Continuous monitors evaluate selected filters about every five minutes, starting after the first five-minute bar and a 15-second provider allowance. A stock is added as Matching filters only after passing every selected criterion and the data-quality checks in an evaluation. The first measured failure is shown as First failed check; two consecutive measured failures remove the stock and its history from the monitor. Unknown or stale data is shown as Data unavailable and interrupts the failure streak without counting as a failure.

Scans run only during the regular NYSE session, respecting holidays, daylight saving and early closes. Monitors expire at that day's close and never restart automatically. A paused monitor cannot be resumed in this release; create a new review from the screener. Missed intervals are not replayed, and long scans do not overlap. Timings depend on server availability and provider speed.

Each scan supports up to 500 stocks matching the market search, before calculated checks, processed in batches of two to limit background memory pressure. Broader searches require tighter filters; they are not silently truncated. Selected daily filters are checked first; a failed or unavailable daily check prevents intraday analysis. The one-minute reading checks selected intraday EMA, VWAP and RVOL plus freshness before any remaining timeframe analysis. Directional scoring stops if the selected score cannot be reached even with the remaining readings; with no score filter, three usable readings satisfy coverage. Some data must be fetched to establish whether calculated filters match; a rejected stock receives no further analysis that scan.

The saved screener settings are fixed when you create a review; changing the screener later does not update it. New stocks that fail any selected criterion or lack required evidence are never added or stored as potential candidates. All retained candidates shows actual matches plus previously matching stocks awaiting the second measured failure or fresh data. Removed stocks can rejoin only after passing every check in a later scan. Qualification is filter consistency, not relative ranking or profitability; inspect observation times because stored qualification is not a live guarantee.

Reduced intraday coverage and developing bars remain provisional. Daily bar dates are disclosed, but daily freshness/completed-bar structure needs manual review. Spread/depth, verified catalysts, borrow, entry/stop/target and costs remain unverified. Intraday direction is not a swing score. Existing analysis and Assess Trade Setup provide additional context, not a trading verdict.

Reviews and history exist only in this backend process's memory for today's session. They are discarded at the close or backend restart/redeployment; no candidate database or persistent-volume configuration is needed. Monitoring with the browser closed requires a running backend. Use one backend worker/instance for consistent access; session state is not shared across instances. This browser's private cookie identifies its reviews, not an account; clearing cookies loses access. Automatic candidates never change your personal watchlist; use Pin explicitly. At most 3 pending/active monitors and 10 session reviews per browser; 20 active monitors and 100 total session reviews per backend process.

The backend retains at most 5,000 candidate statuses and 5,000 history entries across all reviews, with up to 100 evaluations per stock. Oldest history is discarded first and trimming is disclosed without removing latest statuses. A scan that would exceed candidate capacity pauses its review with an explicit error; delete older reviews to free capacity. Connection failures retain displayed evidence as potentially outdated and refresh automatically; stalled requests time out after 10 seconds.