int_pit_loss_circuit

Empirical circuit-specific pit-lane loss prior, read by int_pit_strategy_value in place of the largely-imputed circuit_reference constant.

Part of the Strategy family.

Empirical circuit-specific pit-lane loss prior, read by int_pit_strategy_value in place of the largely-imputed circuit_reference constant. Per stop, loss = (in_lap − ref) + (out_lap − ref) where ref is the driver’s median green lap that race; the value is the median across stops within a sane [pit_loss_min_s, pit_loss_max_s] window, EB-shrunk toward the global median. Stops whose in-lap or out-lap ran under SC, VSC or a red flag are excluded — under a neutralisation the lap the stop is measured against is not the lap the driver would otherwise have run. Estimated per physical venue (circuit_id) so a renamed event or double-header shares one pit lane and one sample, then emitted per event slug so the consumer’s race_to_track join is unchanged. Scoped to races with a venue slug available.

Lineage

Overview

Property	Value
Layer	intermediate
Family	Strategy
Contract enforced	No
Upstream	stg_track_status, stg_results, dim_circuits, stg_laps, stg_pits

9 tests guard this model - 8 generic, 1 singular.

Columns

Column	Type	Nullable	Tests	Description
circuit_slug		Yes	not_null, unique	Event identifier (race name slug); recurs across seasons. PK.
circuit_id		Yes	not_null	Physical-venue identifier the estimate is pooled on. Several event slugs can share one circuit_id (british / 70th_anniversary, brazilian / são_paulo, mexican / mexico_city, austrian / styrian) and then carry identical values by construction.
n_stops		Yes	not_null	Number of clean green-flag pit stops observed at this physical venue.
pit_loss_s_empirical		Yes	not_null, expect_column_values_to_be_between	Median empirical pit-lane loss (seconds) across observed stops.
pit_loss_s_shrunk		Yes	not_null, expect_column_values_to_be_between	Pit-lane loss EB-shrunk toward the global median (pseudo-count pit_loss_prior_stops).