ForecastWorx

AI-Powered Demand Forecasting That Learns Every Pattern

ForecastWorx combines AI Forecast Modeling, using the OpenAI or Anthropic model you choose, with a Horizon-Adaptive Ensemble and a pattern-aware, backtest-scored engine. You get forecasts that follow seasonality over the long term and use the most accurate model over lead time, all at the item-location level.

This feature is part of the ForecastWorx connected planning platform — an AI-native system where demand forecasting, inventory optimization, replenishment automation, and supply planning share the same data layer and work together in real time.

Key Outcomes

Capabilities

Hierarchical Pattern Classification

The engine detects demand patterns in a strict hierarchy: data sufficiency → intermittency → declining → seasonality → trend → spiky overlay → stable vs erratic residual. This prevents misclassification.

AI Forecast Modeling

AI forecast models built on large language models read demand history, patterns, and external signals to build each forecast. You choose which OpenAI or Anthropic model runs it, and every result is still backtested and risk-scored.

Horizon-Adaptive Ensemble

The long-term forecast follows the shape of your seasonality, so it reads as a smooth, realistic profile. Within the ordering horizon, the forecast switches to whichever model was most accurate over lead time, so inventory and orders are tuned to where precision matters most.

Lead-Time Backtesting

ForecastWorx backtests over the item's actual lead time using rolling origin cross-validation, not arbitrary windows. This evaluates models where accuracy matters most — the replenishment horizon.

Tournament Scoring

Each eligible algorithm is scored using WMAPE across backtest folds. The tournament selects the best-fit models per item-location automatically.

Forecast Risk Score (FRS)

Every forecast receives a governance score (0–100) measuring data quality, model fit, and volatility. Planners focus review on high-risk items, not every SKU.

Stability Controls

Dampening and banding rules prevent period-to-period forecast whiplash, ensuring smooth signals for downstream inventory and replenishment systems.

Promotional Uplift Modeling

Learns promotional response curves from historical data and integrates planned promotions into the baseline forecast.

External Signal Integration

Incorporate weather, economic indicators, and market data as regression features to improve forecast accuracy.

Explainable Forecasts

Every forecast comes with decomposition showing the contribution of trend, seasonality, promotions, and anomalies — so planners can validate with confidence.

How It Works

  1. Ingest & Classify: Historical demand is analyzed at the item-location grain. The engine classifies each series into one of seven primary patterns — new/sparse, intermittent, seasonal, seasonal+trend, trending, level/stable, or erratic — plus overlays like declining or spiky.
  2. Narrow Eligible Models: Based on the detected pattern, the engine narrows which algorithms are eligible. Intermittent items get Croston/SBA methods. Seasonal items get Holt-Winters and seasonal decomposition. Trending items get regression models. No algorithm runs where it doesn't belong.
  3. Backtest & Score: Every eligible model is backtested across the item's lead time using rolling origin cross-validation. Models are scored by WMAPE. The tournament ranks them and selects top performers.
  4. AI Model & Horizon-Adaptive Ensemble: AI Forecast Modeling runs on the OpenAI or Anthropic model you choose. The Horizon-Adaptive Ensemble then blends the results: a seasonality-driven shape for the long-term profile, and the best-fit model over lead time for the ordering horizon.
  5. Stabilize & Govern: Final forecasts pass through stability controls (dampening, banding) to prevent whiplash. Each forecast receives a Forecast Risk Score for planner prioritization and governance.

Role Perspectives

How the Engine Works: Pattern-first, backtest-scored forecasting

The ForecastWorx engine doesn't apply one model to everything. It classifies, narrows, tests, and ensembles — automatically.

For Planners: Focus on exceptions, not every SKU

The Forecast Risk Score surfaces only the items that need attention. Low-risk forecasts flow through automatically.

For Executives: Governance and confidence at scale

Real-time visibility into forecast quality, risk distribution, and model performance across the entire portfolio.

Frequently Asked Questions

How does ForecastWorx generate demand forecasts?

ForecastWorx uses a pattern-aware, rules-driven engine. It first classifies each item-location's demand pattern (intermittent, seasonal, trending, etc.), then narrows eligible algorithms, runs backtesting over the item's lead time, and produces an ensemble-weighted forecast from the top performers.

What is AI Forecast Modeling?

AI Forecast Modeling replaces AutoML Ensemble Forecasting. It uses large language models to build forecasts at the item-location level, and you choose which OpenAI or Anthropic model it uses. Forecasts are still backtested, risk-scored, and stabilized.

What is the Horizon-Adaptive Ensemble?

It uses a different forecast logic for each horizon. Long-term periods follow a seasonality-driven shape that is smooth and easy to read. The ordering horizon uses the model that was most accurate over lead time, which improves inventory and order decisions.

What demand patterns does ForecastWorx detect?

Seven primary patterns: New/Sparse, Intermittent, Seasonal, Seasonal+Trend, Trending, Level/Stable, and Erratic. Plus overlays for Declining, Spiky/Lumpy, and High Uncertainty. The hierarchy prevents misclassification.

What is the Forecast Risk Score?

Every forecast gets an FRS from 0–100 measuring data quality, model fit, and volatility. High-risk items surface to planners for review, while low-risk items flow through automatically. This eliminates the need to manually review every SKU.

How does backtesting work in ForecastWorx?

ForecastWorx uses lead-time-oriented backtesting with rolling origin cross-validation. Models are evaluated over the actual replenishment lead time, not arbitrary windows, ensuring accuracy is measured where it matters most.

Do I need a data science team to use ForecastWorx?

No. ForecastWorx handles model selection, training, and evaluation automatically. Planners interact with the results, not the algorithms.

What algorithms does ForecastWorx use?

The ensemble includes ARIMA, ETS, Prophet, Croston, TSB, gradient boosting, and several proprietary models. The system automatically selects the best performer for each item-location.

Can I incorporate external data?

Yes. Weather data, economic indicators, market indices, and custom external signals can be integrated as regression features to improve forecast accuracy.

Related Features

Related Solutions

Book a Demo → | ← All Features