World Platinum Investment Council Supply & Demand Data

The platinum market keeping score on itself. The World Platinum Investment Council is the industry body funded by the major platinum producers, and its supply-and-demand hub has five parts: Platinum Quarterly (the flagship - the 47th edition analysed Q1 2026 with a revised full-year 2026 view, issued in multiple languages), a two-to-five-year outlook, Charts & Tables, an Archive, and the Historical Data core where the long-run numbers live.

That core states the proposition plainly: platinum supply and demand fundamentals from 1975 onwards, spliced across three compiler eras - Johnson Matthey for the decades through 2012, SFA (Oxford) for 2013-2018, Metals Focus from 2019 onward. The current-edition table set runs to roughly 111 sheets, and Table 1 is the one most teams want: the Platinum Supply-demand Balance, with columns for each year and rows running from refined production by region through stock movements, recycling by end-use, total supply, demand across four sectors, total demand, the market balance and cumulative above-ground stocks.

Every figure resolves against seven documented fields, verified during research, all denominated in thousands of troy ounces. Within Datadory's catalog of 1,744 datasets across 159 viable industries, this record scores 8/10 for quality - among the strongest fundamentals sources in the 27-record precious metals & minerals slice.

Get a sample of this dataset

One real column-year from the balance table, flat exactly as its fields arrive:

table                    : Platinum Supply-demand Balance
Year                     : 2013
Refined_Production_koz   : 6070
South_Africa_koz         : 4355
Zimbabwe_koz             : 405
North_America_koz        : 355
Russia_koz               : 740
Total_Mining_Supply_koz  : 5855
Recycling_koz            : 2000
Total_Supply_koz         : 7855
Automotive_Demand_koz    : 3110
Jewellery_Demand_koz     : 2945
Total_Demand_koz         : 8605
Balance_koz              : -750
Above_Ground_Stocks_koz  : 3390

That single year settles three questions buyers usually have to ask twice. First, concentration: South Africa's 4,355 koz of refined production is nearly three-quarters of the 6,070 koz total, which is why the regional split matters more here than in almost any other commodity ledger. Second, breadth: automotive and jewellery demand arrive as separate named lines (3,110 and 2,945 koz respectively), so sector arguments need no reverse-engineering out of narrative text. Third, ledger discipline: the balance prints negative (-750 koz, a deficit year), the arithmetic ties out across rows, and cumulative above-ground stocks carry forward at 3,390 koz - the inventory buffer every deficit story ultimately draws down.

Every other column-year repeats this anatomy, back through 1975 and forward through the current-edition forecast horizon.

Seven documented fields, definitions verified during research - small enough to learn in minutes, complete for what a supply-and-demand ledger owes you. One period column (Year / Quarter) stamps annual columns from 1975 onward, quarterly detail and forecast periods; Refined Production holds total refined output; the regional split assigns it across South Africa, Zimbabwe, North America, Russia and Other; Recycling carries secondary supply by end-use stream; the sector quartet splits demand into automotive, jewellery, industrial and investment; Balance states supply minus demand; and Above Ground Stocks cumulates estimated inventory at period end.

There is no hidden tail beyond these seven. What the dictionary below shows is the entire headline schema; the deeper cut lives in the sub-components - chemical, petroleum, electrical, glass, medical & biomedical and hydrogen stationary inside industrial demand, bars-and-coins, ETF-holdings and exchange-stock changes inside investment - which ride along as named line items and arrive defined on request.

Geography - global totals first, then a producer split that makes concentration measurable: South Africa beside Zimbabwe, North America, Russia and Other, year by year. Demand is reported worldwide by sector rather than by consuming country, so regional demand arguments need an external join - the supply side, where geography actually bites, is fully split.

Temporal - an annual spine from 1975 through the current year, spliced across three compiler eras: Johnson Matthey to 2012, SFA (Oxford) 2013-2018, Metals Focus from 2019. Recent periods carry quarterly columns, and the current-edition tables extend a two-to-five-year forecast horizon. One caveat worth planning around: the oldest era explicitly tracks net demand rather than gross, so era boundaries deserve respect in any splice.

Granularity - annual for the long history, quarterly for recent periods and forecasts, one column-year per observation, every quantity in thousands of troy ounces. A fifty-year panel fits in a spreadsheet; fitting it into your model takes nothing more.

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

Name the sectors, the regions, the era handling you want and the window when you request the sample - the full 1975-onward panel, post-2013 on a single compiler basis, or just the balance and stocks columns your model consumes. The sample ships first either way; extraction, era-splicing and normalization are our problem, and the ongoing feed lands on whatever cadence your workflow runs, shaped to the scope you named rather than as an undifferentiated dump.

A fifty-year market ledger earns its keep in five jobs:

• Investors and quant researchers - the fundamental series underneath the platinum price: balance, stocks and sector demand paired against fixings feeds for factor and deficit-cycle strategies.
• Market researchers and consultants - the exhibit reviewers stop arguing about: production shares, demand mix and deficit years cited straight from the industry body's own arithmetic.
• Data scientists and ML engineers - typed, regularly spaced annual rows that drop into forecasting pipelines as targets or features with no transcription pass up front.
• Developers and builders - dashboards and alerting keyed to balance flips and stock drawdowns, wired against one flat repeated layout.
• Competitive intelligence teams - demand-side context for products exposed to platinum, from autocatalyst programs to hydrogen hardware, evidenced with the market's own numbers.

Deeper guidance sits on the investors & quants, market researchers and data scientists persona pages.

Investors & Quant Researchers get the ledger the price trades on - deficits and stock drawdowns typed per year since 1975, ready to align against any benchmark series. Market Researchers & Consultants get provenance that traces to the producers' own council, compiled per era by named research houses, which is the citation peer review accepts. Data Scientists & ML Engineers get seven typed columns and explicit units throughout, so a deficit never masquerades as a parse error mid-pipeline. Developers & Builders get the smallest plausible integration surface: one balance-table layout, repeated identically across eras and editions. Competitive Intelligence & Product Teams get neutral ground truth - their platinum-exposed product lines measured against the industry's own demand accounting. Start from the precious metals & minerals data hub, then read platinum production by country for the country-level cut of the supply story.

Provenance note - published by the World Platinum Investment Council (WPIC), the industry body funded by major platinum producers, with the underlying series attributed per compiler era: Johnson Matthey through 2012, SFA (Oxford) for 2013-2018 and Metals Focus from 2019. First-party framing with third-party compilation; the compiler attribution rides along with the numbers and belongs in any citation.

Methodology note - era splicing needs care. Series definitions shift across compilers, and the oldest era explicitly tracks net demand rather than gross, so a naive 1975-2025 concatenation mixes bases. Treat the splice points (2012/2013, 2018/2019) as seams to inspect, and expect the sample pass to confirm which basis answers your question before anything recurring ships.

Completeness note - Datadory scores this record 8/10 against a catalog mean of 7.81 across 1,744 datasets, top-quartile within the 27-record precious metals & minerals slice. The ceiling is definitional rather than structural: compiler-era discontinuities cap how cleanly the deepest history merges. For the same house-record treatment of neighbouring metals, start from World Gold Council Goldhub data and Silver Institute market statistics, or read the ranked best precious metals & minerals datasets.