Skip to content

KB/signal

STLFSI4 Nowcast

Last verified

Auto-generated. This article is rebuilt from app/signals/config/signal_definitions.json by scripts/build_signals_kb.py. Edit the registry entry and re-run the script — do not edit this file directly.

STLFSI4 Nowcast

What it is

STLFSI4 Nowcast — registry key stlfsi4_nowcast.

Same-week FRESHNESS CARRIER (SIGPRED #88 W1) for the weekly STLFSI4 Friday print, which is up to 5 business days stale by Thursday. Each arriving Friday print confirms or denies what the nowcast said the prior business day (auto-grading ledger; prereg research/planning/prereg_stlfsi4_nowcast_2026-06-12.md). Not banded / scored / aligned, NOT a drawdown-risk leg (ofr_fsi_funding + nfci already carry financial conditions there — a third leg would double-count), and NEVER substituted into the disorderly_stress gate’s stlfsi4 read — the gate’s CORE AND leg stays the VALIDATED weekly print (coherence audit binding #88 constraint (a)); a regression test pins the gate identical with the nowcast present vs absent. The operator (and, only after forward grading, possibly the gate) reads it as a separate field.

Data-pyramid tier

T1 — Signal. This metric sits at the signal layer of the platform’s six-tier data pyramid (T0 raw inputs → T5 actionable read). The tier reflects how far the value is from a raw measurement — not how strongly it is validated. Abstraction and validation are separate axes: a higher tier is not a claim of stronger evidence.

Horizon & validation

Horizon: same-week — the window this metric’s measured skill speaks to. A horizon stamp is an evidence claim, not a trading signal.

Evidence: Same-week freshness carrier for the weekly STLFSI4 print (state freshening, not a forecast). Dossier SIGPRED_weekly_stress_2026-06-11.md W1: +0.45 OOS RMSE skill vs carry in EVERY era (0.287 -> 0.213), 71.8% sign accuracy, era skill +0.36..+0.60; the OFR-composite bridge alone carries +0.44 of the +0.45 - the single-observable bridge IS the nowcast (BACKTEST-GRADUATE grade; the deployed object is the frozen linear bridge 0.295, stress-week fit, graded forward per research/planning/prereg_stlfsi4_nowcast_2026-06-12.md; kill: forward RMSE skill <= 0 vs carry OR sign accuracy <= 50% over the first ~26 matured prints, ~6 months).

Source

Source module: computed
Data source: back_fillable

Fetched directly from app/sources/computed.py; see source code for the upstream API call and any provider-specific handling.

How it’s computed

Stale weekly print plus the frozen OFR-composite bridge times the OFR move since the print’s observation week-end: stlfsi4_nowcast = stlfsi4_stale + 0.295 * (ofr_fsi_current - ofr_fsi_anchor), where stlfsi4_stale is the row’s stored weekly Friday print, ofr_fsi_current is the row’s OFR FSI composite (T-2 lag — the carrier freshens the up-to-5-day-stale weekly print to T-2, never to today), and ofr_fsi_anchor is the OFR composite as of the print’s observation week-end (the daily_signals row with the LATEST ofr_fsi_obs_date <= stlfsi4_obs_date, non-NULL ofr_fsi). The bridge slope 0.295 is frozen (capture._STLFSI4_NOWCAST_BRIDGE): calibrated 2026-06-12 on the full public history (FRED STLFSI4 x OFR FSI composite, Friday-aligned weekly deltas, n=1378 — full-sample slope +0.2752 at corr +0.670; the frozen choice is the STRESS-WEEK fit |dSTLFSI4| > 0.05, n=990, slope +0.2950, because calm weeks carry ~zero slope and dilute; era-stable 2000-2026, worst era +0.2227, recent eras 0.33-0.37 — research/studies/w1_ofr_stlfsi4_bridge_calibration.py). Dossier evidence (research/findings/SIGPRED_weekly_stress_2026-06-11.md W1): +0.45 OOS RMSE skill vs carry in every era, 71.8% sign accuracy, OFR bridge alone = +0.44 of the +0.45 — the single-observable bridge IS the nowcast; the walk-forward model is the EVIDENCE, the frozen linear bridge is what ships. Abstains (NULL) unless ALL resolve: the stale print, stlfsi4_obs_date, ofr_fsi, and the anchor row — no fake fallbacks (a wrong anchor is a confidently wrong carrier). FRESHNESS CARRIER ONLY: never substituted into the disorderly_stress gate’s stlfsi4 read (binding #88 (a)). Forward grading: research/planning/prereg_stlfsi4_nowcast_2026-06-12.md.

Where it surfaces

Health-score / alignment role

Data carrier — no implication, no health-score contribution.

Persisted for downstream consumers (sparklines, base-rate matcher, calibration substrate) but does not classify into BULLISH / NEUTRAL / BEARISH and does not contribute to the 0-100 health score.

Release cadence

See also