Handling the discontinuity in futures prices when time series modeling of commodity cash and futures prices

Abstract

Futures prices are discontinuous, with each future price series ending at maturity. Differencing before splicing can create a continuous future return series, but still leaves price levels with discrete jumps. When comparing cash and futures prices, there is a need to either make the futures more like the cash price by adding back the changes at rollover or removing the nonstationarity and seasonality from cash prices. In the specific situation of only testing market efficiency of futures prices, we propose using panel unit root tests. Our empirical examples using weekly prices show the null hypothesis of a unit root is not rejected in most cases regardless of the test used.

Commodity price forecasting via neural networks for coffee, corn, cotton, oats, soybeans, soybean oil, sugar, and wheat

“…The first day of the month as the roll date rule and calendar‐weighted rolling as the price adjustment rule are perfectly deterministic, predictable, and smooth, and they do not contaminate economic aspects of the price history, making the resultant series appropriate for economic forecasting purposes (StevensAnalytics, 2022). The small advantage of the neural network model found in this work as compared to the no‐change model could be due to this particular price adjustment since the adjusted series will contain the mean reversion and the seasonality that is in the cash prices (Maples & Brorsen, 2022). It is worth noting that these futures markets have price limits, and not adjusting for price limits is a limitation of this work.…”