|
on Forecasting |
| By: | Emmanuel Boadi |
| Abstract: | This study proposes a hybrid deep learning model for forecasting the price of Bitcoin, as the digital currency is known to exhibit frequent fluctuations. The models used are the Variational Mode Decomposition (VMD) and the Long Short-Term Memory (LSTM) network. First, VMD is used to decompose the original Bitcoin price series into Intrinsic Mode Functions (IMFs). Each IMF is then modeled using an LSTM network to capture temporal patterns more effectively. The individual forecasts from the IMFs are aggregated to produce the final prediction of the original Bitcoin Price Series. To determine the prediction power of the proposed hybrid model, a comparative analysis was conducted against the standard LSTM. The results confirmed that the hybrid VMD+LSTM model outperforms the standard LSTM across all the evaluation metrics, including RMSE, MAE and R2 and also provides a reliable 30-day forecast. |
| Date: | 2025–09 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2510.15900 |
| By: | Mr. Paul Cashin; Mr. Fei Han; Ivy Sabuga; Jing Xie; Fan Zhang |
| Abstract: | This paper evaluates three approaches to address parameter proliferation issue in nowcasting: (i) variable selection using adjusted stepwise autoregressive integrated moving average with exogenous variables (AS-ARIMAX); (ii) regularization in machine learning (ML); and (iii) dimensionality reduction via principal component analysis (PCA). Utilizing 166 variables, we estimate our models from 2007Q2 to 2019Q4 using rolling-window regression, while applying these three approaches. We then conduct a pseudo out-of-sample performance comparison of various nowcasting models—including Bridge, MIDAS, U-MIDAS, dynamic factor model (DFM), and machine learning techniques including Ridge Regression, LASSO, and Elastic Net to predict China's annualized real GDP growth rate from 2020Q1 to 2023Q1. Our findings suggest that the LASSO method outperform all other models, but only when guided by economic judgment and sign restrictions in variable selection. Notably, simpler models like Bridge with AS-ARIMAX variable selection yield reliable estimates nearly comparable to those from LASSO, underscoring the importance of effective variable selection in capturing strong signals. |
| Keywords: | China; GDP; Nowcasting |
| Date: | 2025–10–24 |
| URL: | https://d.repec.org/n?u=RePEc:imf:imfwpa:2025/217 |
| By: | Yimeng Qiu; Feihuang Fang |
| Abstract: | We study whether liquidity and volatility proxies of a core set of cryptoassets generate spillovers that forecast market-wide risk. Our empirical framework integrates three statistical layers: (A) interactions between core liquidity and returns, (B) principal-component relations linking liquidity and returns, and (C) volatility-factor projections that capture cross-sectional volatility crowding. The analysis is complemented by vector autoregression impulse responses and forecast error variance decompositions (see Granger 1969; Sims 1980), heterogeneous autoregressive models with exogenous regressors (HAR-X, Corsi 2009), and a leakage-safe machine learning protocol using temporal splits, early stopping, validation-only thresholding, and SHAP-based interpretation. Using daily data from 2021 to 2025 (1462 observations across 74 assets), we document statistically significant Granger-causal relationships across layers and moderate out-of-sample predictive accuracy. We report the most informative figures, including the pipeline overview, Layer A heatmap, Layer C robustness analysis, vector autoregression variance decompositions, and the test-set precision-recall curve. Full data and figure outputs are provided in the artifact repository. |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2510.20066 |
| By: | Diana Barro (Ca’ Foscari University of Venice); Antonella Basso (Ca’ Foscari University of Venice); Marco Corazza (Ca’ Foscari University of Venice); Guglielmo Alessandro Visentin (Henley Business School, University of Reading) |
| Abstract: | We propose a hybrid approach that combines Neural Networks with a Vector Autoregression (VAR) model to generate long-term forecasts of time series. We apply this methodology to forecast the impact of shifts in monetary policies within the Euro area on a comprehensive set of macroeconomic variables. Our analysis begins with a standard (linear) VAR model, which is then enhanced by incorporating Neural Networks to generate long-term forecasts for key variables such as the interest rate, inflation, real output, narrow money, exchange rate, and corporate bond spread. The results suggest that a Neural Network-VAR model offers improvements over the traditional linear VAR for forecasting certain macroeconomic variables in the long run. However, due to the limited sample size, the nonlinear model does not consistently outperform the linear VAR. |
| Keywords: | Forecasting; VAR; Neural Networks; Monetary policies; Euro area |
| JEL: | C32 C45 C53 E52 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:ven:wpaper:2025:24 |
| By: | Rahul Billakanti; Minchul Shin |
| Abstract: | We propose a simple binarization of predictors—an “at-risk” transformation—as an alternative to the standard practice of using continuous, standardized variables in recession forecasting models. By converting predictors into indicators of unusually weak states, we demonstrate their ability to capture the discrete nature of rare events such as U.S. recessions. Using a large panel of monthly U.S. macroeconomic and financial data, we show that binarized predictors consistently improve out-of-sample forecasting performance—often making linear models competitive with flexible machine learning methods—and that the gains are particularly pronounced around the onset of recessions |
| Keywords: | Recession Forecasting; Machine Learning; Feature Engineering; At-Risk Transformation; Binarized Predictors; Diffusion Index |
| JEL: | C25 C53 E32 E37 |
| Date: | 2025–10–30 |
| URL: | https://d.repec.org/n?u=RePEc:fip:fedpwp:102004 |
| By: | Zhongjun Qu; Wendun Wang; Xiaomeng Zhang |
| Abstract: | A rich set of frequentist model averaging methods has been developed, but their applications have largely been limited to point prediction, as measuring prediction uncertainty in general settings remains an open problem. In this paper we propose prediction intervals for model averaging based on conformal inference. These intervals cover out-of-sample realizations of the outcome variable with a pre-specified probability, providing a way to assess predictive uncertainty beyond point prediction. The framework allows general model misspecification and applies to averaging across multiple models that can be nested, disjoint, overlapping, or any combination thereof, with weights that may depend on the estimation sample. We establish coverage guarantees under two sets of assumptions: exact finite-sample validity under exchangeability, relevant for cross-sectional data, and asymptotic validity under stationarity, relevant for time-series data. We first present a benchmark algorithm and then introduce a locally adaptive refinement and split-sample procedures that broaden applicability. The methods are illustrated with a cross-sectional application to real estate appraisal and a time-series application to equity premium forecasting. |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2510.16224 |
| By: | Abraham Atsiwo |
| Abstract: | This study presents a three-step machine learning framework to predict bubbles in the S&P 500 stock market by combining financial news sentiment with macroeconomic indicators. Building on traditional econometric approaches, the proposed approach predicts bubble formation by integrating textual and quantitative data sources. In the first step, bubble periods in the S&P 500 index are identified using a right-tailed unit root test, a widely recognized real-time bubble detection method. The second step extracts sentiment features from large-scale financial news articles using natural language processing (NLP) techniques, which capture investors' expectations and behavioral patterns. In the final step, ensemble learning methods are applied to predict bubble occurrences based on high sentiment-based and macroeconomic predictors. Model performance is evaluated through k-fold cross-validation and compared against benchmark machine learning algorithms. Empirical results indicate that the proposed three-step ensemble approach significantly improves predictive accuracy and robustness, providing valuable early warning insights for investors, regulators, and policymakers in mitigating systemic financial risks. |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2510.16636 |
| By: | Yaxuan Kong; Yoontae Hwang; Marcus Kaiser; Chris Vryonides; Roel Oomen; Stefan Zohren |
| Abstract: | We introduce M2VN: Multi-Modal Volatility Network, a novel deep learning-based framework for financial volatility forecasting that unifies time series features with unstructured news data. M2VN leverages the representational power of deep neural networks to address two key challenges in this domain: (i) aligning and fusing heterogeneous data modalities, numerical financial data and textual information, and (ii) mitigating look-ahead bias that can undermine the validity of financial models. To achieve this, M2VN combines open-source market features with news embeddings generated by Time Machine GPT, a recently introduced point-in-time LLM, ensuring temporal integrity. An auxiliary alignment loss is introduced to enhance the integration of structured and unstructured data within the deep learning architecture. Extensive experiments demonstrate that M2VN consistently outperforms existing baselines, underscoring its practical value for risk management and financial decision-making in dynamic markets. |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2510.20699 |
| By: | Kronenberg, Philipp |
| Abstract: | This paper presents a weekly GDP indicator for Switzerland, which addresses the limitations of existing economic activity indicators by using alternative highfrequency data created in response to the COVID-19 pandemic. The indicator is derived from a Bayesian mixed-frequency dynamic factor model, which integrates both conventional macroeconomic and alternative high-frequency data at weekly, monthly, and quarterly frequencies. The model extracts business cycle information from a wide range of data frequencies and captures the large and sudden fluctuations during the pandemic by estimating missing observations as latent states through data augmentation, incorporating stochastic volatility in the state equation, and accounting for serial correlation in the measurement errors. An empirical application shows that the indicator accurately approximates weekly GDP growth for Switzerland and provides valuable information on the trajectory of GDP at high frequency, particularly during crisis periods. A pseudo real-time analysis demonstrates high forecast accuracy at short leads and improvements over other GDP indicators for Switzerland. |
| Keywords: | Dynamic Factor Model, High-Frequency Data, Business Cycle Index, Economic Activity Indicator, Covid-19 |
| JEL: | C11 C32 C38 C53 E32 E37 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:esprep:330303 |
| By: | Priscila Espinosa (Department of Applied Economics, Universitat de València, Spain); Priscila Espinosa; Maria Teresa Balaguer-Coll (Department of Finance and Accounting, Universitat Jaume I, Castellón, Spain); José Manuel Pavía (Department of Applied Economics, Universitat de València, Valencia, Spain); Emili Tortosa-Ausina (IVIE, Valencia and IIDL and Department of Economics, Universitat Jaume I, Castellón, Spain) |
| Abstract: | The floods that struck the Valencian region (Spain) in October 2024 illustrate how climate change is intensifying extreme weather events in Mediterranean floodprone areas, challenging regional economic resilience. This disaster, which resulted in numerous fatalities and extensive infrastructure damage, disrupted supply chains across a region already vulnerable due to decades of urban and industrial development in the area. Drawing on regional economic resilience theory and recovery curve methodologies, we present an ex ante framework for rapidly assessing climate disaster impacts on regional economic growth. Our approach combines sectoral recovery dynamics with worker-level impact data to update GDP growth forecasts in real-time, addressing a critical gap in disaster response capabilities for increasingly climate-vulnerable regions. Applied to the Valencia floods, our methodology reveals differential sectoral resilience patterns: while construction demonstrates rapid recovery due to reconstruction demand, agriculture shows prolonged vulnerability reflecting the sector’s exposure to climate risks. Compared to pre-flood forecasts, results indicate economic contractions of up to 0.2 percentage points in 2024 and 2025, followed by a policy-supported rebound adding 0.3 percentage points to growth in 2026. The analysis underscores how government intervention fundamentally shapes postdisaster economic trajectories in flood-prone regions. Beyond providing immediate impact assessment, this framework offers a generalisable tool for enhancing climate resilience planning in Mediterranean and other climatevulnerable territories, enabling policymakers to rapidly adjust economic forecasts and recovery strategies as extreme weather events become more frequent and severe under climate change. |
| Keywords: | macroeconomic forecasting; GDP growth; natural disasters; recovery curves; regional economic resilience |
| JEL: | Q54 R11 H84 C53 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:jau:wpaper:2025/10 |
| By: | Frank, Luis |
| Abstract: | The article proposes a nowcasting model to estimate Argentina's seasonally adjusted Monthly Estimator of Economic Activity (EMAE) using a reduced set of high-frequency economic variables (tax revenue, Portland cement dispatches, automobile sales, and electricity demand), available with a 5–7-day lag, covering data from January 2015 to June 2025. A traditional error correction model (ECM) is compared with a flexible version (FECM) that incorporates time-varying coefficients. The FECM, with $\lambda=1$, outperforms the ECM in accuracy (MAPE of 0.35 versus 1.04). Electricity demand and cement production are the most relevant indicators, while tax revenue has a lower impact. However, it is recommended to retain all variables, as their contribution depends on their joint inclusion. Additionally, a hybrid model that recursively updates parameters is proposed, offering an efficient alternative for real-time economic monitoring. |
| Keywords: | nowcasting, flexible ECM, Argentina, GDP |
| JEL: | C13 C53 |
| Date: | 2025–10–20 |
| URL: | https://d.repec.org/n?u=RePEc:pra:mprapa:126543 |
| By: | Rui Gon\c{c}alves; Vitor Miguel Ribeiro; Roman Chertovskih; Ant\'onio Pedro Aguiar |
| Abstract: | This study presents the implementation of a short-term forecasting system for price movements in exchange markets, using market depth data and a systematic procedure to enable a fully automated trading system. The case study focuses on the UK to Win Horse Racing market during the pre-live stage on the world's leading betting exchange, Betfair. Innovative convolutional attention mechanisms are introduced and applied to multiple recurrent neural networks and bi-dimensional convolutional recurrent neural network layers. Additionally, a novel padding method for convolutional layers is proposed, specifically designed for multivariate time series processing. These innovations are thoroughly detailed, along with their execution process. The proposed architectures follow a standard supervised learning approach, involving model training and subsequent testing on new data, which requires extensive pre-processing and data analysis. The study also presents a complete end-to-end framework for automated feature engineering and market interactions using the developed models in production. The key finding of this research is that all proposed innovations positively impact the performance metrics of the classification task under examination, thereby advancing the current state-of-the-art in convolutional attention mechanisms and padding methods applied to multivariate time series problems. |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2510.16008 |
| By: | Andrew B. Martinez |
| Abstract: | This paper uses an empirical model that incorporates multiple hazards and vulnerabilities to nowcast direct hurricane damages immediately following landfall on the continental United States over the last quarter century using real-time information. I evaluate the performance of the model by constructing a novel database of real-time damage predictions from commercial catastrophe models. I also analyze how official estimates of damage are revised. I find that my empirical model is substantially more accurate than simpler models that only incorporate wind speed and income. While commercial nowcasts are generally accurate, especially when averaging across multiple models, my empirical model is performs best immediately after landfall and when there is a large proportion of uninsured and flood losses. The improved nowcasts are beneficial to many stakeholders including policymakers, insurers, and financial markets. |
| Date: | 2025–10 |
| URL: | https://d.repec.org/n?u=RePEc:gwc:wpaper:2025-006 |
| By: | Marjorie Pampusa; Ashwin Moheeput; Atish Babboo; Rajlukshmee Tengur; Rideema Cunniah; Sharmeen Gariban; Mr. Iaroslav Miller; Shalva Mkhatrishvili; Valeriu Nalban |
| Abstract: | This paper presents the Mauritius Quarterly Projection Model (QPM), the semi-structural analytical tool that underpins the modernized Forecasting and Policy Analysis System of the Bank of Mauritius (BOM). The model is designed to capture the salient features of the domestic economy, including key monetary policy transmission channels and the recently introduced flexible inflation targeting framework. Relative to canonical QPM structures, it also incorporates a parsimonious fiscal block and a labor market block, providing key insights on broader macroeconomic dynamics and enriching the policy advice. The model optimally balances theoretical consistency—evident in coherent shock propagation and policy responses—and empirical reliability, as reflected in its strong in-sample forecasting performance. The practical use of the Mauritius QPM in the context of the BOM’s regular forecasting cycles for the production of baseline projections, counterfactual simulations and alternative scenarios, together with the corresponding model-based economic narratives, make it a critical component of the BOM’s forward-looking monetary policy formulation. |
| Keywords: | Mauritius; Forecasting and Policy Analysis; Quarterly Projection Model; Monetary Policy; Transmission Mechanism |
| Date: | 2025–10–24 |
| URL: | https://d.repec.org/n?u=RePEc:imf:imfwpa:2025/215 |