Atlas reachability screen
Atlas reachability screen is one of 21 measurements this engine publishes in full. It is regenerated from a real run rather than transcribed, and it is shown here with its own claim boundary so the number and its limits arrive together.
The measurement
- Families screened
- 20
- Schema
- canli.alphac-atlas-reachability-screen.v1
By verdict
HELD BUT BLOCKED ON EXTRACTION QUALITY
- active_ownership_escalation
- merger_arbitrage
- tender_offer_spread
HISTORY TOO SHORT FOR THE ADMISSION CONTRACT
- dealer_gamma_pressure
NO POINT IN TIME RECORD WAS EVER PRESERVED
- crypto_liquidation_pressure
OBTAINABLE FROM DATA THIS REPO ALREADY HOLDS
- inflation_breakeven_relative_value
OBTAINABLE PUBLICLY NEEDS A NEW PIPELINE
- credit_equity_relative_value
- municipal_taxable_basis
RECORD OBTAINABLE BUT THE MARKS ARE NOT EXECUTABLE
- catastrophe_bond_event_risk
- freight_derivative_dislocation
VENDOR ONLY AN OWNER SPENDING DECISION
- analyst_revision_drift
- carbon_allowance_carry
- cross_currency_basis
- fallen_angel_flow
- fx_option_risk_reversal
- index_reconstitution_flow
- mortgage_convexity_pressure
- securities_lending_supply
- sovereign_cds_fx_dislocation
- swap_spread_dislocation
Contract
- Minimum history years
- 3
- Minimum oos observations
- 756
- Trading days per year
- 252
Why it is decisive. A candidate's return series cannot be longer than the source that produces it. A source with less history than this cannot supply an admissible sleeve however good the idea, so the arithmetic is applied BEFORE any judgement about the idea.
Families
Nested beyond what this page renders. Read it in the artifact.
Headline. 6 of 20 untouched families are blocked on engineering. The other 14 are blocked on money, on a record that was never preserved, or on prices nobody could trade — none of which more work closes.
Ranking
- OBTAINABLE_FROM_DATA_THIS_REPO_ALREADY_HOLDS
- OBTAINABLE_PUBLICLY_NEEDS_A_NEW_PIPELINE
- HELD_BUT_BLOCKED_ON_EXTRACTION_QUALITY
- VENDOR_ONLY_AN_OWNER_SPENDING_DECISION
- HISTORY_TOO_SHORT_FOR_THE_ADMISSION_CONTRACT
- RECORD_OBTAINABLE_BUT_THE_MARKS_ARE_NOT_EXECUTABLE
- NO_POINT_IN_TIME_RECORD_WAS_EVER_PRESERVED
Sources held
Crypto perp funding multivenue
- First
- 2020
- Last
- 2026
- Measured from
- data/lake/funding/*/year=*
- Partitions
- 7
- Span years
- 6
Deribit option snapshots
- First
- 2026-06-25
- Last
- 2026-08-21
- Measured from
- data/deribit/snapshots/snap_*.jsonl
- Partitions
- 12
- Span years
- 0.160000
Edgar quarterly indexes
- First
- 2010-01
- Last
- 2025-10
- Measured from
- data/raw/sec_active_ownership_13d/indexes/*.idx.gz
- Partitions
- 64
- Span years
- 15.750000
Finra short interest
- First
- 2017-12-29
- Last
- 2026-07-15
- Measured from
- data/lake_shortint/settlement_date=*.parquet
- Partitions
- 206
- Span years
- 8.540000
Fred breakeven 10y
- First
- 2003-01-02
- Last
- 2026-08-20
- Measured from
- data/lake_macro_vintage/meta.json
- Partitions
- 5913
- Span years
- 23.630000
Fred breakeven 5y
- First
- 2003-01-02
- Last
- 2026-08-20
- Measured from
- data/lake_macro_vintage/meta.json
- Partitions
- 5913
- Span years
- 23.630000
Fred treasury 10y
- First
- 1962-01-02
- Last
- 2026-08-19
- Measured from
- data/lake_macro_vintage/meta.json
- Partitions
- 16143
- Span years
- 64.630000
Verdict meanings
HELD BUT BLOCKED ON EXTRACTION QUALITY. The documents are held and the shortfall is in reading them. C1's harness applies directly here: ask what a PERFECT detector would reach BEFORE writing a protocol.
HISTORY TOO SHORT FOR THE ADMISSION CONTRACT. The best source offers less history than the admission contract's minimum observations. The family cannot produce an admissible sleeve today no matter how good the idea is. It becomes screenable only by waiting, or by buying history.
NO POINT IN TIME RECORD WAS EVER PRESERVED. The as-of state was never archived — only the current or restated version survives. No amount of money or engineering recovers it. Forward collection from today is the only route, which means the clock starts now and the contract's minimum sets the wait.
OBTAINABLE FROM DATA THIS REPO ALREADY HOLDS. The record this identity needs is already in this repo's lake, with enough history for the contract. Nothing has to be bought and no new pipeline has to be built.
OBTAINABLE PUBLICLY NEEDS A NEW PIPELINE. The record exists and is public. The cost is engineering, which is the cheapest kind of blocker this programme has.
RECORD OBTAINABLE BUT THE MARKS ARE NOT EXECUTABLE. The prices are assessed, indicated or broker-quoted rather than transacted. A backtest on them measures a price nobody could have traded, which is a more expensive kind of wrong than a null: it produces a number, and the number is not about trading.
VENDOR ONLY AN OWNER SPENDING DECISION. The record exists and is licensed. This is not an engineering question and it is not mine to answer — it is a spending decision with a number attached.
What is judged. Who holds a record this repo does NOT hold, and whether an as-of version of it was preserved at all. Every such row is stamped JUDGEMENT_NOT_VERIFIED_THIS_RUN and carries the concrete check that would settle it. None is presented as measured.
What is measured. The family list, derived from the atlas rather than listed. The contract minimum, read from the contract in force. The history of every source this repo holds, re-derived from the lake on this run. The verdict, computed from those.
Check it yourself
Every figure above is read from /glassbox/atlas_reachability_screen.json, the
artifact the engine wrote. Nothing on this page is typed by hand: the page is generated from that
file, so a figure that moves in the artifact moves here and a figure that is not in the artifact
cannot appear here at all.