White Papers Managing Forward Curves in a Complex Market | Page 4

Managing Forward Curves in a Complex Market RISK MANAGEMENT AND REPORTING A third common use for forward curves is in risk management and reporting; and for these purposes, practices can vary widely amongst market participants. Some companies may wish to have Value-atRisk measurements and limits-monitoring processes match observable market data regardless of granularity. In this case, an exchange-based curve source will likely be the best option for forward curve development. Other companies may wish to apply liquidity and seasonality adjustments if they believe those practic- A ComTech Advisory Whitepaper es provide a more nuanced view of firm risk. For these companies, the use of non-exchange sources, in addition to exchange data, may provide them with the better fit curves that reflect their operations and risk portfolio. Regardless of which situation a company finds themselves, best risk practices dictate that a curve validation process is used in which independent forward curve data is compared to the forward curves that they use for financial reporting, risk measurement, and risk reporting. Companies that utilize forward curves derived from multiple sources, or with internally-developed adjustments, should enshrine a regular testing of those curves and adjustments as part of their risk management policies. And most critically, all forward curve information should be archived indefinitely for audit and compliance purposes. SOURCES OF FORWARD CURVE DATA As previously noted, the data used for the construction of forward curves will likely differ, sometimes dramatically, between different sources. Market activity, knowledge, and insights available to those different sources will impact their views of value of the commodity in the future. Different sources may, and usually will, provide a different set of periods over different time horizons. For example, an exchange may have monthly contracts that will extend for ten years, while an over-the-counter broker may quote future prices as multi-month strips that extend over a period of 5 years. In addition, settlement data from an exchange will be limited to transactions that have been executed across that platform and the “accuracy” of that data may be constrained by the liquidity in those markets – the fewer the deals consummated at any particular market point, the greater the impact an anomalous trade will have in establishing the published price. Further, should a trade not be consummated for a particular market point in any given period, the exchange will still “settle” its open interest using a formula-based approach in order to keep margin accounts in balance. Clearly, simply relying on exchange data to establish forward prices may be insufficient, particularly for illiquid points or markets. So, in order to meet a company’s requirement for forward curves reflective of their markets and curve usage, more robust curves can be internally developed against independent market data aggregated from multiple sources. There are many choices available to market participants seeking forward curve data sources. The most common sources are exchanges, brokers, data publishers, data distributors, ETRM system vendors, and internally-developed models. It is important to understand your company’s intended use of any particular forward curve in order to select the appropriate sources and methodologies for deriving those curves. It is also key to understand the limitations and methodologies inherent in each of the selected data sources. Internally modeled curves may be the only option where reliable market data does not exist (e.g. illiquid points and tenors). In these circumstances, the forward curve’s quality is highly dependent on the quality of the market data inputs, modeling assumptions and methodology. Whenever internally modeled curves are used, calibration and back-testing should be done regularly to validate the quality of the curve and its assumptions. Additionally, when possible, internally developed curves should be compared to independently modeled curves for further validation. © Commodity Technology Advisory LLC, 2015, All Rights R eserved.