Trading Strategy: A Case Study on the Day-Ahead and Week-Ahead Electricity Prices

The study consists of two interrelated parts. In the first part, we aim to identify the best predictive model of day-ahead electricity prices. In particular, we verify the existence of a dependence of the spot price on power generation and variable costs of coal and gas plants in the French and the UK market. In the second part, we capitalize on the findings from the first part. We assess whether a trading strategy based on the forecasting abilities of regressive models can be a relevant decision making tool in order to evaluate if it is preferable to buy at the week ahead price or wait until the following week to buy at the day ahead price. For this purpose, we compare the forecasting performance of an autoregressive model, a generalised additive model, a piecewise linear model and an ARIMA (1,1,7) GARCH(1,1) model and assess their expected shortfalls with a view to identifying the best model for implementing the trading strategy. The results show that for both markets the relationship between the day-ahead price and volume and cost variables is recognized. In addition, the French day ahead price seems to be more dependent on these factors than the UK day ahead price. In the second part, the autoregressive model with the variable cost of gas plants, a volume variable and a long term seasonal component as additional covariates, obtained the best prediction performances and the lowest average shortfall. This model appears to be the most suitable one in order to inform the decisions made by power traders every week and improve their performances.