The publicly accountable sentiment engine

Every alt-data vendor claims their signal works. We are, to our knowledge, the only one that pre-registers its calls in public — hash-stamped before the earnings print, graded honestly after, never deleted. Hits and misses both stay on this page forever.

no entries graded yet — first print pending

The rules of this ledger

Pre-registered: every entry is published before the company reports, with the publication date and a SHA-256 hash of its immutable fields shown below it.
Append-only: facts and predictions are never edited after publication. The only thing ever added is the post-print grade.
Graded against stated rules: each entry defines upfront how it will be judged. No moving the goalposts.
Honest scope: RSAS measures customer-review sentiment. Our validated claim is 14-day sentiment persistence (walk-forward IC +0.30). We do not forecast stock prices, and we say so.
Misses stay public. That is the whole point.

Shake Shack (SHAK)

PENDING — awaiting print

Pre-registered: 2026-07-04 · Earnings print: 2026-07-30 · Data as of: 2026-07-04 (latest captured review: 2026-07-03; 527 reviews in trailing 28d across 40 locations; 0 reviews filtered as suspected manipulation)

What our data showed (facts, frozen at publication)

food_quality: trailing-28d sentiment +0.086 vs 210-day baseline +0.202 (declining, 210 aspect mentions)
service_staff: trailing-28d sentiment +0.176 vs 210-day baseline +0.264 (declining, 189 aspect mentions)
cleanliness: trailing-28d sentiment -0.031 vs 210-day baseline +0.219 (declining, 25 aspect mentions)
speed_wait: trailing-28d sentiment +0.023 vs 210-day baseline +0.105 (declining, 56 aspect mentions)
order_accuracy: trailing-28d sentiment -0.007 vs 210-day baseline +0.077 (declining, 30 aspect mentions)
price_value: trailing-28d sentiment -0.025 vs 210-day baseline +0.086 (declining, 22 aspect mentions)
RSAS v2.1 composite: -4.6 (HOLD), data-sufficiency confidence 91/100.
Context: Shake Shack cut FY2026 guidance on 2026-06-02 (revenue and same-store-sales outlook lowered). One month later, ALL six tracked customer-experience aspects sit below the chain's own 210-day baseline.

Pre-registered expectation

Primary (based on our validated 14-day sentiment-persistence relationship, walk-forward IC +0.30): this broad-based softness persists — when we recompute trailing-28d aspect sentiment on 2026-07-28 (two days before the print), at least 4 of the 6 aspects will still be below their 210-day baselines. Secondary (exploratory, NOT a validated relationship): we would be surprised if management reported a strong positive inflection in traffic/comp momentum for the weeks covered by this window. We do not forecast SHAK's stock price.

How this will be graded

Primary: on 2026-07-28 we re-run the same RSAS v2.1 aspect computation. HIT if >=4 of 6 aspects remain below baseline; MISS otherwise. Objective and reproducible.
Secondary: within 7 days after the 2026-07-30 print we publish a comparison of company-reported traffic/SSS commentary vs these drivers, graded hit/miss/mixed in the notes. Labelled exploratory.
This entry is never edited; only a grade is appended. The SHA-256 hash below covers all pre-registered fields.

SHA-256 (immutable fields): 29f839eea80a684c9eebd6c40ebff2e34efce50958ca3806f0119fd63566a17e

Domino's Pizza (DPZ)

PENDING — awaiting print

Pre-registered: 2026-07-04 · Earnings print: 2026-07-20 · Data as of: 2026-07-04 (latest captured review: 2026-07-04; 409 reviews in trailing 28d across 80 locations; 0 reviews filtered as suspected manipulation)

What our data showed (facts, frozen at publication)

service_staff: trailing-28d sentiment +0.062 vs 210-day baseline -0.122 (improving, 57 aspect mentions)
food_quality: trailing-28d sentiment +0.011 vs 210-day baseline -0.055 (improving, 126 aspect mentions)
speed_wait: trailing-28d sentiment +0.044 vs 210-day baseline +0.310 (declining, 26 aspect mentions)
order_accuracy: trailing-28d sentiment -0.036 vs 210-day baseline -0.126 (improving, 41 aspect mentions)
price_value: trailing-28d sentiment -0.062 vs 210-day baseline -0.171 (improving, 5 aspect mentions)
RSAS v2.1 composite: +52.9 (BUY), data-sufficiency confidence 89/100.
Context: after the Q1 comp miss (US SSS +0.9% vs ~+2.3% expected) and June price-target cuts across the street, our data shows customer-experience sentiment RECOVERING vs baseline on service, food quality, order accuracy and price/value - while speed/wait sentiment has deteriorated sharply vs its baseline.

Pre-registered expectation

Primary (validated persistence relationship, IC +0.30): the split picture persists - when we recompute trailing-28d aspect sentiment on 2026-07-18 (two days before the print), at least 3 of the 4 currently-improving aspects (service_staff, food_quality, order_accuracy, price_value) will still be at or above their 210-day baselines, AND speed_wait will still be below its baseline. Secondary (exploratory, NOT validated): a customer-experience recovery with deteriorating speed/wait is consistent with stabilizing demand plus delivery-capacity strain; we flag delivery service times as the metric to watch in management commentary. We do not forecast DPZ's stock price.

How this will be graded

Primary: on 2026-07-18 we re-run the same RSAS v2.1 aspect computation. HIT if >=3 of the 4 improving aspects remain at/above baseline AND speed_wait remains below baseline; MISS if neither holds; MIXED if only one of the two conditions holds.
Secondary: within 7 days after the 2026-07-20 print we publish a comparison of company commentary vs these drivers, graded in the notes. Labelled exploratory.
This entry is never edited; only a grade is appended. The SHA-256 hash below covers all pre-registered fields.

SHA-256 (immutable fields): 9d5288b8191e7c463e5cc053a6382440999493ab0ae923fcaee880a88c9f91f0

Skin in the game — the Accountability Guarantee

Talk is cheap, even hash-stamped talk. So we put revenue behind this ledger:

If any primary pre-registered prediction on this page is graded a MISS, every active data-license subscriber gets their next month free.
Applies to Monthly and Weekly Data License subscriptions active on the day the grade is published. Credit is applied to the next invoice; capped at one free month per calendar quarter; no cash value.
We grade ourselves by the rules written into each entry before the print — and the entry hashes above make those rules impossible to quietly rewrite.

Large vendors cannot copy this without repricing their entire book. We can, because we would rather lose a month of revenue than a decade of trust.

Verify it yourself — don't trust us

Machine-readable ledger: /proof/ledger.json — the exact source this page is rendered from. Each entry's sha256 is computed over its immutable pre-registration fields (sorted-key canonical JSON of: id, chain, ticker, published_at, earnings_date, data_as_of, facts, prediction, grading_rules).
Bitcoin timestamp: /proof/ledger.json.ots is an OpenTimestamps proof anchoring the ledger's hash in the Bitcoin blockchain. Verify with: ots verify ledger.json.ots -f ledger.json. No blockchain hype — just a free, independent clock nobody (including us) can rewind.
Independent web archive: snapshots of this page are stored by the Internet Archive Wayback Machine — a third party we don't control.