06 · SYSTEM
Universe, sync & freshness
store connected · 12 runs in the ledger tail
snapshot generated 2026-10-27 07:42 ET
snapshot generated 2026-10-27 07:42 ET
WATCHLISTS · DATA UNIVERSE
These decide what this system keeps data for: bars, and the volatility, regimes, fits, chains, moments and betas derived from them. All of it is arithmetic once the bars are in hand, so a wide list is cheap — but not free, and a list nobody is watching should not cost a nightly fetch. The jobs cover the union of the lists marked COVERED; the rest are kept and cost nothing. Mining does not read any of this: it has its own, much shorter list below.
no lists yet — create one, or seed one from the imported book
0 names are covered across 0 lists. Per-ticker nightly cost appears here once a run has measured it. The universe is stored in plaintext because it drives public-data jobs; keeping it wider than the book is what stops it describing the book.
MULTI-DEVICE SYNC · ALWAYS ENCRYPTED
Everything that leaves this device is encrypted before it goes. There is no unencrypted mode to choose. The encryption key is generated here, from this browser’s random source: it is derived with PBKDF2-SHA256 at 600,000 iterations, used for AES-256-GCM, and never sent. Nothing keeps a copy — not this browser, not the server — so it cannot be recovered or reset. Put it in your password manager before you leave this panel. Local data on this device is unaffected either way.
or paste a generated key below
this device has never synced
encryption key · generated on this device · never sent to a server · store it in your password manager · it cannot be recovered or reset. Generated keys are 32 random bytes; a key entered from elsewhere must be base64 of the same form, carrying at least 24 bytes. A phrase is not accepted, because a phrase is the only kind of key the stored blob id can be attacked through.
the server stores ciphertext, an IV, a salt, the KDF parameters and a version. It has no key and no way to obtain one — a failed decrypt here means the key is wrong or the blob was altered, and nothing local is changed either way.
DATA FRESHNESS · PER DATASET
DATASET
SOURCE
AS OF
STATUS
daily OHLCV, watchlist
yfinance
2026-10-26
CURRENT
index strip
yfinance
2026-10-26
CURRENT
options chains / IV
yfinance
2026-10-26
CURRENT
regime model fits
nightly job
2026-10-26
CURRENT
volatility series
nightly job
2026-10-26
CURRENT
distribution fits
nightly job
2026-10-26
CURRENT
MVO moments
nightly job
2026-10-26
CURRENT
signal battery
nightly job
2026-10-24
1 SESSION BEHIND
macro series
FRED
2026-10-23
1 SESSION BEHIND
counted in completed sessions rather than hours, against a 21:15 UTC close — the same boundary the runner uses to decide today’s bar exists. Wall-clock age called a store holding Friday’s close stale all weekend, every weekend, which is the market being shut rather than a source failing. One session behind is left quiet because a holiday is indistinguishable from a late run; two is not. Numbers from a stale dataset carry this as-of date wherever they appear, not only here — source outages are routine rather than exceptional, and the interface treats them that way.
JOB LEDGER
11s/ticker · run 184 · idle — checking on focus and every 5 min
RUN
KIND
STATUS
STARTED
RAN
TICKERS
NOTES
184
nightly
SUCCEEDED
2026-10-26 18:24 ET
2m
12
183
chain_capture
SUCCEEDED
2026-10-26 16:26 ET
32s
7
session in progress — today's bar (2026-10-26) not stored
182
chain_capture
SUCCEEDED
2026-10-26 11:25 ET
34s
7
session in progress — today's bar (2026-10-26) not stored
181
nightly
SUCCEEDED
2026-10-24 10:54 ET
43s
12
unchanged, skipped: distribution_fits (9) · moments (1) · regimes (9) · volatility (9)
180
study_run
SUCCEEDED
2026-10-24 06:58 ET
2m
12
PG: rebased, full refetch (adj_close moved 0.6577% over 24 shared sessions, the oldest (2026-09-22) included (dividend or split)) · mined IWM, PG, SPY, TLT, XLE, XLU · pruned 129 superseded signal rows
179
nightly
SUCCEEDED
2026-10-23 18:24 ET
3m
12
178
chain_capture
SUCCEEDED
2026-10-23 15:08 ET
36s
7
session in progress — today's bar (2026-10-23) not stored
177
nightly
SUCCEEDED
2026-10-23 11:56 ET
42s
12
session in progress — today's bar (2026-10-23) not stored · unchanged, skipped: distribution_fits (9) · moments (1) · regimes (9) · volatility (9)
176
chain_capture
SUCCEEDED
2026-10-23 11:25 ET
30s
7
session in progress — today's bar (2026-10-23) not stored
175
nightly
SUCCEEDED
2026-10-22 18:24 ET
2m
12
174
chain_capture
SUCCEEDED
2026-10-22 15:11 ET
40s
7
session in progress — today's bar (2026-10-22) not stored
173
nightly
SUCCEEDED
2026-10-22 12:02 ET
52s
12
session in progress — today's bar (2026-10-22) not stored · unchanged, skipped: distribution_fits (9) · regimes (9) · volatility (9)
a dispatch queues a run and returns; it does not wait, and nothing on this page streams
stage timings are recorded per run, which is where the per-ticker cost figure above comes from. Per-ticker failures are recorded against the run rather than being absorbed by writing yesterday’s value under today’s date.