Why do Prophet + NeuralProphet + StatsForecast + Amazon Forecast + Vertex AI Forecast + Anaplan + Vena + Pigment break at multi-location-sub-month-cadence-cohort-relative scale?

Each forecasting vendor ships per-account flat point-forecast primitive at single-horizon level. None coordinates per-location per-cohort per-horizon calibration against Brier + ECE + MCE + reliability diagram + isotonic regression + Platt scaling + conformal prediction intervals + Bayesian posterior calibration. None handles cohort-relative trend-line decomposition (per-cohort intercept + per-cohort slope + per-cohort interaction) at sub-month cadence. None gates against replication-crisis statistical discipline (Bonferroni + BH FDR + E-value + Rosenbaum sensitivity + negative-control outcome + per-cohort placebo test). None enforces FDD Item 19 financial performance representations when forecast shared with franchisees + FINRA Rule 2210 when public-company forecast + SOX 302/404/906 when public-company forecasting material + FASB ASC 280 segment reporting. None writes a per-location per-cohort per-horizon WORM forecast-calibration audit trail. The four-skill bundle Calibrate + Decompose + Score + Audit sits above the warehouse + forecasting + Bayesian substrate — it does not replace it.

What does this skill connect to on the per-location-forecasting agent and across the swarm?

On the per-location-forecasting agent: per-location-cohort two-sigma anomaly detection (sibling) + per-cohort benchmarking + root-cause attribution sketch + peer-cohort computation. Across the swarm: per-location per-cohort two-sigma anomaly detection (same per-location-cohort substrate) + per-location demographic feeds (DOWNSTREAM consumer of per-cohort trend) + per-location metric ingestion (UPSTREAM source of per-location metric history) + integration-drift-monitor agent (#562 + #569 + #570) + governance-decision-router five-destination routing + tiered pre-filter deterministic gates for AI content + per-state-overlay-composer (#599 UPSTREAM canonical for FDD Item 19 + per-state attorney advertising + FINRA per-state). Commercial-pillar parent: /pre-emptive-churn-and-cohort-relative-trends.

What does the 6-workstream pre-engagement-baseline reporting cycle look like for this skill?

Every two weeks during the Tier 3 Fractional CMO with AI Swarm engagement, six workstreams report against the pre-engagement baseline. Workstream 1: per-portfolio per-location per-cohort per-horizon forecasting coverage — locations monitored + cohorts decomposed + horizons calibrated. Workstream 2: Calibrate per-location per-horizon calibration flow — Brier score + ECE + MCE + reliability diagram + isotonic + Platt + conformal + Bayesian metric absorbed. Workstream 3: Decompose cohort-relative trend-line flow — per-cohort intercept + per-cohort slope + per-cohort seasonality + per-cohort interaction effect decomposed. Workstream 4: Score per-horizon scoring-rule flow — logarithmic + Brier + spherical + RPS + CRPS + quantile + pinball + Brier-decomposition + Murphy diagram. Workstream 5: Regulatory-defense audit coverage — sub-month horizon + per-location calibration + cohort-relative decomposition + replication-crisis discipline + EU AI Act Article 50 + FDD Item 19 + FINRA 2210 + SOX + FASB ASC 280. Workstream 6: FBC feedback-loop pattern-learning — per-location per-cohort realized-vs-predicted calibration + per-horizon score retrospective + per-cohort decomposition retrospective.

Build pillar · per-location-forecasting agent

How to build per-location AI-calibrated forecasting at sub- month cadence for cohort-relative trend lines

Snowflake + BigQuery + Databricks + Redshift + ClickHouse + dbt + Dataform + Apache Airflow + Prefect + Dagster + Prophet + NeuralProphet + StatsForecast + Nixtla + Amazon Forecast + Vertex AI Forecast + TimeGPT + Lag-Llama + Chronos + Moirai + TimesFM + PyMC + NumPyro + Stan + GluonTS + DeepAR + N-BEATS + Temporal Fusion Transformer ship per- account flat forecasting primitives. The Calibrate + Decompose + Score + Audit skill bundle on the per-location- forecasting agent sits above the warehouse + forecasting + Bayesian substrate and writes a per-location per-cohort per-horizon canonical forecast-calibration record with named regulatory anchors covering sub-month forecast horizon (4d + 7d + 14d + 21d + 28d) + per-location calibration (Brier + ECE + MCE + reliability diagram + isotonic + Platt + conformal + Mondrian conformal + ACI + Bayesian posterior + BMA + WAIC + LOO-CV) + cohort- relative trend-line decomposition (STL + X-13ARIMA-SEATS + TBATS + per-cohort intercept/slope/interaction) + replication-crisis statistical discipline (Ioannidis 2005 + Amrhein-Greenland-McShane 2019 + Bonferroni + BH FDR + E-value + Rosenbaum Γ + Cornfield + falsification test + negative-control + per-cohort placebo + reverse causality) + EU AI Act Article 50 + FDD Item 19 + FINRA 2210 + SOX 302/404/906 + FASB ASC 280.

Published January 13, 2027 · 3,200 words

The 4-skill bundle on the per-location-forecasting agent

One agent. Four coordinated skills. The Calibrate + Decompose + Score + Audit bundle runs above the warehouse + forecasting + Bayesian substrate and writes one canonical per-location per-cohort per-horizon forecast-calibration record.

Calibrate

Per-location per-cohort per-horizon (4d + 7d + 14d + 21d + 28d) forecast generation from per-vendor forecasting backbone + per-location calibration (Brier + Expected Calibration Error + Maximum Calibration Error + reliability diagram + isotonic regression + Platt scaling + sigmoid + conformal prediction intervals + Mondrian conformal + Adaptive conformal inference + Bayesian posterior + Bayesian model averaging + WAIC + LOO-CV).

Decompose

Cohort-relative trend-line decomposition: STL decomposition + X-13ARIMA-SEATS + TBATS + Prophet additive + multiplicative seasonality + per-location- cohort trend isolation + per-cohort intercept + per- cohort slope + per-cohort seasonality + per-cohort interaction effects against portfolio average. Per- location per-cohort decomposition into trend + seasonal + residual components.

Score

Per-location per-cohort per-horizon scoring against proper scoring rules: logarithmic + Brier + spherical + ranked probability score + continuous ranked probability score + quantile loss + pinball loss + Brier-decomposition (reliability + resolution + uncertainty) + Murphy diagram. Per-horizon score severity P0-P4.

Audit

Per-location per-cohort per-horizon WORM record: forecast distribution snapshot + per-location calibration metric + cohort-decomposition + per- anchor gate-pass + AI-ML provenance + EU AI Act FRIA. Retention: 7-year FTC + 7-year IRS + 7-year HIPAA + 7-year state bar + 6-year SEC + 3-year FINRA + 7-year SOX + GDPR Article 30 + EU AI Act Article 12 + SOC 2 CC7/CC8.

The real ecosystem this sits above

Calibrate + Decompose + Score + Audit does not replace warehouses, forecasting libraries, or Bayesian samplers. It sits above them and writes one canonical per-location per- cohort per-horizon forecast-calibration record.

Warehouse + orchestration

Snowflake + BigQuery + Databricks + Redshift + ClickHouse
dbt + Dataform + SQLMesh + Apache Airflow + Prefect + Dagster
Iceberg + Hudi + Delta Lake time-travel
Apache Kafka + Confluent + AWS MSK + Azure Event Hubs
Apache Spark + Apache Flink + Apache Beam + Ray distributed

Forecasting + time-series LLM

Prophet + NeuralProphet + StatsForecast + sktime + Darts + Nixtla
Amazon Forecast + Vertex AI Forecast + Azure ML Forecast
Anaplan + Workday Adaptive + Vena + Pigment FP&A
TimeGPT + Lag-Llama + Chronos + Moirai + TimesFM
GluonTS + DeepAR + N-BEATS + Temporal Fusion Transformer

Bayesian + calibration + scoring

PyMC + NumPyro + Stan + brms Bayesian samplers
scikit-learn + XGBoost + LightGBM + CatBoost
Brier + ECE + MCE + reliability diagram
Isotonic regression + Platt scaling + conformal + Mondrian
Replication-crisis discipline + per-cohort placebo

Compliance overlay

Five anchors run per-location per-cohort per-horizon before any forecast distributes to downstream decision systems. The first anchor is operationally distinctive: sub-month forecast horizon + per-location calibration (Brier + ECE + isotonic + Platt + conformal + Bayesian) + cohort-relative decomposition + replication-crisis statistical discipline converge on every forecast.

Anchor 1: Sub-month horizon + per-location calibration + cohort-relative decomposition + replication-crisis discipline (operationally distinctive)

Sub-month forecast horizon (4d + 7d + 14d + 21d + 28d). Per-location calibration: Brier score + Expected Calibration Error (ECE) + Maximum Calibration Error (MCE) + reliability diagram + isotonic regression calibration + Platt scaling + sigmoid calibration + conformal prediction intervals + Mondrian conformal (per-location bin) + Adaptive conformal inference (ACI) + Bayesian posterior calibration + Bayesian model averaging (BMA) + WAIC + LOO-CV. Cohort-relative trend- line decomposition: STL + X-13ARIMA-SEATS + TBATS + Prophet additive + multiplicative seasonality + per- location-cohort trend isolation + per-cohort intercept + per-cohort slope + per-cohort seasonality + per- cohort interaction effects. Brier-decomposition (reliability + resolution + uncertainty) + Murphy diagram + proper scoring rules (logarithmic + Brier + spherical + RPS + CRPS + quantile + pinball). Replication-crisis statistical discipline: Ioannidis 2005 “Why Most Published Research Findings Are False” + Amrhein Greenland McShane 2019 “Retire statistical significance” + Benjamini-Hochberg FDR + Bonferroni correction + E- value (VanderWeele Ding 2017) + Rosenbaum sensitivity (Γ) + Cornfield inequality + falsification test + negative-control outcome + per-cohort placebo test + reverse causality test.

Anchor 2: FTC + FDD Item 19 + Lanham

FTC Section 5 + Pfizer 1972 + CFPB UDAAP + Lanham + USPTO + Robinson-Patman + FDD Item 19 financial performance representations when forecast shared with franchisees + 15-state franchise + per-state attorney advertising.

Anchor 3: HIPAA + FINRA + per-vertical

HIPAA 45 CFR 164.502/504/514 + state mini-HIPAA + FINRA Rule 2210 + Rule 3110 + SEC Regulation FD + per-state professional licensing.

Anchor 4: EU AI Act + AI-ML forecasting

EU AI Act Article 50 transparency when AI-generated forecast + Article 13/14/15 + Annex III when AI-ML forecasting drives capital/inventory decisions + Article 6/27 FRIA + DSA + DMA. GDPR Article 6/7/28/30 + LGPD + DPDP + PIPEDA + Quebec Law 25 + CCPA + CPRA + 18- state.

Anchor 5: Accessibility + SOX + FASB + WORM retention

WCAG 2.2 AA + ARIA + EAA + ADA Title III + Section 508. SOX 302/404/906 when public-company forecasting material + COSO + Exchange Act 13(b)(2) + FASB ASC 280 segment reporting + SEC Reg S-K. NIST AI RMF + ISO 42001 + ISO 27001 + SOC 2 Type II. Per-vendor LLM zero-retention + per-source DPA + per-API rate-limit. Storage: AWS S3 Object Lock + Azure Blob immutable + GCS + Wasabi WORM. Retention: 7-year FTC + 7-year IRS + 7-year HIPAA + 7-year state bar + 6-year SEC + 3- year FINRA + 7-year SOX + GDPR Article 30 + EU AI Act Article 12 + SOC 2 CC7/CC8.

6-workstream reporting cycle

Every two weeks during a Tier 3 Fractional CMO engagement, six workstreams report against the pre-engagement baseline. No forecast accuracy claims. Process commitments only.

1. Per-portfolio per-location per-cohort per- horizon forecasting coverage. Locations monitored + cohorts decomposed + horizons calibrated.
2. Calibrate per-location per-horizon flow. Brier + ECE + MCE + reliability diagram + isotonic + Platt + conformal + Bayesian metric absorbed.
3. Decompose cohort-relative trend-line flow. Per-cohort intercept + slope + seasonality + interaction effect decomposed.
4. Score per-horizon scoring-rule flow. Logarithmic + Brier + spherical + RPS + CRPS + quantile + pinball + Brier-decomposition + Murphy diagram.
5. Regulatory-defense audit coverage. Sub-month horizon + per-location calibration + cohort- relative decomposition + replication-crisis discipline + EU AI Act Article 50 + FDD Item 19 + FINRA 2210 + SOX + FASB ASC 280.
6. FBC feedback-loop pattern-learning. Per-location per-cohort realized-vs-predicted calibration + per-horizon score retrospective + per-cohort decomposition retrospective.

FAQ

What is per-location AI-calibrated forecasting at sub-month cadence — and what is the sub-month-horizon-times-per-location-calibration-times-cohort-relative-decomposition-times-replication-crisis-statistical-discipline problem distinctive to this skill?: A multi-location retail operator with 50-300 stores ships per-location forecasting at sub-month cadence (4-day + 7-day + 14-day + 21-day + 28-day forecast horizons) for cohort-relative trend lines (per-cohort intercept + per-cohort slope + per-cohort interaction effects against portfolio average). Standard FP&A platforms (Anaplan + Workday Adaptive + Vena + Pigment) report monthly or quarterly. Standard forecasting libraries (Prophet + NeuralProphet + StatsForecast + sktime + Darts + Nixtla + Amazon Forecast + Vertex AI Forecast) ship point forecasts without per-location calibration. The four-skill bundle on the per-location-forecasting agent — Calibrate, Decompose, Score, Audit — sits above the warehouse + forecasting + Bayesian substrate (Snowflake + BigQuery + Databricks + Redshift + ClickHouse + dbt + Apache Airflow + Prefect + Dagster + Prophet + NeuralProphet + StatsForecast + Nixtla + TimeGPT + Lag-Llama + Chronos + Moirai + TimesFM + PyMC + NumPyro + Stan + brms + GluonTS + DeepAR + N-BEATS + TFT) and writes a per-location per-cohort per-horizon canonical forecast-calibration record. The operationally distinctive anchor: sub-month forecast horizon (4-day + 7-day + 14-day + 21-day + 28-day) + per-location calibration (Brier score + Expected Calibration Error (ECE) + Maximum Calibration Error (MCE) + reliability diagram + isotonic regression calibration + Platt scaling + sigmoid calibration + conformal prediction intervals + Mondrian conformal + Adaptive conformal inference (ACI) + Bayesian posterior calibration + Bayesian model averaging (BMA) + WAIC + LOO-CV) + cohort-relative trend-line decomposition (STL decomposition + X-13ARIMA-SEATS + TBATS + Prophet additive + multiplicative seasonality + per-location-cohort trend isolation) + replication-crisis statistical discipline (Ioannidis 2005 + Amrhein Greenland McShane 2019 + Benjamini-Hochberg FDR + Bonferroni + E-value VanderWeele Ding 2017 + Rosenbaum sensitivity Γ + Cornfield inequality + falsification test + negative-control outcome + per-cohort placebo test + reverse causality test).
Why do Prophet + NeuralProphet + StatsForecast + Amazon Forecast + Vertex AI Forecast + Anaplan + Vena + Pigment break at multi-location-sub-month-cadence-cohort-relative scale?: Each forecasting vendor ships per-account flat point-forecast primitive at single-horizon level. None coordinates per-location per-cohort per-horizon calibration against Brier + ECE + MCE + reliability diagram + isotonic regression + Platt scaling + conformal prediction intervals + Bayesian posterior calibration. None handles cohort-relative trend-line decomposition (per-cohort intercept + per-cohort slope + per-cohort interaction) at sub-month cadence. None gates against replication-crisis statistical discipline (Bonferroni + BH FDR + E-value + Rosenbaum sensitivity + negative-control outcome + per-cohort placebo test). None enforces FDD Item 19 financial performance representations when forecast shared with franchisees + FINRA Rule 2210 when public-company forecast + SOX 302/404/906 when public-company forecasting material + FASB ASC 280 segment reporting. None writes a per-location per-cohort per-horizon WORM forecast-calibration audit trail. The four-skill bundle Calibrate + Decompose + Score + Audit sits above the warehouse + forecasting + Bayesian substrate — it does not replace it.
How does Calibrate + Decompose work?: Calibrate runs per-location per-cohort per-horizon (4-day + 7-day + 14-day + 21-day + 28-day) forecast generation from per-vendor forecasting backbone (Prophet + NeuralProphet + StatsForecast + Nixtla + Amazon Forecast + Vertex AI Forecast + TimeGPT + Lag-Llama + Chronos + Moirai + TimesFM + PyMC + NumPyro + Stan + GluonTS + DeepAR + N-BEATS + TFT) then runs per-location calibration: Brier score + Expected Calibration Error (ECE) + Maximum Calibration Error (MCE) + reliability diagram + isotonic regression calibration + Platt scaling + sigmoid calibration + conformal prediction intervals + Mondrian conformal (per-location bin) + Adaptive conformal inference (ACI) + Bayesian posterior calibration + Bayesian model averaging (BMA) + WAIC + LOO-CV. Per-location per-horizon calibrated forecast distribution. Decompose runs cohort-relative trend-line decomposition: STL decomposition + X-13ARIMA-SEATS + TBATS + Prophet additive + multiplicative seasonality + per-location-cohort trend isolation + per-cohort intercept + per-cohort slope + per-cohort seasonality + per-cohort interaction effects against portfolio average. Per-location per-cohort decomposition into trend + seasonal + residual components.
What does Score + Audit do?: Score runs per-location per-cohort per-horizon scoring against proper scoring rules: logarithmic score + Brier score + spherical score + ranked probability score (RPS) + continuous ranked probability score (CRPS) + quantile loss + pinball loss + Brier-decomposition (reliability + resolution + uncertainty) + Murphy diagram. Per-location per-cohort per-horizon score evolution against historical baseline. Per-horizon score severity classification: P0 calibration failure (ECE > 0.10 over 4-week window) + P1 reliability-diagram drift 72-hour + P2 conformal-interval coverage drift 7-day + P3 cohort-decomposition drift 30-day + P4 docs-only. Gate runs 5 anchors per-location per-cohort per-horizon before any forecast distributes to downstream decision systems. (1) Sub-month horizon + per-location calibration (Brier + ECE + MCE + reliability + isotonic + Platt + conformal + Bayesian) + cohort-relative decomposition + replication-crisis discipline (Ioannidis + Amrhein-Greenland-McShane + Bonferroni + BH FDR + E-value + Rosenbaum Γ + Cornfield + falsification + negative-control + per-cohort placebo + reverse causality). (2) FTC Section 5 + Pfizer 1972 + CFPB UDAAP + Lanham + USPTO + Robinson-Patman + FDD Item 19 financial performance representations when forecast shared with franchisees + 15-state franchise + per-state attorney advertising. (3) HIPAA + state mini-HIPAA + FINRA Rule 2210 + Rule 3110 + SEC Regulation FD + per-state professional licensing. (4) EU AI Act Article 50 transparency when AI-generated forecast + Article 13/14/15 + Annex III when AI-ML forecasting drives capital/inventory decisions + Article 6/27 FRIA + DSA + DMA + GDPR Article 6/7/28/30 + LGPD + DPDP + PIPEDA + Quebec Law 25 + CCPA + CPRA + 18-state. (5) WCAG 2.2 AA + ARIA + EAA + ADA Title III + Section 508 + SOX 302/404/906 + COSO + Exchange Act 13(b)(2) + FASB ASC 280 segment reporting + SEC Reg S-K. Audit writes a per-location per-cohort per-horizon WORM record: forecast distribution snapshot + per-location calibration metric + cohort-decomposition + per-anchor gate-pass + AI-ML provenance + EU AI Act FRIA. Retention: 7-year FTC + 7-year IRS + 7-year HIPAA + 7-year state bar + 6-year SEC + 3-year FINRA + 7-year SOX + GDPR Article 30 + EU AI Act Article 12 + SOC 2 CC7/CC8.
What does this skill connect to on the per-location-forecasting agent and across the swarm?: On the per-location-forecasting agent: per-location-cohort two-sigma anomaly detection (sibling) + per-cohort benchmarking + root-cause attribution sketch + peer-cohort computation. Across the swarm: per-location per-cohort two-sigma anomaly detection (same per-location-cohort substrate) + per-location demographic feeds (DOWNSTREAM consumer of per-cohort trend) + per-location metric ingestion (UPSTREAM source of per-location metric history) + integration-drift-monitor agent (#562 + #569 + #570) + governance-decision-router five-destination routing + tiered pre-filter deterministic gates for AI content + per-state-overlay-composer (#599 UPSTREAM canonical for FDD Item 19 + per-state attorney advertising + FINRA per-state). Commercial-pillar parent: /pre-emptive-churn-and-cohort-relative-trends.
What does the 6-workstream pre-engagement-baseline reporting cycle look like for this skill?: Every two weeks during the Tier 3 Fractional CMO with AI Swarm engagement, six workstreams report against the pre-engagement baseline. Workstream 1: per-portfolio per-location per-cohort per-horizon forecasting coverage — locations monitored + cohorts decomposed + horizons calibrated. Workstream 2: Calibrate per-location per-horizon calibration flow — Brier score + ECE + MCE + reliability diagram + isotonic + Platt + conformal + Bayesian metric absorbed. Workstream 3: Decompose cohort-relative trend-line flow — per-cohort intercept + per-cohort slope + per-cohort seasonality + per-cohort interaction effect decomposed. Workstream 4: Score per-horizon scoring-rule flow — logarithmic + Brier + spherical + RPS + CRPS + quantile + pinball + Brier-decomposition + Murphy diagram. Workstream 5: Regulatory-defense audit coverage — sub-month horizon + per-location calibration + cohort-relative decomposition + replication-crisis discipline + EU AI Act Article 50 + FDD Item 19 + FINRA 2210 + SOX + FASB ASC 280. Workstream 6: FBC feedback-loop pattern-learning — per-location per-cohort realized-vs-predicted calibration + per-horizon score retrospective + per-cohort decomposition retrospective.

Engage Completions

Two ways to engage. The Tier 1 AI Readiness Assessment maps the warehouse + forecasting + Bayesian substrate + sub-month horizon + per-location calibration + cohort- relative decomposition + replication-crisis discipline surface against the Calibrate + Decompose + Score + Audit bundle. The Tier 3 Fractional CMO with AI Swarm embeds 1-2 days per week for 6+ months and runs the bundle end-to-end against the per-location-forecasting agent across the swarm.

Tier 1 AI Readiness Assessment — $10k, 2-3 weeks Tier 3 Fractional CMO with AI Swarm — $15-25k/mo Not sure which tier? Take the 3-question quiz