stETH Yield Update — Week of September 14–20, 2026

1. The rate board

The board is half fresh this week, which is the best it has looked since early September. Two rows are new dated prints, one row changed source, and the rest still carry their original dates rather than an interpolation.

Two disclosures attach to that table. The network APR row changed source this week: our series had been validatorqueue (2.61% Sep 2, 2.59% Sep 10) and now reads Ebunker's weekly at 2.57%. That is a 2bp step between two different providers, not a 2bp move in the market, and we would rather say so than present it as a trend. The stETH 7-day average finally refreshed after six weeks at 2.60% — and it refreshed to 2.60%, which is either a coincidence or a rate that has genuinely stopped moving. We cannot distinguish the two on two prints, so we simply report both.

Source: Ebunker Ethereum staking weekly (Sep 14, 2026); validatorqueue.com series as previously cited (Sep 2 and Sep 10); US Treasury 3M print (Sep 1, carried); 2Y per the Aug 28 keynote-day read.

2. The carry trade, on this week's price

Position value uses the Sep 20 same-source read: 10 ETH = $26,205.90 ($2,620.59). Rates are the board above. Every line is DERIVED, on one price and one set of rates:

The gap is $446 a year against bills and $556 against the two-year — 168–171bp and 210–213bp respectively across the 2.21–2.24% user-net range (DERIVED).

Last week this column printed a translation table before the FOMC met: every 25bp the committee delivered would move the gap 25bp, and a hike would widen it. The committee delivered a hike. Measured against the policy rate rather than the carried bill, the staking shortfall is now 143bp (DERIVED: 4.00 − 2.57), against 135bp on the carried 3.92% bill at the current 2.57% network APR (DERIVED: 3.92 − 2.57). Both readings are on the table because they measure different things: the bill is a market price that has not yet been re-read since the hike, and the policy rate is an administered one that has.

Source: position value from the DefiLlama coingecko feed, Sep 20 print; rates as tabled in section 1; FOMC statement and implementation note, Sep 16, 2026; all conversions our own arithmetic.

3. Peg watch — a 9bp discount, and a 5bp premium

The peg printed its second discount of the quarter this week, and the sequence is the interesting part:

One basis change to disclose, because it changes the series rather than the market. Edition four printed 0.9987 for Sep 15 — a live same-timestamp pair read at the snapshot moment. That read was honest but not reproducible. From this edition the ratio is computed from the daily series' same-source same-day prints, so a reader can re-derive every row above. On the new basis Sep 15 reads 0.99969, a 9bp difference in basis from the 0.9987 we printed; that is why the Sep 16 row, not Sep 15, is this week's widest.

Source: DefiLlama coingecko feed, stETH and ETH daily prints at the same fixed UTC hour, read Sep 20, 2026.

4. LDO's best week since August

LDO rose 12.7% in-window close-to-close ($0.37158 on Sep 13 → $0.41873 on Sep 20). The path was not a straight line: it fell 10.1% from its Sep 13 close to $0.33397 on the Fed day, then gained 25.4% off that low into the Sep 20 close — the sharpest round trip in our series.

The market cap we print changes basis this week, and the change goes against the price move. We had carried 895.8M LDO outstanding from a tracker we could not corroborate; a September 17 market data set puts circulating supply at 834,151,322 against a 1B total. On the new basis LDO's market cap is $349M; the old basis would produce $375M. Every LDO cap this site has printed since August was therefore about 7% too high, and we would rather correct the level than bank a number built on a basis we could not stand behind.

No Lido-specific disclosure in the window explains the move. Two protocol events landed during it — the CMv2 consolidation began Sep 17 and Simple DVT was retired Sep 18 — but LDO's Sep 17 gain of 6.2% on the day came with a market that was repricing risk appetite, not a governance catalyst. We describe the flow; we do not assign it.

Source: LDO daily series from the DefiLlama coingecko feed, read Sep 20; circulating supply per Analytics Insight market data (Sep 18, 2026); LDO's Sep 17 move per the same source.

5. Flows that touch the corridor

ETF flows turned negative. The Farside daily table reads +$121.1M (Sep 14), −$142.0M (Sep 15), −$224.1M (Sep 16), −$39.3M (Sep 17), +$143.7M (Sep 18) — a window net of −$140.6M (DERIVED), the first negative five-session stretch since August and a $337.5M swing from the prior week's +$196.9M. The two heaviest redemption days were the Fed statement day and the day after; the two inflows bracketed them.

The queue stays one-directional. Entry: 2,480,000 ETH, 43.1 days. Exit: 182,352 ETH, 3.2 days. A ratio of 13.6× (DERIVED) against a protocol churn limit of roughly 57,600 ETH per day. For the corridor this is the single most important structural fact on the page: a month-long activation queue means new stake cannot respond to a rate change quickly, in either direction.

Concentration, updated. BitMine reported more than 5 million ETH staked, with its treasury put at $15.8 billion and expected annual staking revenue of $334 million. Against 43.18M ETH staked network-wide, that remains roughly 11.5% of all staked ETH (DERIVED). One corporate balance sheet sits behind about an eighth of the network's security — and it is the most rate-insensitive staker on the board, because it is not comparing against a Treasury bill.

Source: Farside Investors Ethereum ETF daily flow table (accessed Sep 20, 2026); validatorqueue.com via Tremplin (Sep 20, 2026); BitMine figures via the Ebunker weekly (Sep 20, 2026), which cites the company's disclosure.

6. What would move this update next week

Three things, each with a number attached. A fresh dated stETH supply print would settle the 1.6% cross-check gap that this site's data report is now carrying open — at $2,621.87 the gap is worth about 150,000 stETH, and a single dated read either closes it or converts it from staleness into growth. A second same-source rate print on the frontend would re-arm the top half of the board, which has been carried since Sep 7 and is the weakest leg of section 1. And the next policy-rate print matters more than usual: this week's translation table was tested once and held, which is a sample of one.

What we would not read as a signal: another 20bp of peg premium. The ratio has now printed above par on six of the last ten sessions and below it on four, all inside ±19bp. That is a wrapper tracking its redemption mechanics, not a market making a statement about Lido.

Source: supply and TVL reads as cited in section 1 and the data report; peg series from the DefiLlama coingecko feed, read Sep 20, 2026; FOMC decision, Sep 16, 2026.

7. Gross versus net: what a staker actually keeps

The rate board in section 1 prints three different stETH yields, and the distinction decides every number in the carry table. The network staking APR is 2.57% — protocol-level, what the chain pays. The stETH 7-day average is 2.60% — Lido's smoothed headline. But the user-net band we actually use is 2.21–2.24% (carried since Sep 2): what a holder keeps after Lido's fee and the smoothing buffer. The carry table runs on 2.22% for a reason — it is the number a staker can bank, not the number a dashboard shows.

The 38bp spread between the 2.60% headline and the 2.22% net (DERIVED: 2.60 − 2.22) is not an error; it is the cost of the wrapper, and it is the most common misread of this column. It is also why the "staking yield" a casual reader sees overstates the carry appeal by roughly that 38bp. We flag it here because the 7-day average refreshed this week to the same 2.60% it has held for six weeks — the net band did not move either, so the gap widened by exactly nothing, and a staker's effective yield is unchanged despite a 9% move in the underlying.

Source: network APR and stETH 7-day average per Ebunker Ethereum staking weekly (Sep 14, 2026); user-net band as previously cited (Sep 2); all conversions our own arithmetic.

8. The LSDfi shelf competes for the same ETH

The carry trade is not Lido's alone. The restaking and liquid-staking shelf holds real capital that would otherwise sit in stETH. EtherFi carries $4.91B in TVL (DefiLlama, Sep 20) on its liquid-restaking token, and Rocket Pool $1.37B on rETH, whose APY mid-point is 2.19% (carried since Sep 2). These are not direct substitutes for stETH's redemption-backed liquidity, but they compete for the same yield-seeking ETH, and they are the reason a 143bp policy gap does not automatically trigger redemptions — there is no obviously better on-chain dollar yield either.

The structural question this raises ties straight back to CMv2. If Lido's migration pushes its operator set toward 34 curated, bonded operators, the permissionless, higher-decentralisation shelf — Rocket Pool especially — becomes relatively more attractive to the subset of stakers who price decentralisation above the 2.22% net. We are not claiming flows have moved; the entry queue says they have not. Only that the competitive axis just tilted, and the next edition's queue and TVL reads will tell us whether it mattered.

Source: EtherFi and Rocket Pool TVL per DefiLlama API direct reads (Sep 20, 2026); rETH APY per the section 1 board (carried Sep 2).

9. Where the 143bp gap sits in the series

The carry gap is the through-line of this column, and this week it reached a series high. Measured against the policy rate the staking shortfall is 143bp (DERIVED: 4.00 − 2.57); measured against the carried 3.92% three-month bill it is 135bp (DERIVED: 3.92 − 2.57). The bill reading is the one we have printed for weeks, and it barely moved — but the bill is a market price that has not been re-read since the hike, while the policy rate is an administered rate that has.

The reason 143bp is a record for this column is arithmetic, not drama: the policy rate just rose 25bp to its highest since the series began, while the network APR fell 2bp. Every prior edition printed a smaller policy gap, because the policy rate was lower and the network APR higher. That is the whole macro story in one figure — staking yield is being squeezed from both sides, and the only thing keeping stETH attractive is that the alternatives are squeezed harder. We will keep printing both readings, because they measure different things, and the gap between them is itself a signal.

Source: policy rate and SEP per FOMC (Sep 16, 2026); network APR per Ebunker weekly (Sep 14); bill rate carried (Sep 1); all gaps DERIVED.

FAQ

Related reading

• Lido & stETH Weekly Data Report — Week of September 14–20, 2026
• stETH Yield Update — Week of September 7–13, 2026
• 8.4M ETH on the Move: What Lido's CMv2 Consolidation Actually Changes