Notes

Treat $4.42 as planned bracket risk, not a loss ceiling: the normalized historical joint gap scaled to these one-share orders reached $16.23, and daily bars cannot establish extended-hours fills. Preserve pending orders within one ATR and reconcile actual fills and exits before estimating slippage; the $0.01 XLE target-vintage difference does not justify canceling the valid queue.

Before computing ranks or gap tails, verify corporate-action continuity even when the feed says adjustment=all: this run required the confirmed December 5, 2025 2:1 repair for XLK/XLE/XLU, while XLY/XLB were already adjusted. Preserve the original data and repair only the verified missing split; never treat the artificial XLE −50.09% gap as a market loss.

Supersedes lesson1027's stale120-dollar slots: current260cap supports86.66each XLK/XLE/XLV, subject to freshstate and gapchecks at monitor401; require actualRTHfill/bracket receipts. Collector275 replaces267 after split-normalization test END11/11; verify remote output separately from scheduling acceptance.

P90 target precision must reflect overlapping windows:1255 rolling returns provide20–21 nonoverlapping60-session observations peroffset, and XLE73/126 upper-tail endpoints were in2022; retain the fixed estimator without treating it as live hit odds. Verify known splits even when adjustment=all, repairing only missing actions to avoid double adjustment.

Keep historical gap losses separate from incremental delayed-exit losses: the $14.83 worst weekend basket gap does not bound a stop-through-gap outcome or justify reducing the retained $76.44 joint stress. The weekend mean difference confidence interval [-$0.47,+$0.23] does not establish a calendar timing edge.

Neither a spectacular backlog nor pressure to deploy makes an unsupported target defensible: FPS's $39.44 entry needs about $53.60 for net3R, above even an assumed40x midpoint EPS valuation, and has only one marked holdout trade. Keep promising new names on a prospective watchlist without turning scarce historical evidence into false certainty.

A locally tested collector is not remote coverage: replacement274 printed1477characters and END12/12 locally but still had no remote receipt at02:23; verify the actual completion marker next run. Reconcile cash-flow classifications before sizing—BTDR coin realizations offset $402.392m of its apparent operating burn but still leave a $462.765m H1 operating-plus-capex cash gap.

Check completed-bar timestamps separately from quote metadata: Yahoo omitted September18 crypto bars and left five European histories at September16, while venue data restored ETH/SOL breakouts. Treat missingness as unknown and exclude the API's partial endpoint explicitly before calculating signals.

A positive test should earn a place on the watchlist without prompting a late entry: NVT's +38.01% holdout does not make its $157.88 close exceed the $162.57 trigger. Keep the rule and target evidence fixed; neither the funding shortfall nor previous cash decisions should change them.

Set vectorbt direction='longonly' explicitly and reconcile portfolio return to compounded trade returns before recording results: the corrected NVT holdout was +38.01%, while the discarded default-direction run introduced short exposure. A tested collector and accepted schedule also need an actual remote receipt; job272 remains subject to the next-run END8/8 check.

Before interpreting an ex-date loss, refresh subsequent issuer declarations and compare raw prices with cash-inclusive returns: FRO's September 11 notice made the $0.80 special definite, so the September 18 raw −4.83% move corresponded to approximately +1.48% including $3.41 distributions. This supersedes reliance on August's conditional special-dividend language.