Skip to content
AlpineDataWorks.AI
← All products
Energy · Index Gold

Energy-Inventory Tightness

How tight are US petroleum stocks?

Refresh
weekly
History
2.8 yrs
Plan
Gold
100.0/ 100
Stable

US crude oil inventories (national, ex-SPR)

2023-09-01 2.8 yrs · 148 pts 2026-06-26

Top drivers

crude_stocks_kbz_score_vs_2yr_baseline4wk_stock_change_kb
⌁ mcp.call("adw-062") vADW-062-live-1.0
Use cases

What it unlocks

For an agent

A commodity trading agent polls ADW-062 weekly; when the EIA petroleum stocks z-score hits 100 (current: 100.0, 100th percentile — maximum tightness in the 148-week backtest history since September 2023) with a rising trend, the agent flags upstream crude long positions for hold and queues a review of any short-dated refinery supply contracts, citing source_lineage (EIA weekly petroleum status report) in the trade rationale. The 100th-percentile score means physical inventories are tighter than any prior week in the tracked history, giving the agent a concrete threshold to escalate to a human trader rather than relying on oil-price moves alone, which are coincident.

📈

For the business

An energy supply-chain lead at a large airline or industrial manufacturer uses ADW-062 to time jet-fuel and feedstock forward purchases. A score of 100 — maximum tightness ever recorded in the tracked EIA series — signals that spot price spikes are likely imminent as refiners draw on already-depleted buffers. Versus the status quo of watching WTI spot prices (which reflect tightness only after it has occurred), the inventory-tightness score provides a structural leading indicator, allowing the procurement team to buy forwards before the spot curve fully reprices.

Forward outlook

Prediction

Horizon
Recommended use
Measure US crude oil market tightness for energy trading, procurement hedging, and macro oil-price risk assessment. High score = tight supply. Weekly cadence. Monthly EIA report validation lags ~2 weeks.
Methodology

How it's built

composite z-score 0-100

Version ADW-062-live-1.0 · validated to beat a naive baseline · benchmark: n/a