Trading in the energy market is tricky when it comes to wind power. Energy is typically traded in two types of markets: spot and futures markets. Pricing in the spot market is volatile as it driven by the real-time demand and supply situation, which requires the trader to focus on shortfalls and perform balancing to close the gaps in the earlier trades. Here, the small wind power generator has little say in the selling price. However, in a futures (typically a 'day-ahead') market, the producer can set a selling price for the power. Therefore, trading in the futures market is more attractive for the producer. Wind energy producers would prefer to bid on these markets in order to maximize profits. Missed forecasts (if the bid quantity is not delivered), on the other hand, entail penalties.
Most producers therefore decide on the quantity of a bid in a futures markets based on farm yield predictions and the price in the time-ahead market. Renewable power portfolio managers, power brokers, and producers have to consider constraints such as the varying accuracy in yield prediction, the limitations of buffers available for energy storage, and the fluctuations in price (involving penalties). In this challenging environment, data-driven approaches can be employed to develop better trading strategies.
Recommendations for a Robust Bidding Strategy
Optimizing bids for 'one period ahead' ignores valuable information in the projected prices (troughs and peaks) and yield estimations obtained through appropriate statistics. The supplier should account for the interplay between the projected price, a specific buffer capacity, and the predicted yield across a longer time horizon.
A tool that best aids the supplier must take into account:
- Prediction algorithms, error margins, and the associated statistics
- Multiple bidding timeframes, including longer horizons
- The nature of a specific market, including factors like local market conditions and wind penetration
- Pricing variations over a time period
- Penalties and regulation costs for under-delivery, as well as storage capacity and operating costs
- Discount factors to translate cash flows between slots for an accurate net present value (NPV)
Such a tool can be created by first modeling the bidding process specific to a particular market. Next, under a set of simplifying assumptions, an optimal bid can be determined that maximizes the cash flows over a future horizon. The solution can be evaluated on realistic data sets, and can be improved to also handle situations where the assumptions are invalid.