Using LSTM for predictions of the regulating direction on the Swedish electricity intraday market
(2023) STAN40 20231Department of Statistics
- Abstract
- Electricity price prediction is increasingly important as the prices become more volatile due to the expan- sion of weather-dependent renewable electricity sources. The electricity market of focus in this thesis is the Nord pool intraday market and the regulating prices occurring on it. For actors trading on the intra- day market increased volatility implies greater risks since the regulating prices can cause huge imbalance fees.
The main objective of this thesis is therefore to create multivariate multi-step predictive LSTM models for predicting the regulating direction on the intraday market. These predictions could help trading actors minimize their risks and avoid high imbalance fees. LSTM models are used since they have previously... (More) - Electricity price prediction is increasingly important as the prices become more volatile due to the expan- sion of weather-dependent renewable electricity sources. The electricity market of focus in this thesis is the Nord pool intraday market and the regulating prices occurring on it. For actors trading on the intra- day market increased volatility implies greater risks since the regulating prices can cause huge imbalance fees.
The main objective of this thesis is therefore to create multivariate multi-step predictive LSTM models for predicting the regulating direction on the intraday market. These predictions could help trading actors minimize their risks and avoid high imbalance fees. LSTM models are used since they have previously been proven efficient for time series predictions where historical data and periodicity must be accounted for.
Two LSTM models and an ARIMA model are compared. The simpler LSTM model performs surpris- ingly well compared to the more complicated LSTM model. The models are benchmarked against the most common regulating direction percentage during 2022. On average the results are close to the benchmark. The results of this thesis underline the complexity of predicting the regulating direction. (Less) - Popular Abstract
- Electricity price prediction is increasingly important as the prices become more volatile due to the expan- sion of weather-dependent renewable electricity sources. The electricity market of focus in this thesis is the Nord pool intraday market and the regulating prices occurring on it. For actors trading on the intra- day market increased volatility implies greater risks since the regulating prices can cause huge imbalance fees.
The main objective of this thesis is therefore to create multivariate multi-step predictive LSTM models for predicting the regulating direction on the intraday market. These predictions could help trading actors minimize their risks and avoid high imbalance fees. LSTM models are used since they have previously... (More) - Electricity price prediction is increasingly important as the prices become more volatile due to the expan- sion of weather-dependent renewable electricity sources. The electricity market of focus in this thesis is the Nord pool intraday market and the regulating prices occurring on it. For actors trading on the intra- day market increased volatility implies greater risks since the regulating prices can cause huge imbalance fees.
The main objective of this thesis is therefore to create multivariate multi-step predictive LSTM models for predicting the regulating direction on the intraday market. These predictions could help trading actors minimize their risks and avoid high imbalance fees. LSTM models are used since they have previously been proven efficient for time series predictions where historical data and periodicity must be accounted for.
Two LSTM models and an ARIMA model are compared. The simpler LSTM model performs surpris- ingly well compared to the more complicated LSTM model. The models are benchmarked against the most common regulating direction percentage during 2022. On average the results are close to the benchmark. The results of this thesis underline the complexity of predicting the regulating direction. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9121894
- author
- Johansson, Anna LU
- supervisor
-
- Rani Basna LU
- organization
- course
- STAN40 20231
- year
- 2023
- type
- H1 - Master's Degree (One Year)
- subject
- keywords
- LSTM ARIMA Electricity market Nord Pool Intraday market Forecasting Regulating direction
- language
- English
- id
- 9121894
- date added to LUP
- 2023-06-28 09:36:45
- date last changed
- 2023-06-28 09:36:45
@misc{9121894,
abstract = {{Electricity price prediction is increasingly important as the prices become more volatile due to the expan- sion of weather-dependent renewable electricity sources. The electricity market of focus in this thesis is the Nord pool intraday market and the regulating prices occurring on it. For actors trading on the intra- day market increased volatility implies greater risks since the regulating prices can cause huge imbalance fees.
The main objective of this thesis is therefore to create multivariate multi-step predictive LSTM models for predicting the regulating direction on the intraday market. These predictions could help trading actors minimize their risks and avoid high imbalance fees. LSTM models are used since they have previously been proven efficient for time series predictions where historical data and periodicity must be accounted for.
Two LSTM models and an ARIMA model are compared. The simpler LSTM model performs surpris- ingly well compared to the more complicated LSTM model. The models are benchmarked against the most common regulating direction percentage during 2022. On average the results are close to the benchmark. The results of this thesis underline the complexity of predicting the regulating direction.}},
author = {{Johansson, Anna}},
language = {{eng}},
note = {{Student Paper}},
title = {{Using LSTM for predictions of the regulating direction on the Swedish electricity intraday market}},
year = {{2023}},
}