The record · public · graded nightly

Every number here has a shadow.

We grade our levels, our bands and our alerts every night — and every rate is printed beside what a random level of the same distance did. Anyone can publish a percentage. The control column is the part that costs something.

Sealed before the openGraded after the closeControls beside every rateNulls published
The seal · a hash chain, not a promise

Yesterday cannot be rewritten.

Each morning's levels are hashed at 9:12 ET, and each day's seal contains the previous day's — a chain. Edit one historical level and every seal after it breaks. These are the real, current seals:

2026-09-10c97e536e43c159cd1d172f5b5b1d0ae8ed4b9a431bc265ff6c082bd1c5d9ebc1 · 9 boards
↳ containsa61c49af52a27a0bcf8b23427d63529112= 2026-09-09's seal, verbatim

Why this matters: we measured that the industry's favorite level "walks to price" intraday — evening charts show price at the level 55% more often than the morning's actually did. So ask any levels service for their morning level. Ours is math; it can't move.

Levels · graded against distance-matched controls

The test that kills most level claims: compare every touch against a random strike the same distance from price. What survives it, we show. What doesn't, we say so below.

the claimmeasuredthe controlsample
The Lid holdsheaviest cap overhead — price stays capped+9–11 ptsover controlrandom level, same distance70,000+ boards
The gatekeeper blocks the patha heavy node between price and target — the target rarely trades16.4%47.2%reach with no gatekeeper, distance-matched0.5–1% band
The board's edges fadetouch the extreme node, fade it56–60%≈50%midpoint of the same board — coin flipboth sample halves
All three index books negative → range expandsthe day-type read on the strip75%35%every other configurationmatched on prior range
The expected-move band · target beside achieved
80%the target80.1%achieved · daily80.9%achieved · weekly680sessions graded

A band that claims 80% and delivers 80.1% is calibrated. One that delivers 95% is padded; one at 60% is broken. The claim is its own control — walk-forward, per symbol, refused where history is thin.

Alerts · counts, never win rates
601fired · 60 days305ran +50%198doubled100%delivery-receipted

Entries are stamped from the real trade nearest the post — not the day's convenient print. The board grades the same population Discord received. Peaks are the move that was there, not a result: entry, size and exits are yours.

What we tested and refuse to sell

The moat is the disproof. Each claim below is sold as a feature somewhere in this market today — and each one died against its control in our data. That's why you won't find them here.

the folk claimmeasuredthe controlsample
“Price is drawn to the biggest level”the magnet — a competitor's headline stat50.5%50%a coinn=44,958
“The squeeze comes before the takeoff”tight ranges predict launchesinvertedquiet stays quietmonotone, both halvesn≈650,000
“Levels turn price on touch — 95%+”the touch-decay pitchany levelmirror identicala property of touching, not the leveldistance-matched
“The end-of-day gamma rug”short the ceiling into the close−0.01%mirror faded moremedian slide ≈ one index point118 sessions

A page like this cannot be copied in a night — publishing your nulls means having run the controls that would kill your own marketing.

469,523 graded level outcomes590 boardsnightly · misses included, none removedstructure, not advice