The market
$25B+
of tokenized US equities cleared through xStocks; ~95%+ settles on Solana. Tokenized RWAs on-chain now sit near $29B outstanding.
M6 deployed · Paper 1 → arXiv · Open source
Real-world assets
Real-world assets
don't sleep on the weekend.
But the oracles that price them do. Soothsayer serves a fair-value band for the hours the market is closed — with a receipt anyone can check. Don't trust the price. Verify the band.
The gap
Every weekend, the incumbents stop thinking.
Incumbent feeds either freeze at Friday's close, route to an opaque tokenized mark, or extend into 24/5 and still hand back the weekend — none of it auditable. And the market has noticed: these tokens trade roughly 100× more than they're held as collateral (>$25B vs ~$225M) — Kamino disables borrowing on half its xStock reserves, and MarginFi lists none.
MonTueWedThuFriSatSun
19% of the week the reference market is open — 32.5 of 168 hours. That green sliver is the only time every incumbent oracle has a real price to serve.
81% is off-hours — capped by the 65.5-hour Friday-close→Monday-open window — where feeds go stale, synthetic, or opaque. Soothsayer serves a checkable band across all of it.
What current providers tell you
Auditable risk. Tighter reserves. Free and open.
What each surface actually publishes once the reference market closes — measured on the tape and from primary docs. Only Soothsayer serves a calibrated band with a receipt you can re-derive from public data.
| Provider | Weekday overnight | Weekend | Halts | Verifiable coverage claim? |
|---|---|---|---|---|
| Chainlink Data Streamsv8 · v11 | stale-hold | stale-hold / synthetic marks |
stale-hold | No |
| Pythcore aggregate | dispersion → 0 at the close |
dispersion → 0 | status Unknown | No |
| Pyth Pro · Blue Oceanpaid, executable | real book Sun–Thu 20–04 ET |
not covered | not covered | No |
| RedStone Liveoff-chain | 24/7 scalar, undisclosed method |
24/7 scalar, undisclosed |
— | No |
| Kamino Scoperoutes Chainlink | ±500 bps clamp on Fri close |
±500 bps clamp | clamp | No |
| Soothsayer | calibrated band + receipt |
calibrated band + receipt |
regime-conditioned band |
Yes — re-derivable |
serves a real, priced signal
degraded — no width, collapses, clamped, or undisclosed
stale, synthetic, or unknown
not covered
Market admission
Incumbents understand the risk.
This isn't a hypothetical edge case. The providers themselves are warning that tokenized markets drift from any defensible reference once the venue shuts and the oracle stack can't adapt.
“We might see a dislocation of the tokenized stock versus the real value on Nasdaq.”
Marcin Kazmierczak · RedStone · quoted in CoinDesk
“If the oracle doesn't update until markets reopen, on-chain protocols could be trading on ghost prices.”
CoinDesk weekend-gap coverage · On-chain stocks mispricing risk
Incumbent admission
“Without these capabilities, tokenized products are exposed to potential mispricings, unfair liquidations, regulatory gaps, and other critical risks.”
Chainlink · v10/v11 launch · blog.chain.link
How you use it
Ask for a coverage level. Get a band and a receipt.
One read gives you a band — not a naked point — and a receipt anyone can re-derive from public data. No proprietary feeds, and nothing you have to take on trust.
01
τ ∈ [0.68, 0.99]
Request your coverage level
Ask for a band at a target coverage τ — 0.85 by default, up to 0.99. The oracle returns the price interval that has historically delivered exactly that coverage for this symbol and regime, not a fixed-width guess.
02
PricePoint · soothsayer-consumer
Read the band and receipt on-chain
Every read carries its own receipt — target_coverage, claimed_served, forecaster, regime — right on the
PricePoint. Decode it in your program with the no_std soothsayer-consumer crate. Rust ↔ Python parity is 180/180.
03
reports/m6_validation.md
Verify it yourself
Replay the claimed coverage against the public calibration surface and check it delivers what it says. Held out by symbol: 0.9497 ± 0.0128 at τ=0.95, per-symbol Kupiec 10/10, across 173 weekends and 11 forward-tape weekends since the freeze. Don't trust the price — verify the band.
TSLAxxStock · Sat 14:32 UTC · τ = 0.85 default
Regime · normal
$241.07
point · half-width ±2.30%
band $235.52 — $246.62
claimed_served 0.850
SPYxxStock · Sun 03:10 UTC · τ = 0.85
Regime · normal
$604.11
point · half-width ±1.42%
band $595.54 — $612.68
claimed_served 0.850
MSTRxxStock · Sun 18:40 UTC · τ = 0.85
Regime · high vol
$412.66
point · half-width ±3.94%
band $396.39 — $428.93
forecaster lwc
What it protects against
When the market closes, the risk compounds.
Price uncertainty compounds with every hour the reference market is shut — and it's shut 81% of the week, from a single overnight to a 65-hour weekend. These are the three places that widening gap turns into loss.
01 · LIQUIDATION
A weekend flash crash that reverts by Monday.
732 bps — p99 single-name weekend gap · 5,996 windows, 12-yr panel
WithoutA transient 700+ bps print on a thin weekend book liquidates a healthy borrower, then reverts by Monday. The borrower still eats the liquidation penalty; the protocol booked a loss on a price that never held.
With SoothsayerThe receipt shows the print landing outside the empirically calibrated band, giving the protocol an audit-logged basis to pause liquidations or tighten policy — a rule, not a judgment call.
02 · GAP RISK
An overnight earnings gap the clamp can't hold.
5–10% — routine earnings-night gap · a ±500 bps clamp caps at 5%
WithoutA name reports after Tuesday's close and gaps 8% by Wednesday's open. A fixed ±500 bps clamp still marks the position healthy; the ~300 bps beyond the clamp is bad debt the protocol only sees at the bell.
With SoothsayerThe band widens ahead of the scheduled print — session-timed to the after-hours report — and the point moves with the overnight signal set, so risk engines price the gap before it opens, not after.
03 · ORACLE OUTAGE
The primary feed goes stale mid-window.
65.5 h — Fri close → Mon open with no reference print
WithoutTwo bad options: freeze the market and forgo fees on flow that trades $25B+, or keep serving the frozen number and absorb the gap as bad debt. Today that's the only choice for 65.5 straight hours.
With SoothsayerAn independent, calibration-aware cross-check and fallback that keeps serving through the outage — with uncertainty surfaced as a band, not silently implied by a frozen point.
Partner · support the project
Price the closed window against a band you can verify.
We're looking for a small number of lending, perp, and AMM teams to design against a calibration-transparent closed-market reference — you bring the integration surface, we bring the band and the receipt. Everything is open source and reproducible from public data.
- Integrate — read the band on-chain via the
soothsayer-consumerdecoder. Integration guide → - Fund the work — it's self-funded and grant-seeking; a runway keeps the forward tape and the mainnet path moving.
- Contribute — the oracle, the Rust/Anchor stack, and the research are open source. Star / fork the repo →
- Cite or endorse — Paper 1 (coverage-inversion) is near arXiv submission and seeking a q-fin.RM endorsement.