Comparability and Normalization
What is normalization in a price assessment?
Normalization is the adjustment of gathered market information so it matches the base standard reflected in the price assessments. Markets are diverse: trades differ in delivery location, data quality, quantity, specification, and timing — merely averaging them would not represent the assessment value accurately. Normalization occurs concomitantly and iteratively with modeling, not as a strictly sequential step before it.
Each kind of difference has its own correction:
How are prices adjusted for location differences?
Trades reflecting different delivery locations are adjusted using freight rates, so every figure is expressed at the published location basis — typically the location where trading of that commodity is most common. This is how, for example, Energy Prices & Markets publishes energy commodity prices on a per-country basis: the same underlying market information, translated to each country's reference location through freight adjustments.
How are currencies and units handled?
The conversion runs both ways. Internally, every monetary figure is standardized to US dollars using a single set of reference exchange rates, and every physical quantity to the International System of Units (SI) using fixed conversion factors — the common basis on which all comparisons and models operate (see the Currency and Units pages for the underlying references). At publication, figures are then converted from this default basis to the currencies and units commonly used in each country, so readers see values in their familiar local conventions.
How are slower series aligned to the monthly cadence?
Some official series are published quarterly or annually — less often than the monthly reporting cadence. These slower series are aligned to the monthly grid through interpolation, so that all series share a common timeline and any month can be compared across the full dataset.
How are published figures rounded?
At the end of the transformation stage, historical and preliminary data are rounded to three significant figures, and forecast data to two. The difference is deliberate honesty about precision: forward-looking estimates carry greater uncertainty than observed values, and the published rounding reflects that.