Salmon Market Volatility Prediction using LSTM

The salmon industry is an essential component of the global food economy and, like any market, is subject to price fluctuations and volatility.

Salmon is a volatile commodity, with increased demand in recent decades resulting in rising prices and increased volatility. The high price volatility poses a challenge for all market participants, including producers, processors, traders, and intermediaries.

For participants in this market, the ability to accurately forecast these fluctuations is crucial for making informed decisions and mitigating risks.

A study conducted by Mikaella Zitti of the Department of Economics and Resource Management at the Norwegian University of Life Sciences has focused on predicting salmon market volatility and explored the effectiveness of a deep learning technique known as Long Short-term Memory (LSTM) networks.

Deep Learning Technique

Deep learning techniques, such as neural networks, have shown great potential for forecasting financial data, including commodity prices.

Among these techniques, Recurrent Neural Networks (RNNs) are particularly suited for predicting financial market volatility due to their ability to learn temporal dependencies from time series data.

Specifically, Long Short-Term Memory (LSTM) networks, a type of RNN, are capable of capturing long-term dependencies, producing promising results in various forecasting tasks.

Importance of Predicting Salmon Market Volatility

Salmon market volatility is a phenomenon that can significantly impact industry participants, from producers and distributors to investors and consumers. Variability in salmon prices can have a direct impact on company profitability and the stability of international food markets.

To address this challenge, the study aimed to evaluate the ability of LSTM, a deep learning technique, to predict salmon market volatility multiple steps ahead. Additionally, the performance of LSTM was compared to the Autoregressive Moving Average (ARMA) model, a traditional approach in time series analysis commonly used as a benchmark.

Study Results: LSTM vs. ARMA

The study results indicate that the ARMA model outperforms LSTM in predicting salmon market volatility. This suggests that, in this specific context, LSTM may not be able to effectively exploit nonlinear patterns in salmon market volatility. In other words, deep learning methods may not necessarily be the best choice for this type of prediction, at least with the configuration and data available in the study.

However, it is important to note that both LSTM and the ARMA model showed significant discrepancies between actual volatility values and predictions. This underscores the intrinsic complexity of accurately predicting volatility in the salmon market and possibly in other similar food markets.

Implications for the Salmon Industry

While LSTM did not outperform the ARMA model in this study, it is essential to recognize that each market is unique and may require different modeling approaches.

Participants in the salmon industry can use these findings as a signal to consider a variety of approaches to predict volatility. The use of deep learning techniques should not be entirely dismissed, as their effectiveness may vary depending on the data and market conditions.

Furthermore, the discrepancy between predictions and actual values highlights the importance of ongoing research and improvement of prediction methods in the salmon industry. Companies can invest in collecting and analyzing more accurate data and developing customized models that better fit their specific needs.


In summary, the study on salmon market volatility prediction provides valuable insights for industry participants and highlights the importance of flexibility in choosing prediction methods.

The volatility of food markets is a constant challenge, and continuous research is essential to enhance the ability to anticipate and manage these risks in the salmon industry and beyond.

Mikaella Zitti
Department of Economics and Resource Management
Norwegian University of Life Sciences
Ås, Norway

Reference (open access)
Mikaella Zitti (2023) Forecasting salmon market volatility using long short-term memory (LSTM), Aquaculture Economics & Management, DOI: 10.1080/13657305.2023.2255346

Leave a Comment