IQR vs PCT reward bands

FTSO providers aren't paid for showing up — they're paid for being accurate. Accuracy rewards are split across two bands measured against the consensus median every round.

Two bands, 40 / 60

• Primary — the IQR band (≈40% of accuracy rewards). The interquartile range is the spread of the middle 50% of providers by vote weight. It is narrow when the field agrees and wide when it doesn't. Landing inside it is what separates providers, because it moves with the market instead of sitting still.
• Secondary — the PCT band (≈60%). A fixed half-width in parts per million set per feed by governance (tight for stablecoins, wider for volatile assets). Easier to hit; the baseline.

The 40/60 split is a governance parameter (set in FIP.11). Both bands pay out proportionally to a provider's (capped) weight.

What a miss costs you

The bands are pass/fail per round, per feed. A submission inside the IQR earns its weighted share of that round's primary pool; a submission outside earns nothing from it — there is no partial credit for being nearly right. Miss both bands and the round pays zero. Across an epoch of 90-second rounds that arithmetic is the whole difference between two providers charging the same fee, and it lands on the delegator as a lower reward rate per FLR. This is why we score ourselves on band hit rate rather than on uptime, and why Providers prints our row in the same sort as everyone else's.

Why "weighted" changes everything

The quartiles are computed on a weight axis, not a head-count. Providers are sorted by submitted price; weight is accumulated; Q1 is the price where cumulative weight first passes 25%, Q3 where it passes 75% (a step function — no interpolation, with ties broken deterministically). A handful of heavy providers can define the band, so a light provider has to track the heavy cluster, not the simple average.

The 2.5% cap

No single provider's weight can exceed 2.5% of the total when bands and payouts are computed; the excess is redistributed. This keeps one large entity from defining "consensus" by itself — and it is why the reward rate per delegated FLR can be higher with a mid-sized provider than with the very largest. See Delegation 101.

The same idea, live

Below is BTC/USD from the live network — the gold line is the consensus median, the inner band the IQR, the outer band the secondary band, and the bright dots are our own submissions. Switch to % deviation to see how tight "agreement" really is.

Last reviewed June 2026

Convinced by the data? Delegating takes two minutes. Delegate to Seraphim →