414 (Sep 13) → 906,000 (Sep 20/21) — CMv2 consolidation, visible in the carried reads; no fresh count was obtainable later in the month, so the monthly's last validator data point is the Sep 20/21 figure.

Source: Ebunker weeklies (Sep 7, 14, 28 editions); validatorqueue reads (Sep 2, 12, 20); editions 3–6 as published.

7. The peg: a month inside a 63bp band

On the single-read basis, the stETH/ETH ratio opened September at a −18.2bp discount (Aug 31), printed its month-high premium of +28.8bp (Sep 14), its month-low discount of −33.9bp (Sep 25), and closed the month at −5.2bp (Sep 30). Two structural facts survive the noise band. First, the discount never approached levels that imply redemption stress — the worst print of the month, 33.9bp, came on the month's strongest tape. Second, edition six's “first full week at or below par” observation (Sep 21–27) was real on its prints, but the month-wide view shows the ratio oscillating through par in both directions all month — the weekly framing was a function of where the week sat in the oscillation, not a regime change. The month-long view is the honest one, and this report exists to provide it.

Source: stETH and ETH series, DefiLlama coingecko feed, one read session (Aug 31 and Sep 30 points); feed re-print noise disclosed in the ledger below.

8. Treasury demand and the rate path

The month's institutional arc, dated: Bitmine's staked position printed 5.07M ETH (Sep 8), its treasury 5.98M (Sep 21) and 6,001,302 ETH (Sep 28 SEC exhibit) — ending September behind roughly 11.6% of all staked ETH (DERIVED on its 5,067,309 staked against the 43,681,037 ETH Sep 28 board), with its 7-day yield print easing 2.62% → 2.57% → 2.62% across the month's disclosures. The rate path ran the other way: the Fed's 25bp hike on Sep 16 (first since 2023) to 3.75–4.00%, the 10-year Treasury at 5.11% on Sep 23 (its highest since 2007), the 30-year at a 22-year high the next day. The cash-versus-staking gap that defines this site's carry table set a new high every week of the month. The September 30 MetaMask disclosure closes the month's record — the single event that connects a quiet staking month to an unsettled start to October.

Source: Bitmine disclosures via Cryptopolitan and MEXC research; Fed decision and Treasury prints per editions 5–6 and Gate research (Sep 25, 2026); MetaMask per the Sep 30 disclosure.

9. The September data-quality ledger

The consolidated list of every correction, revision and unresolved conflict the month produced. Nothing here is new except the assembly — each item was disclosed in its edition of origin.

• The September 2025 queue misattribution (Sep 29 correction). Edition four built its queue story on figures that belonged to September 2025 — a 2.656M ETH exit queue and a 46-day wait. The error propagated across five pages; all five now carry correction banners linking to the corrected record, and the site's year-check discipline (confirm the year of every undated queue reference against physical churn limits) dates from this catch.
• Price-feed re-prints, disclosed twice. The feed re-prints its own history: Sep 20 prints re-read on Sep 29 (edition five's page keeps its values; edition six restated its cross-week math), and Sep 27 prints re-read in a single later session (edition six's baseline revised −0.4% to −0.6%; this month's table runs entirely on that session). The Sep 27 peg print flipped sign across re-reads (5bp discount to 5bp premium) — inside feed noise, both prints on file.
• Two arithmetic fixes in edition six (Sep 29 audit). Bitmine's projected staking revenue restated $357M → $359M on the printed price, and the two-year carry gap $574 → $575. Both were our own derivation errors, caught by re-computation, corrected in place.
• The rewards-paid counter conflict (open). lido.fi's cumulative “rewards paid since 2020” printed $3.085B at our Aug 27 read and $2.107B since — a cumulative counter cannot fall. The conflict is carried open; the lower figure prints as current with the conflict labeled.
• The stETH supply print age (open). The 9,659,594 dated print (Sep 9) served all month; the supply×price market cap sits ~1–2% below the TVL read throughout, and the two-source supply conflict (aggregator 9.66M vs lido.fi counter 8.89M) remains unresolved at month-end.
• Series changes, disclosed at the point of use. Prices to the DefiLlama coingecko feed (disclosed in edition 4, section 6, after the CoinGecko close page became unreachable); queue days to Ebunker (edition 6 onward). Where a fresh staking board was unobtainable inside September, the board is carried rather than refreshed, and the validator-count path in this ledger stops at the Sep 20/21 figure accordingly.
• Provisional ETF prints. Every September session in this report's totals uses Farside final prints; where a provisional read diverged intraday, the final print is the one carried.

Source: editions 4–7 data-quality sections; correction banners on the archived pages listed above.

10. Method and coverage note

Coverage: stETHHub published its first edition on Aug 27, so September is the first calendar month the site covered end to end on its weekly cadence — editions 3, 4, 5 and 6 plus this report. Unlike August, a month-open anchor exists: the Aug 31 prints come from the same read session as the Sep 30 close, so every monthly change in sections 2, 5 and 7 is computed on one basis. Weekly figures inside sections 3, 4, 6 and 8 are quoted from the editions of origin on the bases those editions disclosed (including the Sep 20 and Sep 27 re-read chains). Where a weekly print and this month's re-read differ, the ledger in section 9 says which basis each number runs on. Nothing in this report is a forecast; the rate board below the fold updates weekly.

Source: site_data.py (Snapshot #7) and the edition archive; full method notes in the weekly data report and the about page.

FAQ

Related reading

• stETH Market Intelligence — August 2026 Monthly Report
• stETH Yield Update — Week of September 28–October 4, 2026
• Lido & stETH Weekly Data Report — Week of September 28–October 4, 2026