There are so many prices used in commodity trading, and we frequently need a way of organizing these prices to simplify the task of trading and risk management. In reality, there are prices that are closely related to each other, and exists a hierarchy among them. For example, my trader would answer like this, “The main price is XXX; then you add sulfur contents premium to get price YYY; add delivery cost then you get ZZZ”. So there are some ‘base’ curves, and others are priced at differentials, spreads, or netback priced to the base curves, and this is a hierarchy based on ‘Expert Opinion’.
How about doing this through statistical analysis? You may run a data-driven factor analysis to find out base prices, or simply looking for a highly correlated block in correlation matrix to find clusters (families) of prices. However, there is simpler, and visually intuitive way of doing that – hierarchical clusters. Continue Reading in LinkedIn Article1 and Article 2