pbs step 5 FAILEDp.433 -- content and relationships, in the SOURCE — RuntimeError: `pbs`'s natural_key does not resolve to bare columns of `pbs_story_class`. …
edgar_docs step 5 FAILEDp.433 -- content and relationships, in the SOURCE — MemoryError: …
aapb step 5Now (2026-10-22 21:19, OWED): 113527 table(s) had two EQUALLY good provider surfaces offered and `aapb` declares nothing that separates them. …
What tripped it (2026-10-16 04:28): p.433 -- content and relationships, in the SOURCE — AttributeError: 'dict' object has no attribute 'lower'
… and 6 more.
The walk stops here: the next rank is not released until this is answered.
L1 is the warehouse itself — the Kimball star. Thirty-four subsystems walked in the book’s own order, every source profiled, screened and certified against the page it came from, so that what reaches a decision has already been measured rather than assumed.
Wall Street quality. Detective grade. Delivered to the kitchen table.
Financial data arrives dirty — quietly restated, survivorship-biased, revised after the fact without a word. The industry’s answer is to sample it, average over the mess, and call it clean. Ours is to read every value end to end, and publish what we find.
Three things most vendors will not do. We keep the companies that died. A universe that quietly drops the bankruptcies and the buyouts makes every strategy look brilliant; ours holds the delisted beside the living, because a test that never meets failure has not been tested. We keep the number as it was known. Every filing carries the moment it became public, so a figure restated two years later cannot leak backwards into a decision nobody could have made at the time. And we publish our own defects. The flaws this system finds in its own work sit on this board beside the successes — a record that can only say yes is a brochure.
The work is the moat. Not a sample, not an average: every column in every source read wall-to-wall for nulls, extremes and cardinality, and counted. The figure below is what that adds up to — and it is not a marketing number, it is arithmetic on the record, re-measured the moment you refresh this page. No shop reaches it by hand, which is the whole reason the kitchen table has never been offered data like this before.
And we are graded against a standard we did not write. The Data Warehouse Toolkit, 3rd ed. — the book the banks and the exchanges build on. Thirty-four subsystems, walked in its own order, every box a certification against its page, signed and dated by BrainTrust Staff. Anyone can buy that book and mark our homework; that is the difference between a claim and a proof.
What the rest of the field actually sells
The other competitors fall into five tiers. We name none of them — a logo dates a page and invites an argument about a brand; the SHAPE of each tier is the durable fact, and it is the one that makes the case.
Nobody at any price sells the join. That is the finding, and it held across every tier surveyed. Insider transactions, ownership edges, failures-to-deliver, short interest, fundamentals and prices — on one conformed bus, anchored to the filer, resolved to the entity. The tiers above show you filings. This shows you the people and the money as one fabric, which is where a tell actually lives.
And the honest other side, because a pitch with no gaps in it is the thing we are arguing against. We are US-centric where the broad screeners are global. We carry no analyst estimates. The polished interface is a later level and is not built. Those are on the record here for the same reason the defect count is: you should not have to find them yourself.
The delivery is the point: all of it is for our Campers — we grow Investors through our Summer Camp for Kitchen Table Investors, and the kitchen table deserves institutional-grade truth. What you are watching is that build, live. Refresh, and every number on this page re-measures from the record.
We don’t guess and we don’t trust, We Measure!Nothing here is an estimate or on someone else’s word.
What an ETL process is — and why your kitchen table needs one
ETL stands for Extract, Transform, Load — the three jobs that stand between raw data and a question you can actually answer.
Extract pulls records from wherever the world publishes them: an SEC filing, a price history, a government release. Transform repairs what is wrong and makes the pieces agree — one company spelled four ways becomes one company, and a figure quietly restated years later is still recorded as it was first known. Load puts the result where it can be asked hard questions quickly.
Here is why that reaches your table. A wrong number never announces itself. A price that silently excludes the companies that went bankrupt makes every strategy look brilliant. An earnings figure revised two years after the fact makes a decision you could never have made look obvious. Nobody sends you a correction. You simply act on it, and the mistake shows up later as money.
An institution spends millions on an ETL process so its analysts never touch a number that has not been through one. A household normally gets the raw feed and is expected to cope. This is the same machinery, built for the kitchen table — run on a schedule, the same way every time, so the answer you get on Friday was built exactly like the answer you got on Monday. Every serious warehouse runs one. The difference is whether anyone will show you theirs.
Who was Dr. Ralph Kimball — and why he matters at your table
The man who wrote the field’s standard. Dr. Kimball took a PhD at Stanford, helped design the Xerox Star at PARC — the workstation that put windows, icons and a mouse in front of the world — and founded Red Brick Systems, a database built for exactly this kind of work.
His Data Warehouse Toolkit, written with Margy Ross, is the book most warehouses are measured against. It is where the 34 ETL subsystems on this page come from — in his order, in his four groups, cited to his page numbers.
Why that matters to you, and not just to us. Anyone can claim their data is clean. The claim is worth nothing unless there is a standard outside the claimant to check it against. Dr. Kimball wrote that standard, banks and exchanges are built on it, and it is published — you can buy the book and mark our homework.
So BrainTrust does not improvise a method and ask you to trust it. It follows his, page by cited page, and every box below is graded against the page it came from. The same standard the institutions are held to, pointed at your table instead of theirs.
How to read this board
Each card is one subsystem. Its colour is the walk — whether that stage has actually been run and certified. Its badges are the code — how many programs serve it, whether they sit in their official place, and whether the register knows them.
Those two can disagree, and the disagreements are the point: a stage marked done while still carrying gaps, or one holding programs it has never run. A dashed outline means work that is declared but not built.
We publish the gaps. A board that showed only finished work would be a brochure; this one is measured fresh on every visit and is allowed to say no.
The work, mounting
Read live from the record on every refresh — measured by BrainTrust Staff, never estimated.
33,291,123,997 values measured wall-to-wall by the star’s instruments — every one of them read wall-to-wall, not sampled. One person completing these complex calculations on a calculator, one every minute, around the clock without stopping, would need 554,852,067 man-hours to do the same work by hand.
L1 — The Kimball Star — the ledger
| what | Last Night | Last 7 Days | Last Month | Last Year | Total |
|---|---|---|---|---|---|
| Values measured | 2,020 | 2,020 | 1,374,943,799 | 33,291,123,997 | 33,291,123,997 |
| Facts held in the provider record | — | — | — | — | 2,544,461,235 |
| Certifications signed | 95 | 2,009 | 3,224 | 3,564 | 3,564 |
| every one a named claim against the book's own page | |||||
| Acts witnessed in the record | 9,513 | 26,792 | 388,896 | 634,673 | 634,673 |
| nothing the machinery does goes unrecorded | |||||
Room reserved: L0 — The Chaplain (governance and self-correction) · and the levels beyond.
The start-up, walking
Certified work fills from the left; the run under way follows after the gap; the white behind the black edge is the road ahead. Measured from the walk record on every refresh.