SteelHome Database - Coke/Coal & Steel Raw Materials

It is the price tape for the raw-material side of the world's largest steel industry. Shanghai SteelHome, an independent price reporter operating since 2004, organizes its database into nine sections - Price Index, Market Price, Ex-works Price, Market Inventory, Production & Consumption, Import & Export, Macro Data, Steel-consuming Sectors and Financial Figures - and the coal-and-coke slice carries exactly what blast furnaces burn: metallurgical coke, foundry coke, coking coal, PCI coal, anthracite and thermal coal, plus the coking by-products (coal tar, crude benzol, ammonium sulfate) that tell you how the ovens themselves are running.

What separates it from a statistical series is vantage point. Official accounts count tonnes produced and consumed; SteelHome quotes prices where they clear. Metallurgical coke gets five named ports - Jingtang, Tianjin, Lianyungang, Qingdao and Rizhao. Coking coal gets producing provinces - Shaanxi, Lvliang, Shandong. Thermal coal gets the northern Chinese ports that load the seaborne trade. DCE coking-coal and coke futures settlements sit beside the physical prints, more than 70 large-and-medium Chinese steelmakers report ex-works, and the house index family - Coal Price Index, Coke Price Index, Raw Materials Price Index among them - condenses each day into single readings.

Five rows from the August 21, 2026 capture, reproduced flat as they arrive:

# price records as delivered (Aug 21 2026 capture)
layer=market-price  currency=RMB/t

headline : Market price of Jingtang Port Metcoke
  product = Metallurgical coke       market = Jingtang Port           freq = daily

headline : Market price of Imported Coking Coal at Chinese ports
  product = Coking coal (imported)   market = Chinese ports           freq = daily

headline : Market price of Shaanxi Coking Coal
  product = Coking coal              market = Shaanxi province        freq = daily

headline : Thermal Coal Price at Northern Ports
  product = Thermal coal             market = Northern Chinese ports  freq = daily

headline : Weekly Data Summary of China Coal and Coke Market
  product = Coal/coke aggregate      market = China national          freq = weekly

Read them as one chain seen from three checkpoints. The Shaanxi row is the mine gate - coking coal priced where it leaves the producing province. The Jingtang row is the berth gate - met coke sitting at a named port waiting for a furnace. The imported coking-coal row is the import gate, where seaborne cargoes meet the domestic price and an arbitrage either exists or does not. The northern-ports thermal row covers the seaborne interface, and the weekly national summary rolls the whole coal-and-coke market up for readers who need the week, not the day. Each checkpoint keeps its own clock and its own RMB quote; nothing arrives pre-blended into an average you cannot take apart.

Seven columns carry a quote, and they behave like coordinates rather than a spreadsheet: pick a product, a market, a date, and the row resolves. Product names the grade - metallurgical coke through thermal coal. Market/Region pins the assessment point, a named berth for port quotes or a province for regional ranges. Price holds the level in RMB per tonne, with a published USD/RMB reference rate travelling beside it for conversion (6.7817 on the August 21, 2026 capture). Date stamps the assessment. Three further layers extend the same vocabulary: Index value for the house-index prints, Inventory for port stocks of coke and coal, and Futures settlement for the DCE coking-coal and coke closes.

One honesty note belongs in the dictionary itself. The column vocabulary above is mapped from the published assessment language - headline series, daily index reports and section documentation - rather than read off a screen nobody outside the vendor has parsed, so exact table headers get confirmed at sampling. Deeper cuts fold under additional fields on request: by-product series, customs-level import/export detail, provincial production and consumption statistics, mill operating rates and cost lines. They are named when your sample is cut rather than promised blanket here.

Geography - China-first, deliberately. Five named met-coke ports, three producing provinces for coking coal, the northern load ports for thermal coal, and ex-works feeds from more than 70 large-and-medium steelmakers spread mill-gate pricing across the steelmaking heartland. Supplementary international coverage - USA, CIS, Japan, South Korea, India, the Middle East - rides along as trade context for the domestic assessments rather than as standalone country files.

Temporal - daily assessments and daily index reports form the top layer, a weekly national summary rolls the coal-and-coke market up, and provincial production, consumption and trade series run monthly to annual. Historical depth varies by section and is confirmed per series when your sample is cut. For scale: of the 1,744 datasets in Datadory's catalog, only 89 carry any daily-resolution cadence, so a record built on daily assessments sits near the top of the pile for recency.

Granularity - product-by-market price points (one grade at one port or province), port-level inventory volumes, provincial aggregates, and mill-level ex-works quotes where the mill is named. Nothing forces you through a composite before you can see the print underneath it.

API, files, or your warehouse. Daily, weekly, or hourly.

Pick the channel your stack already speaks. The rows arrive identical either way - product, market, price, date, with the index, inventory and futures layers keyed to the same vocabulary - extracted so nothing in your pipeline re-assembles a price reporting house by hand.

Every delivery ships with the field dictionary above plus these sample rows for validation, so your first join attempt happens against evidence, not hope.

Scope the slice you actually touch when you request the sample - the five met-coke ports alone, provincial coking coal beside imported cargoes, northern-ports thermal, the index layer, inventories, DCE settlements, or the lot - and the sample comes back shaped to those choices with the dictionary unchanged.

A daily, port-level read on the world's biggest met-coal market earns its keep in six jobs:

• Procurement and price benchmarking - mills, coke buyers and traders track RMB moves port by port instead of averaging a continent; the price-monitoring playbook shows the alerting pattern.
• Import-parity arithmetic - imported coking coal at Chinese ports against the Shaanxi and Lvliang provincial prints builds the seaborne-versus-domestic spread in two rows and one subtraction; the quant-backtesting playbook handles the panel version.
• Supply-chain mapping - ex-works feeds tie price points to named steelmakers, so a cost move traces to specific mills; see supply-chain mapping.
• Demand forecasting - provincial production and consumption statistics beside port inventories turn stock builds into forward-demand signals; the demand-forecasting playbook fits the cadence.
• Credit and counterparty screening - operating rates, production costs and the financial figures section give mill-level health checks a numeric base; the credit-risk-screening playbook shows the screen.
• Market sizing - vendors selling into Chinese coke, refractory or steelmaking chains size segments bottom-up from named mills and provinces; start from market sizing.

Market researchers and consultants get the chapter-one exhibit for any Chinese steelmaking study - port, province and mill-gate prices measured one way across the whole raw-materials chain; the workflow lives on market researchers x coal & consumable fuels. Investors and quants get daily-resolution inputs for steel-margin and arbitrage models, with futures settlements beside the physicals so basis needs no second source; see investors & quants. Data scientists and ML engineers get a typed panel ready for feature work - the ml-model-training playbook covers the joins. Developers and builders integrate one parser across prices, indices, inventories and futures instead of maintaining one per layer; guidance at developers & builders. Journalists and academics cite a named price reporter's daily assessment rather than a screenshot of one; background at journalists & academics.

Provenance note - published by Shanghai SteelHome Information Technology Co., Ltd, an independent steel-industry price reporter operating since 2004 and serving steel-related clients globally. One vocabulary - product by market by date - runs across the price, index, inventory and futures layers.

Methodology note - the index family (Coal Price Index, Coke Price Index, Raw Materials Price Index, plus the steel, iron ore, semis, scrap, ferroalloy and global steel indices) is the reporter's own construction, published as daily readings that condense its assessments. Use the indices for trend and the underlying prints for basis work; the two answer different questions.

Completeness note - Datadory scores this record 5/10, and the honest edges are structural: coverage is China-centered with international detail riding as context rather than depth, the English-language mirror may trail the Chinese-language edition, and exact subscriber-table headers stay unconfirmed until sampling. All three are restated when your sample is cut rather than smoothed over here.

Where to go next - set the price tape beside its peers: the EIA Annual Coal Report for US production-side statistics, the Global Coal Plant Tracker for the consuming fleet, and OilPriceAPI coal benchmarks for seaborne markers outside China. Browse the full vertical on the coal & consumable fuels data hub or the best coal & consumable fuels datasets ranking, and read the industry data guide for how practitioners combine them.