Staking yield looks like one number. On a dashboard it often is. Underneath, it moves every day, and a single day can be badly misleading.
Lido's own data shows this clearly. Over the seven daily reports to 6 November 2026, the stETH staking APR ranged from 3.227% to 2.263%. The 7-day average was 2.486%. Anyone who read only the highest day would have overstated the yield by almost a third.
This is a worked example. We take Lido's published APR series, read it three ways and show why the average and a defined middle band tell a more honest story than any single day.
The Yes/No/Maybe mechanic, briefly
Oddup markets have three outcomes: Yes, No and Maybe. Yes and No split 90% of the pool reserve. Maybe wins 10% of the pool reserve. Maybe is a hedge, not a guaranteed win on every pool. It suits questions with a real middle state that the rules define before trading opens. Yield questions are a natural fit, because "roughly where it was" is often the most likely answer.
Where the stETH APR comes from
stETH is Lido's liquid staking token for Ether. Holders do not claim rewards. Their balance grows instead. As Lido's help centre puts it, every day at 12PM UTC the amount of stETH in your address increases with the current APR.
That daily balance change is called a rebase. Lido publishes two APR figures through its public API:
- Last APR. The latest staking APR value, calculated from the most recent rebase. It compares the share rate before and after the report and annualises the change over the time elapsed.
- SMA APR. In Lido's words, an APR "based on Simple Moving Average of APR values over a period of 7 days".
Both come from the same on-chain events. The difference is the window. One reads a single day. The other reads a week.
The rewards themselves come from running Ethereum validators. As ethereum.org explains, stakers earn rewards for helping secure the network. Part of that reward is steady protocol issuance. Part comes from transaction activity, which rises and falls with demand for block space. That second part is a key reason one day can differ sharply from the next.
The live numbers
We pulled Lido's 7-day series from its public API on 6 November 2026. Each value is the APR reported at the daily rebase, all at roughly 12:22 UTC.
- 31 October: 3.227%
- 1 November: 2.343%
- 2 November: 2.279%
- 3 November: 2.301%
- 4 November: 2.275%
- 5 November: 2.263%
- 6 November: 2.713%
Lido's API reported the 7-day simple moving average as 2.486%. We checked it against the seven values above. The arithmetic matches.
Two things stand out. First, the series is not smooth. It fell almost a full percentage point in one day, from 3.227% to 2.343%. Second, five of the seven days sat in a tight cluster between 2.263% and 2.343%. The two outliers pulled the average up.
Worked example: read the same week three ways
Read 1: the latest day. On 6 November, the last APR was 2.713%. A dashboard showing only that figure would suggest yield was rising. It was the second-highest day of the week. It was also 0.45 points above the day before.
Read 2: the best day. The highest value was 3.227% on 31 October. A reader who anchored on that day would expect almost 30% more yield than the 7-day average delivered. That gap is the cost of reading the peak.
Read 3: the 7-day average. At 2.486%, the average sits between the cluster and the outliers. It is not the yield on any single day. It is a fairer estimate of what a holder earned across the week.
Now turn each read into a market question. Suppose an illustrative binary asks: "Will the stETH APR be above 2.5%?" This is not a live market. It is a teaching example.
- Settled on each daily value, the answer would have been Yes on 31 October, No for the next five days, then Yes again on 6 November. Two flips in seven days.
- Settled on the 7-day average, the answer would have been No. But only just. The average sat 0.014 points below the strike.
A binary on the daily figure was noisy. A binary on the average sat almost exactly on the line. Neither gave a trader a clean read.
Add a Maybe band
Now take the same illustrative question and add a defined middle. Suppose the rules say:
- Yes: APR above 2.75%.
- Maybe: APR from 2.25% to 2.75%.
- No: APR below 2.25%.
Against the daily values, six of the seven days land in Maybe. Only 31 October lands in Yes. None land in No. Against the 7-day average, the answer is Maybe, comfortably inside the band.
That is a much more stable question. The middle band absorbs the day-to-day noise. A trader can now ask a real question: will something push yield clearly out of its recent range? That is a view worth holding. "Will one daily print land above 2.5%?" mostly is not.
The band only works if it is fixed before trading opens. Drawing it after the data arrives is hindsight, not analysis. We cover this in define the Maybe band first.
Pick the source and the clock
Any yield market needs two more rules before it can settle cleanly.
The source. "Staking APR" can mean several things. It could be Lido's last APR, its 7-day SMA, a third-party dashboard or a figure calculated directly from rebase events. Each can show a different number on the same day. A market has to name one. As our piece on oracles argues, you should always read the data feed first.
The clock. Lido's rebase happens once a day, around 12PM UTC. A market that settles at 11AM UTC reads yesterday's value. One that settles at 1PM UTC reads today's. If an oracle report is late, the rules need to say what happens. Every yield market needs a market clock, not just a strike.
Get either rule wrong and the market can settle on a number nobody expected, even when every trader read the data correctly.
Why averages still need care
The 7-day average is better than a single day. It is not perfect.
- Averages lag. If yield shifts to a new level, a 7-day average takes a week to fully reflect it. On 6 November, the average still included the 3.227% reading from 31 October.
- Outliers move averages. One high day lifted the week. Without the 31 October value, the remaining six days average 2.362%.
- APR is not a promise. Past rebases describe what holders earned. They do not set what holders will earn next week.
None of this makes the average useless. It means the average is a starting point. A band around it is a better question than a point on it.
Three checks before reading any staking yield
- Ask which window. Is the figure a single day or an average? If it is an average, over how many days?
- Look at the range, not the point. Read the high, the low and where most days cluster. A tight cluster with two outliers tells a different story from a steady drift.
- Match the market to the view. A view on "roughly unchanged" belongs in a band. A view on a single print belongs in a binary, with the noise priced in.
A common misreading, worked through
Suppose someone checked the dashboard on 31 October, saw 3.227%, and concluded that staking yield was rising. Six days later the daily figure was 2.713%, still above every reading from 1 to 5 November. A quick glance could make the week look strong. The series says otherwise. Once you leave out the 31 October spike, the other six days average roughly 2.36%. Five of the seven readings sat between 2.263% and 2.343%. The 6 November figure was the second-highest day of the week, which is why it stands out. That does not make it a new trend.
The lesson is not that the high readings were wrong. Each one was a real rebase. The mistake is treating one rebase as the level. Write the market question around the level, and give the edge cases a defined place in the middle.
Why this matters for prediction traders
DeFi yield is one of the most quoted numbers in crypto. It is also one of the most misread. A single daily APR is a snapshot of one rebase. It can sit far from what holders earned across the week.
In Lido's data, the daily APR ranged from 3.227% to 2.263% over seven days. The 7-day average was 2.486%. A binary at 2.5% would have flipped twice on daily values and sat almost on the line on the average. A defined Maybe band from 2.25% to 2.75% captured six of seven days.
That is the case for three-outcome design in DeFi. Yield clusters. Outliers happen. When the middle is the most likely result, it should be a tradable outcome with its own rules. The same logic applies to Uniswap liquidity ranges: in, out or on the edge is a more honest question than above or below.
This article is for educational and informational purposes only. It is not financial, investment, legal or tax advice, and it does not recommend staking, any protocol or any trade. The example market is illustrative and not a live Oddup market. APR figures are quoted from Lido's public API at the stated time, describe past rebases and do not predict future yield. Staking and DeFi protocols carry risks, including smart-contract risk and loss of funds. Prediction markets involve risk. Availability and eligibility vary by jurisdiction. Oddup does not make price or yield predictions or guarantee outcomes.
