Applied Quants is currently testing a separate Alpaca Trading API client in PAPER, not a funded live account and not a LEAN deployment. We are interested in practical reconciliation patterns from the LEAN/Alpaca integration community, while keeping those implementation differences explicit.

We reviewed “Self.liquidate() does not work 5 minutes before market close”: https://www.quantconnect.com/forum/discussion/19276/self-liquidate-does-not-work-5-minutes-before-market-close/p1
The posted chronology submits a replacement exit before the previous stop's terminal cancellation update. In April 2026 the original author confirmed that sequential operations and careful cancel-status checks resolved it. We do not take the earlier automated response's proposed five-minute broker cutoff as verified policy.

Building on that resolution, is there a current minimal event/state example or integration test covering partial entry fills, cancellation of the unfilled remainder, and protective-exit resizing when a fill races cancellation? What broker evidence does it use before a replacement exit relies on the released closing-share reservation? We are looking for the actual reconciliation sequence rather than a fixed delay after requesting cancellation.

Current Alpaca documentation describes quote/NBBO-based simulated fills (https://docs.alpaca.markets/us/docs/paper-trading), and a staff forum response described additional PAPER latency near the open (https://forum.alpaca.markets/t/consistant-6-30am-6-35am-unfilled-marketable-paper-trade-orders/19416). We want to distinguish simulator limitations from client ownership/state defects. We have no live execution evidence yet.