|
on Forecasting |
| By: | Katarzyna Maciejowska; Tomasz Serafin; Bartosz Uniejewski |
| Abstract: | This paper presents a novel approach for constructing probabilistic forecasts, which combines both the Quantile Regression Averaging (QRA) method and the Principal Component Analysis (PCA) averaging scheme. The performance of the approach is evaluated on datasets from two European energy markets - the German EPEX SPOT and the Polish Power Exchange (TGE). The results indicate that newly proposed solutions yield results, which are more accurate than the literature benchmarks. Additionally, empirical evidence indicates that the proposed method outperforms its competitors in terms of the empirical coverage and the Christoffersen test. In addition, the economic value of the probabilistic forecast is evaluated on the basis of financial metrics. We test the performance of forecasting models taking into account a day-ahead market trading strategy that utilizes probabilistic price predictions and an energy storage system. The results indicate that profits of up to 10 EUR per 1 MWh transaction can be obtained when predictions are generated using the novel approach. |
| Date: | 2023–03 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2303.08565&r=for |
| By: | Chang Liu; Sandra Paterlini |
| Abstract: | Stock price prediction is a crucial element in financial trading as it allows traders to make informed decisions about buying, selling, and holding stocks. Accurate predictions of future stock prices can help traders optimize their trading strategies and maximize their profits. In this paper, we introduce a neural network-based stock return prediction method, the Long Short-Term Memory Graph Convolutional Neural Network (LSTM-GCN) model, which combines the Graph Convolutional Network (GCN) and Long Short-Term Memory (LSTM) Cells. Specifically, the GCN is used to capture complex topological structures and spatial dependence from value chain data, while the LSTM captures temporal dependence and dynamic changes in stock returns data. We evaluated the LSTM-GCN model on two datasets consisting of constituents of Eurostoxx 600 and S&P 500. Our experiments demonstrate that the LSTM-GCN model can capture additional information from value chain data that are not fully reflected in price data, and the predictions outperform baseline models on both datasets. |
| Date: | 2023–03 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2303.09406&r=for |
| By: | Haritha GB; Sahana N. B |
| Abstract: | The cryptocurrency ecosystem has been the centre of discussion on many social media platforms, following its noted volatility and varied opinions. Twitter is rapidly being utilised as a news source and a medium for bitcoin discussion. Our algorithm seeks to use historical prices and sentiment of tweets to forecast the price of Bitcoin. In this study, we develop an end-to-end model that can forecast the sentiment of a set of tweets (using a Bidirectional Encoder Representations from Transformers - based Neural Network Model) and forecast the price of Bitcoin (using Gated Recurrent Unit) using the predicted sentiment and other metrics like historical cryptocurrency price data, tweet volume, a user's following, and whether or not a user is verified. The sentiment prediction gave a Mean Absolute Percentage Error of 9.45%, an average of real-time data, and test data. The mean absolute percent error for the price prediction was 3.6%. |
| Date: | 2023–03 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2303.09397&r=for |
| By: | Largier, John L; Munger, Sophie; Shilling, Fraser; Roettger, Robin |
| Abstract: | Like most coastal states in the U.S., California’s shoreline communities and ecosystems have been exposed to flooding related to sea level rise and storms, which jeopardize their persistence and well-being. Shoreline transportation is especially vulnerable in certain places to flooding and failure, and because it is part of a continuously used network with little redundancy, it transfers its vulnerability to regional transportation networks. Forward-projected inundation/flooding risk is typically modeled at coarse spatial and temporal scales, which are useful at regional and decadal scales, but less useful for coastal managers and flood responders. This project improved assessment of both overall probability and short-term forecasts of water level for specific locations in San Francisco Bay that are vulnerable to flooding associated with sea level rise. The authors have developed probability assessment and forecasts through developing data-based, site-specific, model-independent approaches, which can be compared with and help to improve regional models of coastal flooding (e.g., CoSMoS). Water level data were collected across fine-scale arrays at fluvial-bay junctures in Sonoma and Marin Counties. The primary analysis is based on deconstructing water level records into multiple quasi-independent signals, which can be better predicted and recombined to produce probability of extreme events and to produce short-term forecasts during a flooding event based on predicted weather, wind, rain, and tide. In addition, real-time water level data are now available to first responders at critical locations in Novato Creek and Petaluma River when there is potential for flooding, as well as during a flood event. This is a pilot project that could be replicated at many other vulnerable locations around San Francisco Bay and elsewhere. View the NCST Project Webpage |
| Keywords: | Engineering, Physical Sciences and Mathematics, flood forecast, sea-level rise, storm surge, king tides, extreme event |
| Date: | 2023–03–01 |
| URL: | http://d.repec.org/n?u=RePEc:cdl:itsdav:qt4ts3d98k&r=for |