Open observability, closed execution

Understand the machine.

Kaelum exposes aggregate telemetry, market coverage, archive scale, system flow, and rounded operating burn while private execution remains closed.

updatinggenerated Jul 11, 13:48 UTC · refreshes every 30s143d runtime4/6 markets
Market data analyzed623K
bucketed public row scale
Source items scanned28K
451 today
Archive records3,335
0 today
System decisions2,563
0 today
Scanned markets

Market coverage

Scanned means data and analysis coverage — the loop reads these markets continuously during trading hours. Each heat cell below is one of the last 24 UTC hours, filled when that hour produced public market rows. Not a recommendation, and no private decision is implied.

ESS&P 500 futures
monitored · market closed
chart loading…
price unknown
24h data coverage
coverage loading…
NQNasdaq 100 futures
monitored · market closed
chart loading…
price unknown
24h data coverage
coverage loading…
GCGold futures
monitored · market closed
chart loading…
price unknown
24h data coverage
coverage loading…
CLCrude oil futures
monitored · market closed
chart loading…
price unknown
24h data coverage
coverage loading…
BTCBitcoin futures
stale · stale
chart loading…
price unknown
24h data coverage
coverage loading…

XAUUSDLEGACYGold spotretired feed · no live window

Operating loop

System flow

High-level pipeline stages, shown without private execution mechanics.

  1. Read

    Fetch public market and source context on schedule.

  2. Normalize

    Clean, align, group, and convert inputs into reusable context.

  3. Decide

    Record a measured system decision or stand-aside record.

  4. Archive

    Persist tuples for review, calibration, and audits.

  5. Review

    Evaluate outcomes, drift, failure modes, and operating burn.

Cost window · since Feb 2026

Operational burn

What it costs to run — all of it. Subscriptions, model usage, data feeds, and infrastructure, aggregated by public category and rounded. Months are gaps when nothing is recorded, never zeros.

Loading burn history…

Transparency boundary

Rules of the window

Research-only diagnostics must carry sample size and window context. Percentages stay conservative unless uncertainty is visible. Freshness failures become unknown, never a false all-clear.

Public

  • aggregate telemetry
  • market-data coverage
  • pipeline stages
  • archive scale
  • freshness and health status
  • rounded operating burn by category

Closed

  • private instruction logs
  • private source full text
  • secrets and keys
  • private execution rules
  • private risk geometry
  • private operator controls