Alpaca Fill Slippage Readiness

Run alpaca_fill_slippage_features_20260718T220412Z | Generated 2026-07-18T22:04:12.662884+00:00

Paper-only guardrail: this report is read-only, submits no orders, makes no provider calls, and does not enable live trading. Next action: No clean paper-fill candidates are available yet; refresh/review market-hours intraday proxy context for 1 eligible intent(s), then collect 98 additional linked, fresh-quote, tight-spread paper fills before slippage-sensitive training.

Readiness Summary

Feature rows3

Candidate samples2

Latest run rows3

Required samples100

Remaining samples98

Wide-spread review1

Training allowedfalse

Orders submitted0

Provider calls0

Live tradingfalse

Source views: paper.alpaca_fill_slippage_training_sample_inventory_operator_view and paper.alpaca_fill_slippage_readiness_operator_view.

Overall average adverse slippage: 14.4226 bps; average fill spread: 28.8452 bps.

Feature Status Groups

Feature Status	Training Sample Status	Rows	Symbols	First Fill	Latest Fill	Avg Adverse Slippage bps	Avg Spread bps
feature_ready	candidate_paper_slippage_training_sample	2	2	2026-06-30T17:54:59.393349+00:00	2026-06-30T17:55:00.615623+00:00	8.7239	17.4478
feature_ready_wide_spread_review	review_only_feature_ready_wide_spread_review	1	1	2026-06-30T17:54:59.087465+00:00	2026-06-30T17:54:59.087465+00:00	25.8200	51.6400

Slippage Training Sample Segmentation

This report-only segmentation separates clean tight-spread samples from wide-spread stress-review and other review-only evidence. Clean samples still cannot be used for slippage-sensitive stop-loss training until sample-depth and guardrail gates clear.

Segment	Status	Rows	Symbols	Required	Gap/Review	Avg Spread bps	Avg Adverse bps	Sample Symbols	Model Usage Policy	Operator Action
clean_tight_spread_candidate_samples	clean_sample_depth_blocked	2	2	100	98	17.4478	8.7239	STNG, TNK	clean_training_candidate_after_sample_depth_gate	Keep clean tight-spread fills separated and continue collecting until sample depth clears.
wide_spread_stress_review_samples	wide_spread_stress_review_required	1	1	0	1	51.6400	25.8200	ASA	stress_bucket_or_exclusion_review_only	Keep wide-spread fills out of the clean training sample; use only as stress/liquidity review evidence.
other_review_only_samples	no_other_review_only_samples	0	0	0	0	n/a	n/a		exclude_from_clean_training_until_reviewed	No non-wide review-only fill rows are visible in the latest slippage sample.
aggregate_slippage_training_sample_policy	training_blocked_segmented_evidence_available	3	3	100	99	28.8452	14.4226	STNG, TNK, ASA	report_only_no_slippage_sensitive_training_until_segments_clear	Do not train slippage-sensitive stop-loss models until clean sample depth clears and review/stress segments are excluded or explicitly modeled.

Readiness Gates

Gate	Status	Observed	Required	Operator Action
paper_fill_sample_depth	blocked	2.0000	100.0000	Keep slippage evidence collection running after paper fills; do not train slippage-sensitive stop models yet.
fill_quote_match_coverage	passed	3.0000	3.0000	Use matched quote context as paper evidence only; validate latency before live use.
nearest_trade_match_coverage	passed	3.0000	3.0000	Use trade matches to audit fill realism against prints before modeling.
wide_spread_review	review	1.0000	0.0000	Segment slippage by spread and liquidity before any stop-loss model uses these rows.
slippage_training_guardrail	blocked	2.0000	4.0000	Accumulate more paper fills and quote/trade matches before using slippage features in training.

Paper Execution Readiness

This is plan/readiness context only. It does not submit orders and does not authorize paper or live execution.

Run	Status	Account	Plan Only	Eligible Intents	Blocked Intents	Orders Submitted	Submission Enabled	Operator Action
alpaca_paper_execution_manager_20260717T195216Z	paper_plan_ready	ACTIVE	true	1	3	0	false	Risk-capped paper order plan is ready. Submit only after operator approval and guard switches.

Paper Fill Collection Control

Collection stageproxy_review_required_before_paper_collection

Collection allowed nowfalse

Possible when authorizedfalse

Clean candidates0

Proxy review intents1

Blocked reasons6

Suggested batch size0

Max batch size5

Gap after next batch98

Batches remaining20

Batches after next20

Market-hours refresh targets1

This section separates collection readiness from order authorization. A clean candidate means a row is suitable for a future bounded paper-fill collection run after explicit approval; it does not authorize an order from this report.

Batch policy: At most 5 explicitly authorized paper fills per batch, then reconcile fills and refresh slippage features before any next batch.

Block Reasons

sample_depth_gap_98
wide_spread_review_rows_1
slippage_training_guardrail_blocked
paper_order_submission_switch_disabled
no_clean_collection_candidate
proxy_review_intents_1

Authorization Requirements

Explicit operator approval in the current work session.
Paper-order submission guard switch enabled only for a bounded collection run.
Use only rows marked paper_fill_candidate_when_authorized with passed execution-proxy gates.
Rerun paper fill reconciliation and slippage feature refresh immediately after collection.
Keep live trading disabled.

Forbidden Actions

Do not treat this report as paper-order authorization.
Do not submit live orders.
Do not submit paper orders from review or blocked queue rows.
Do not train slippage-sensitive models on review-only or wide-spread samples.

Paper Fill Collection Diagnostics

This turns the queue into an operator checklist: why rows are not clean candidates, which rows need market-hours proxy refresh, and how large a future bounded paper-fill collection batch could be if explicitly authorized.

Diagnostic status: market_hours_proxy_refresh_required; batch action: No batch is clean now; refresh intraday monitor/proxy during market hours and recheck before any paper collection authorization..

Diagnostic	Count	Severity	Operator Action
clean_collection_candidates	0	blocked	No candidate rows can be used for paper fill collection right now.
market_hours_proxy_refresh_targets	1	review	Refresh intraday monitor/proxy during market hours before collecting paper fills.
wide_spread_review_rows	3	review	Segment or exclude wide-spread symbols before slippage-sensitive training.
stale_after_hours_rows	4	review	Treat after-hours proxy evidence as review-only until a fresh market-hours snapshot exists.
blocked_order_intents	1	blocked	Resolve execution risk reasons before using these rows for collection planning.
hold_rows_no_fill_needed	2	info	Do not manufacture fill evidence from hold rows.
paper_order_submission_switch	0	ok	Submission switch is disabled, which is the expected safe state.

Paper Fill Collection Manifest

This is the single report-only authorization packet for the next possible evidence-collection step. It never authorizes orders by itself; it records whether a future bounded paper-only batch could be requested after the listed preflight checks pass.

Manifest statusblocked_market_hours_proxy_refresh_required

Authorization statenot_authorized

Proposed batch size0

Max batch size5

Clean candidates0

Proxy refresh symbolsTNK

Remaining samples98

Orders submitted0

Manifest action: Run the market-hours proxy refresh wrapper for the listed symbols, then rebuild this manifest before asking for any paper-fill collection approval.

Preflight Checks

Check	Status	Observed	Required	Operator Action
sample_depth_gap	open	2	100	Need 98 more clean linked fills before training depth clears.
market_hours_proxy_refresh	required	1	0	Refresh proxy context for: TNK
clean_candidate_batch	not_available	0	1	Rebuild after proxy refresh or risk review creates clean candidates.
execution_guardrail	safe_plan_only_state	1	1	Plan-only, submission disabled, zero submitted orders.
current_report_authorization	not_authorized	0	1	This report cannot authorize paper orders; approval must happen explicitly in the current session.

Proposed Batch Rows

Symbol	Intent	Side	Risk	Proxy Gate	Monitor	Quote Age min	Spread bps	Action
No proposed paper-fill batch rows

Required Operator Actions

Keep live trading disabled.
Keep paper order submission disabled while the manifest status is blocked or review-only.
Run the market-hours proxy refresh wrapper first when proxy-refresh symbols are listed.
Rebuild fill/slippage features and live readiness after any proxy refresh.
Request explicit current-session approval before any bounded paper-fill collection batch.
After any future approved paper-fill batch, immediately run broker fill reconciliation and refresh slippage features.

Market-Hours Proxy Refresh Plan

This is a report-only runbook for refreshing local Alpaca bars/trades/quotes and recomputing execution proxy evidence during regular US equity market hours. It does not submit paper orders, does not enable live trading, and should be skipped outside market hours.

Market-hours onlytrue

Run-now policyrun_only_during_regular_us_equity_market_hours

Clock policydry_run_anytime_execute_only_regular_us_equity_market_hours

Target symbolsTNK

Target count1

Estimated provider requests3

Order step includedfalse

Operator action: During regular US equity market hours, run the bounded proxy refresh sequence for the target symbols, then recheck the collection queue before any paper-fill authorization.

Preferred Wrapper Commands

.\scripts\run-alpaca-market-hours-proxy-refresh.ps1 -Symbols 'TNK'
.\scripts\run-alpaca-market-hours-proxy-refresh.ps1 -Execute -Symbols 'TNK'

Command Sequence

.\scripts\run-alpaca-market-data-foundation.ps1 -Symbols 'TNK' -DataTypes 'bars,trades,quotes' -Feed 'sip' -Timeframe '1Min' -Limit 500 -MaxPages 1 -Sort desc -RunKind manual
.\scripts\run-alpaca-intraday-market-monitor.ps1 -Symbols 'TNK' -Feed 'sip' -LookbackMinutes 1440 -MaxStaleMinutes 30 -WideSpreadBps 50 -MonitorMode paper -RunKind manual
.\scripts\refresh-alpaca-intraday-execution-proxy-rollup.ps1
.\scripts\run-alpaca-fill-slippage-features.ps1 -RunKind manual
.\scripts\write-alpaca-live-trading-readiness.ps1
.\scripts\write-high-value-data-progress-review.ps1

Forbidden Commands

.\scripts\run-alpaca-paper-execution-manager.ps1 -SubmitOrders
Any live-trading command or broker order submission switch.

Paper Fill Collection Policy

This derived policy keeps the next collection decision separate from order authorization. It tells us whether sample depth, clean candidates, proxy quality, execution guardrails, and training use are blocked, review-only, or ready for a future explicitly authorized paper-only collection run.

Policy Area	Status	Observed	Required	Gap/Review	Ratio	Model Usage Policy	Operator Action
sample_depth_gate	sample_depth_blocked	2	100	98	0.0200	do_not_train_slippage_sensitive_models	Collect 98 more linked, fresh-quote, tight-spread paper fills before training review.
clean_collection_candidate_gate	proxy_review_required_before_collection	0	1	1	0.0000	no_order_submission_from_current_queue	Refresh/review collection queue until at least one clean candidate exists before asking for paper-fill collection approval.
execution_proxy_quality_gate	execution_proxy_review_required	8	0	8	n/a	proxy_context_only_not_fill_evidence	Refresh market-hours proxy rows and segment wide-spread/stale evidence before collection.
paper_execution_guardrail	safe_plan_only_state	1	1	0	1.0000	plan_only_no_orders_from_report	Expected safe state: plan-only, submission switch disabled, zero orders submitted.
paper_collection_authorization_gate	paper_collection_not_authorized	0	1	1	0.0000	no_paper_order_submission	Do not submit paper orders; this report is collection planning only.
slippage_training_use_gate	training_blocked	0	1	1	0.0000	do_not_use_for_slippage_sensitive_training	Keep slippage-sensitive training blocked while guardrail status is blocked.
blocked_intent_review_gate	blocked_intents_review_required	1	0	1	n/a	exclude_blocked_intents_from_collection	Do not use blocked intents for collection planning until risk reasons are resolved.
aggregate_paper_fill_collection_policy	paper_fill_collection_research_only_blocked	1	7	6	0.1429	paper_only_research_until_collection_and_training_gates_clear	Use this report to plan the next safe evidence step; it does not authorize orders or training.

Paper Fill Collection Queue

Plan-only order intents joined to local intraday proxy evidence. Rows marked as candidates still require operator approval and guard switches before any paper order submission.

Symbol	Intent	Side	Risk	Reason	Collection Status	Proxy Gate	Monitor	Quote Age min	Spread bps	Action
TNK	exit	sell	eligible	not_in_latest_target_list	review_proxy_before_paper_fill_collection	review	stale_after_hours	1533.6339	97.4578	Plan-only paper intent is eligible, but proxy evidence is not passed; refresh/review intraday context before paper fill collection.
PTC	entry	buy	blocked	max_positions_reached	blocked_paper_intent	review	stale_after_hours	1498.8486	39.7899	Intent is not eligible in the latest plan; resolve the risk reason before using it for fill collection.
ASA	hold	n/a	blocked	already_held_target	no_fill_collection_needed	review	stale_after_hours	1559.4884	230.4527	Latest plan says no order is required for this symbol; do not create fill evidence from a hold row.
STNG	hold	n/a	blocked	already_held_target	no_fill_collection_needed	review	stale_after_hours	1546.8222	68.8312	Latest plan says no order is required for this symbol; do not create fill evidence from a hold row.

Stop-Loss Sample Design Targets

Use these case buckets to shape future paper-fill collection after operator approval. They are labels for balanced evidence collection, not trading instructions.

Priority	Case Dimension	Case Value	Training Role	Gate	Current Symbols	Min Rank	Example Symbols	Recommended Stop	Operator Action
high	value_bucket	value_below_55	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	1596	53	TRMB #53, LSCC #104, AGX #111, KEYS #144, FIX #147, UI #157, CALX #158, ULTA #160	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
high	quality_bucket	quality_below_55	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	1420	135	UHS #135, ULTA #160, RMT #187, FDX #189, ELV #253, UNH #271, EXLS #273, CWEN #293	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
high	dominant_positive_signal_family	dominant_positive_family_return_prediction	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	1068	53	TRMB #53, ULTA #160, BKE #267, LDOS #300, EME #312, DPZ #333, GNRC #339, EYE #390	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
medium	dominant_positive_signal_family	dominant_positive_family_quality	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	288	29	SII #29, GOOGL #90, LRCX #92, GOOG #94, SCCO #100, LSCC #104, MNST #110, AGX #111	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
medium	dominant_negative_signal_family	dominant_negative_family_value	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	583	645	TRP #645, CM #708, STN #723, TSLA #733, GLBE #802, ROL #818, CCJ #832, VIAV #880	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
low	confidence_bucket	confidence_70_plus	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	9	775	EA #775, NXP #828, RZC #1066, SLAB #1287, NATH #1349, TALK #1917, AGNCP #2159, SOJE #2176	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
low	confidence_bucket	confidence_below_55	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	19	282	JHG #282, KALV #662, STEL #1251, BLD #1341, CWAN #1432, NREF #1530, LC #1556, MASI #1578	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
low	risk_bucket	risk_elevated_55_plus	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	71	292	HLNE #292, PTGX #433, WVE #591, IBRX #612, VRDN #639, MDGL #781, TRIN #900, ONDS #994	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
low	dominant_positive_signal_family	dominant_positive_family_event	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	58	891	WEC #891, BBY #988, NTRA #1000, CASS #1046, NET #1106, EPAM #1124, TD #1139, BLKB #1164	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
low	dominant_positive_signal_family	dominant_positive_family_no_positive_signal	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	31	1840	BK #1840, CMSD #1906, SAFE #1934, PRA #1969, FFIC #1989, RZB #2017, YEXT #2065, FNKO #2088	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
low	dominant_negative_signal_family	dominant_negative_family_quality	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	17	597	MGA #597, DB #855, GAM #1115, PVH #1297, GPI #1463, VET #1545, ALIT #1605, UNFI #1626	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.
low	market_cap_bucket	market_cap_unknown	stop_selector_coverage_gap_backfill	review_only_optimizer_date_gate	15	282	JHG #282, KALV #662, STEL #1251, BLD #1341, CWAN #1432, LC #1556, MASI #1578, TBRG #1604	Collect optimizer evidence before selecting a stop for this case	Do not train or promote a stop selector from this row yet; collect more distinct point-in-time signal dates.

Sample Feature Rows

Symbol	Side	Filled At	Quantity	Avg Price	Fill Spread bps	Quote Age sec	Nearest Trade Age sec	Adverse Slippage bps	Feature Status	Training Status	Operator Action
STNG	buy	2026-06-30T17:55:00.615623+00:00	14.4591	69.1600	8.6793	0.6084	2.0078	4.3396	feature_ready	candidate_paper_slippage_training_sample	Candidate paper-only slippage training sample; keep live use disabled.
TNK	buy	2026-06-30T17:54:59.393349+00:00	15.4010	64.9300	26.2164	0.6115	2.2720	13.1082	feature_ready	candidate_paper_slippage_training_sample	Candidate paper-only slippage training sample; keep live use disabled.
ASA	buy	2026-06-30T17:54:59.087465+00:00	19.0765	52.4200	51.6400	0.9196	1.6357	25.8200	feature_ready_wide_spread_review	review_only_feature_ready_wide_spread_review	Wide-spread paper fill; segment by liquidity/spread before model training.

Sample Quality Rules

Use linked Alpaca paper order intents only.
Require fresh fill-time quote context inside the configured max quote age.
Require nearby trade-print context for realism checks.
Keep wide-spread fills segmented or review-only until liquidity controls are validated.
Balance future paper-fill collection across stop-loss case buckets and liquidity regimes.
Use paper execution readiness as a preflight only; eligible intents are not order authorization.
Keep all outputs paper-only; this plan does not authorize order submission.