UNDERWRITERWhere the run hit friction.
Distinct decision cycles with a recorded live-strategy service failure.
Select a bar to open that period's failure records. A failure in one stage does not prove that an entire cycle stopped; these counts are historical evidence, not an uptime measurement.
Service failure ledger.
One primary category per live failed record. Category links open related evidence carrying that reason.
Why candidates stopped.
Live strategy reason mentions. Service failures are reported separately above.
A rejected record may carry several reasons. These are stage observations, not counts of unique trade opportunities.
Every stage leaves a trace.
Recorded events across the decision path, including the exploratory track.
| Stage | Events | Accepted | Rejected | Service failures |
|---|
Stage events are not a conversion funnel: a cycle can record several symbols and rejections, and an accepted stage event does not mean a trade occurred.
A fair view of the previous period.
| Recorded measure | Selected | Previous | Change |
|---|
Intent to execution.
Orders are selected by intent date; fills by execution date. Both use Eastern time.
Order status reflects its latest journal state. A fill can land outside its order's intent period. Spread units are never option contracts.
The controls behind the book.
Reconciliation passes and global regime verdicts in the selected period.
A completed check is evidence at its recorded moment. Regime blocks can name several reasons; successful reconciliation does not imply a trade was placed.
Three series. Clear provenance.
Account equity and the latest P&L snapshot for each source, reported independently.
Equity change includes account adjustments and is not a strategy return. Realised and unrealised P&L are the latest recorded snapshot values; daily snapshots are never added together. An absent measurement stays unknown.
A receipt for every readout.
Full selected-period counts, beside the complete journal inventory.
| Journal evidence | Selected records | All-time records | First selected observation | Last selected observation |
|---|
No recent-row cap is used for these totals. Calendar gaps represent no observations, not a measured zero balance or a confirmed outage.
How to read this report.
The counting rules behind the headline.
Decision cycles. A cycle is a distinct ID present in decision records. The cycle count includes exploratory records; operational impact counts only live-strategy rejected records.
Recorded days. Days with decision records in America/New_York. This is a data-coverage measure, not a trading calendar or a measure of uptime.
Operational classification. Known scan, preflight, veto and backend failure codes, plus the historical “Model unavailable” detail marker. Read the original records to investigate a category.
Time comparisons. The immediately preceding window has the same number of calendar days. Holidays and missing observations are not normalised; coverage is shown beside counts.
Historical context. A period containing old failures is not a claim that the service is failing now. The overview carries the current operating picture.
Complete counts, legible charts. Charts use daily, weekly or monthly buckets for long selections. Aggregation preserves every selected record and leaves missing money measurements blank.