Plain-language map of heavy jobs: what they do, where to click, and whether you start them, the API does, or a server timer does.
Copy is edited in webShares/admin_automation_catalog.py.
| Control | Where | How triggered | Notes |
|---|---|---|---|
| Strategy experiments (rank & backtest candidates) | Strategies tab → workbench |
Manual
API
Started from the UI, or by scripts posting to the API. |
Runs deterministic backtests on the ranked shortlist for a chosen day—different weights, filters, or horizons—and stores each outcome as a recorded run. Endpoints: /api/experiments/run/, /api/experiments/design/, /api/experiments/recent/. No cron job in this repo starts these automatically. |
| Model Center automation (per strategy lane) | Strategies tab → Model Center automation |
Scheduled
API
A repeating server job and/or on-demand from the UI or API. |
Here a lane means one strategy line (Trade, Growth, or Yield) plus its timeline—the horizon you rank names against. When automation is enabled for that lane, the host should run `manage.py run_model_center_automation` on a timer. You can also fire a run manually with POST /api/model-center/automation/trigger/; status at /api/model-center/automation/status/. |
| Strategy optimiser (tune each lane) | Strategies tab → Regression Model Configuration |
Manual
API
Saved schedule
From the UI or API; clock settings are reminders unless a host calls the trigger. |
Searches for better scoring-model settings for the lane’s regression / ranking setup and saves the winner for promotion. Hosts should run `manage.py run_strategy_optimization_schedule` on a timer (see scripts/scRunStrategyOptimisationTick.sh and scripts/crontab-current.conf); that command processes due schedules and drains the optimisation queue. Operators can also POST /api/model-center/optimization/trigger/ to enqueue a run for the next tick. |
| After-hours batch (prices, analysis, mail) | This page → Daily close pipeline |
Shortcut
Same as cron
Same workload as the nightly cron; this tile reruns it on demand. |
Downloads prices, refreshes analysis, updates watchlists, and sends summary mail for the post-close routine. Host wrappers and recovery flows are described in scripts/README.md (for example scStockBatchRunClose.sh, scRecoverMissedCloseBatch.sh). |
| Live candidate refresh | This page → Daily close pipeline |
Shortcut
After rankings
Normally follows the ranking step after close; use this to regenerate sooner. |
Rebuilds the saved candidate rows the workspace reads from. scripts/README.md explains generate_live_candidates and how it relates to the overnight batch. |
| Trading dry-run | This page → Command Execution |
Manual
Started manually from here unless you wire your own cron. |
Runs trading-batch checks without sending live orders—useful before trusting automation. |
| Platform regression tests | This page → Command Execution |
Manual
Started manually from here unless you wire your own cron. |
Runs the pytest suite through the legacy menu runner. |
Starting close_stage_daily_updates... close_stage_daily_updates finished.
Exit: n/a
System events, error summary, AI diagnosis, broker gateways, and MySQL status.
Loading...
Loading AI diagnosis...
Loading...
Running reconciliation...
Usually under a minute; may take up to 2 minutes when syncing many orders.