|
on Forecasting |
| By: | Heinisch, Katja; van Norden, Simon; Wildi, Marc |
| Abstract: | Forecasts that minimize mean squared forecast error (MSE) often exhibit excessive volatility, limiting their practical applicability. We address this accuracy smoothness trade-off by introducing a Multivariate Smooth Sign Accuracy (M-SSA) framework, which extracts smoothed components from leading indicators to enhance the signal-to-noise ratio and control the forecast volatility and timing. Applied to quarterly German GDP growth, our method yields smoothed forecasts that can improve forecasting accuracy, particularly over medium-term horizons. We find that while smoother forecasts tend to lag slightly around turning points, this can be offset by adjusting the forecast horizon. These findings highlight the practicality of the M-SSA framework for both forecasters and policymakers, |
| Keywords: | forecast smoothing, Smooth Sign Accuracy, time-series filtering |
| JEL: | C53 E37 E66 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:iwhdps:335681 |
| By: | Minh Nguyen; Farshid Vahid; Shanika L. Wickramasuriya |
| Abstract: | In hierarchical forecasting, the process of forecast reconciliation transforms a set of base forecasts, which do not satisfy hierarchical aggregation constraints, into coherent forecasts that do satisfy those constraints. Traditional improvements due to reconciliation have been attributed to imposing aggregation constraints. However, when base forecasts are based on different information sets and historical data are available, additional gains may be achieved by combining the information contained in the base forecasts. We propose a new method, called the information combination (IComb) method, which combines the information content of forecasts during the reconciliation process using penalised regression. We provide simulation evidence on the role of information sets, distinct from aggregation constraints, in hierarchical time series forecasting and show that the IComb method produces superior results compared to traditional reconciliation approaches. |
| Keywords: | coherent forecasts, forecast reconciliation, hierarchical time series, information combination, multivariate regression |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:msh:ebswps:2025-11 |
| By: | Le Wang; Boyuan Zhang |
| Abstract: | Forecasting agricultural markets remains a core challenge in business analytics, where nonlinear dynamics, structural breaks, and sparse data have historically limited the gains from increasingly complex econometric and machine learning models. As a result, a long-standing belief in the literature is that simple time-series methods often outperform more advanced alternatives. This paper provides the first systematic evidence that this belief no longer holds in the modern era of time-series foundation models (TSFMs). Using USDA ERS data from 1997-2025, we evaluate 17 forecasting approaches across four model classes, assessing monthly forecasting performance and benchmarking against Market Year Average (MYA) price predictions. This period spans multiple agricultural cycles, major policy changes, and major market disruptions, with substantial cross-commodity price volatility. Focusing on five state-of-the-art TSFMs, we show that zero-shot foundation models (with only historical prices and without any additional covariates) consistently outperform traditional time-series methods, machine learning models, and deep learning architectures trained from scratch. Among them, Time-MoE delivers the largest accuracy gains, improving forecasts by 45% (MAE) overall and by more than 50% for corn and soybeans relative to USDA benchmarks. These results point to a paradigm shift in agricultural forecasting: while earlier generations of advanced models struggled to surpass simple benchmarks, modern pre-trained foundation models achieve substantial and robust improvements, offering a scalable and powerful new framework for highstakes predictive analytics. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.06371 |
| By: | Mohammed Alruqimi; Luca Di Persio |
| Abstract: | This paper investigates an optimal integration of deep learning with financial models for robust asset price forecasting. Specifically, we developed a hybrid framework combining a Long Short-Term Memory (LSTM) network with the Merton-L\'evy jump-diffusion model. To optimise this framework, we employed the Grey Wolf Optimizer (GWO) for the LSTM hyperparameter tuning, and we explored three calibration methods for the Merton-Levy model parameters: Artificial Neural Networks (ANNs), the Marine Predators Algorithm (MPA), and the PyTorch-based TorchSDE library. To evaluate the predictive performance of our hybrid model, we compared it against several benchmark models, including a standard LSTM and an LSTM combined with the Fractional Heston model. This evaluation used three real-world financial datasets: Brent oil prices, the STOXX 600 index, and the IT40 index. Performance was assessed using standard metrics, including Mean Squared Error (MSE), Mean Absolute Error(MAE), Mean Squared Percentage Error (MSPE), and the coefficient of determination (R2). Our experimental results demonstrate that the hybrid model, combining a GWO-optimized LSTM network with the Levy-Merton Jump-Diffusion model calibrated using an ANN, outperformed the base LSTM model and all other models developed in this study. |
| Date: | 2025–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2512.07860 |
| By: | Fredy Pokou (MRE, CRIStAL); Jules Sadefo Kamdem (MRE); Kpante Emmanuel Gnandi (ENAC-LAB) |
| Abstract: | This paper investigates whether structural econometric models can rival machine learning in forecasting energy--macro dynamics while retaining causal interpretability. Using monthly data from 1999 to 2025, we develop a unified framework that integrates Time-Varying Parameter Structural VARs (TVP-SVAR) with advanced dependence structures, including DCC-GARCH, t-copulas, and mixed Clayton--Frank--Gumbel copulas. These models are empirically evaluated against leading machine learning techniques Gaussian Process Regression (GPR), Artificial Neural Networks, Random Forests, and Support Vector Regression across seven macro-financial and energy variables, with Brent crude oil as the central asset. The findings reveal three major insights. First, TVP-SVAR consistently outperforms standard VAR models, confirming structural instability in energy transmission channels. Second, copula-based extensions capture non-linear and tail dependence more effectively than symmetric DCC models, particularly during periods of macroeconomic stress. Third, despite their methodological differences, copula-enhanced econometric models and GPR achieve statistically equivalent predictive accuracy (t-test p = 0.8444). However, only the econometric approach provides interpretable impulse responses, regime shifts, and tail-risk diagnostics. We conclude that machine learning can replicate predictive performance but cannot substitute the explanatory power of structural econometrics. This synthesis offers a pathway where AI accuracy and economic interpretability jointly inform energy policy and risk management. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.19321 |
| By: | Nicolas Hardy; Dimitris Korobilis |
| Abstract: | We revisit macroeconomic time-varying parameter vector autoregressions (TVP-VARs), whose persistent coefficients may adapt too slowly to large, abrupt shifts such as those during major crises. We explore the performance of an adaptively-varying parameter (AVP) VAR that incorporates deterministic adjustments driven by observable exogenous variables, replacing latent state innovations with linear combinations of macroeconomic and financial indicators. This reformulation collapses the state equation into the measurement equation, enabling simple linear estimation of the model. Simulations show that adaptive parameters are substantially more parsimonious than conventional TVPs, effectively disciplining parameter dynamics without sacrificing flexibility. Using macroeconomic datasets for both the U.S. and the euro area, we demonstrate that AVP-VAR consistently improves out-of-sample forecasts, especially during periods of heightened volatility. |
| Keywords: | Bayesian VAR; time-varying parameters; stochastic volatility; macroeconomic forecasting; uncertainty. |
| JEL: | C11 C32 C53 E32 E37 |
| Date: | 2025–12 |
| URL: | https://d.repec.org/n?u=RePEc:gla:glaewp:2025_12 |
| By: | Ficura, Milan; Ibragimov, Rustam; Janda, Karel |
| Abstract: | This working paper investigates the application of modern artificial intelligence techniques to financial time-series forecasting, with a specific focus on crude oil futures markets. Building on advances in deep learning and natural language processing, the study evaluates the predictive performance and economic relevance of several neural network architectures, including univariate and multivariate LSTM, CNN, and N-HiTS models. In addition to statistical accuracy, the models are assessed through trading-based performance metrics and factor regressions to examine the presence of economically and statistically significant returns. The paper contributes to the growing literature on AI-driven asset price forecasting by demonstrating that multivariate deep learning models incorporating additional market information and sentiment measures can improve both forecast precision and trading performance in commodity markets. |
| Keywords: | Artificial intelligence, Deep learning, Oil futures, Time-series forecasting |
| JEL: | C45 Q47 G13 G17 |
| Date: | 2025 |
| URL: | https://d.repec.org/n?u=RePEc:zbw:esprep:335571 |
| By: | Jinjun Liu; Ming-Yen Cheng |
| Abstract: | We study U.S. Treasury yield curve forecasting under distributional uncertainty and recast forecasting as an operations research and managerial decision problem. Rather than minimizing average forecast error, the forecaster selects a decision rule that minimizes worst case expected loss over an ambiguity set of forecast error distributions. To this end, we propose a distributionally robust ensemble forecasting framework that integrates parametric factor models with high dimensional nonparametric machine learning models through adaptive forecast combinations. The framework consists of three machine learning components. First, a rolling window Factor Augmented Dynamic Nelson Siegel model captures level, slope, and curvature dynamics using principal components extracted from economic indicators. Second, Random Forest models capture nonlinear interactions among macro financial drivers and lagged Treasury yields. Third, distributionally robust forecast combination schemes aggregate heterogeneous forecasts under moment uncertainty, penalizing downside tail risk via expected shortfall and stabilizing second moment estimation through ridge regularized covariance matrices. The severity of the worst case criterion is adjustable, allowing the forecaster to regulate the trade off between robustness and statistical efficiency. Using monthly data, we evaluate out of sample forecasts across maturities and horizons from one to twelve months ahead. Adaptive combinations deliver superior performance at short horizons, while Random Forest forecasts dominate at longer horizons. Extensions to global sovereign bond yields confirm the stability and generalizability of the proposed framework. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.04608 |
| By: | Jack Fanshawe; Rumi Masih; Alexander Cameron |
| Abstract: | This paper studies forward-looking stock-stock correlation forecasting for S\&P 500 constituents and evaluates whether learned correlation forecasts can improve graph-based clustering used in basket trading strategies. We cast 10-day ahead correlation prediction in Fisher-z space and train a Temporal-Heterogeneous Graph Neural Network (THGNN) to predict residual deviations from a rolling historical baseline. The architecture combines a Transformer-based temporal encoder, which captures non-stationary, complex, temporal dependencies, with an edge-aware graph attention network that propagates cross-asset information over the equity network. Inputs span daily returns, technicals, sector structure, previous correlations, and macro signals, enabling regime-aware forecasts and attention-based feature and neighbor importance to provide interpretability. Out-of-sample results from 2019-2024 show that the proposed model meaningfully reduces correlation forecasting error relative to rolling-window estimates. When integrated into a graph-based clustering framework, forward-looking correlations produce adaptable and economically meaningfully baskets, particularly during periods of market stress. These findings suggest that improvements in correlation forecasts translate into meaningful gains during portfolio construction tasks. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.04602 |
| By: | Degiannakis, Stavros; Delis, Panagiotis; Filis, George |
| Abstract: | The aim of this study is to develop a forecasting framework for household-level inflation in Greece using domestic, global and energy-related predictors for the period 2009-2022. We show that significant forecasts gains are obtained when models incorporate global conditions and energy prices, relative to our benchmark model, the AR(1). More importantly, though, we find that although the global economic activity, global supply chain pressure and geopolitical risk are important predictors for all households, there are other predictors which demonstrate a household-specific forecast performance. Even more, we show that the energy factors are more important predictors for the low-income households. Overall, these results demonstrate (i) that aggregate inflation forecasts are not representative of the Greek households and (ii) the importance of household-specific inflation forecasting, which could be used as an early warning system that identifies the factors that could drive inflation inequality across the different households. |
| Keywords: | Household-level inflation, inflation inequality, Greece, forecasting, DMA, quantile regression. |
| JEL: | C52 C53 D14 E31 E37 |
| Date: | 2025–10–30 |
| URL: | https://d.repec.org/n?u=RePEc:pra:mprapa:127228 |
| By: | Marc Wildi |
| Abstract: | Forecasting entails a complex estimation challenge, as it requires balancing multiple, often conflicting, priorities and objectives. Traditional forecast optimization criteria typically focus on a single metric -- such as minimizing the mean squared error (MSE) -- which may overlook other important aspects of predictive performance. In response, we introduce a novel approach called the Smooth Sign Accuracy (SSA) framework, which simultaneously considers sign accuracy, MSE, and the frequency of sign changes in the predictor. This addresses a fundamental trade-off (the so-called accuracy-smoothness (AS) dilemma) in prediction. The SSA criterion thus enables the integration of various design objectives related to AS forecasting performance, effectively generalizing conventional MSE-based metrics. We further extend this methodology to accommodate non-stationary, integrated processes, with particular emphasis on controlling the predictor's monotonicity. Moreover, we demonstrate the broad applicability of our approach through an application to, and customization of, established business cycle analysis tools, highlighting its versatility across diverse forecasting contexts. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.06547 |
| By: | Alessio Brini; Ekaterina Seregina |
| Abstract: | We propose Mixed-Panels-Transformer Encoder (MPTE), a novel framework for estimating factor models in panel datasets with mixed frequencies and nonlinear signals. Traditional factor models rely on linear signal extraction and require homogeneous sampling frequencies, limiting their applicability to modern high-dimensional datasets where variables are observed at different temporal resolutions. Our approach leverages Transformer-style attention mechanisms to enable context-aware signal construction through flexible, data-dependent weighting schemes that replace fixed linear combinations with adaptive reweighting based on similarity and relevance. We extend classical principal component analysis (PCA) to accommodate general temporal and cross-sectional attention matrices, allowing the model to learn how to aggregate information across frequencies without manual alignment or pre-specified weights. For linear activation functions, we establish consistency and asymptotic normality of factor and loading estimators, showing that our framework nests Target PCA as a special case while providing efficiency gains through transfer learning across auxiliary datasets. The nonlinear extension uses a Transformer architecture to capture complex hierarchical interactions while preserving the theoretical foundations. In simulations, MPTE demonstrates superior performance in nonlinear environments, and in an empirical application to 13 macroeconomic forecasting targets using a selected set of 48 monthly and quarterly series from the FRED-MD and FRED-QD databases, our method achieves competitive performance against established benchmarks. We further analyze attention patterns and systematically ablate model components to assess variable importance and temporal dependence. The resulting patterns highlight which indicators and horizons are most influential for forecasting. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.16274 |
| By: | Alexander Eliseev (Bank of Russia, Russian Federation) |
| Abstract: | This study focuses on improving the accuracy of nowcasting in DSGE models. We extend one of the general equilibrium models of the Russian economy by incorporating mixed-frequency data. Specifically, we introduce an equation that links a panel of non-modelled high-frequency indicators to observable variables, whose dynamics are determined directly by the model. The out-of-sample pseudo-real-time forecasting procedure demonstrates that incorporating these additional variables enhances the accuracy of Russian GDP nowcasting using the DSGE model. This improvement makes the model’s forecasts comparable in accuracy to state-of-the-art econometric models and superior to univariate models. We also investigate the extent to which fluctuations in high-frequency indicators are associated with macroeconomic factors, as well as the economic shocks driving the explained portion of these fluctuations. While the structural interpretation of non-modelled variables is a potential strength of the model, caution is warranted due to the econometric methodology employed. |
| Keywords: | nowcasting, GDP, DSGE model, mixed frequency data, pseudo real-time forecasting |
| JEL: | C53 C82 E32 E37 |
| Date: | 2025–02 |
| URL: | https://d.repec.org/n?u=RePEc:bkr:wpaper:wps145 |
| By: | Craig S Wright |
| Abstract: | Forecasting accuracy is routinely optimised in financial prediction tasks even though investment and risk-management decisions are executed under transaction costs, market impact, capacity limits, and binding risk constraints. This paper treats forecasting as an econometric input to a constrained decision problem. A predictive distribution induces a decision rule through a utility objective combined with an explicit friction operator consisting of both a cost functional and a feasible-set constraint system. The econometric target becomes minimisation of expected decision loss net of costs rather than minimisation of prediction error. The paper develops a utility-weighted calibration criterion aligned to the decision loss and establishes sufficient conditions under which calibrated predictive distributions weakly dominate uncalibrated alternatives. An empirical study using a pre-committed nested walk-forward protocol on liquid equity index futures confirms the theory: the proposed utility-weighted calibration reduces realised decision loss by over 30\% relative to an uncalibrated baseline ($t$-stat -30.31) for loss differential and improves the Sharpe ratio from -3.62 to -2.29 during a drawdown regime. The mechanism is identified as a structural reduction in the frequency of binding constraints (from 16.0\% to 5.1\%), preventing the "corner solution" failures that characterize overconfident forecasts in high-friction environments. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.07852 |
| By: | Alexander Eliseev; Sergei Seleznev |
| Abstract: | Large language models (LLMs) are a type of machine learning tool that economists have started to apply in their empirical research. One such application is macroeconomic forecasting with backtesting of LLMs, even though they are trained on the same data that is used to estimate their forecasting performance. Can these in-sample accuracy results be extrapolated to the model's out-of-sample performance? To answer this question, we developed a family of prompt sensitivity tests and two members of this family, which we call the fake date tests. These tests aim to detect two types of biases in LLMs' in-sample forecasts: lookahead bias and context bias. According to the empirical results, none of the modern LLMs tested in this study passed our first test, signaling the presence of lookahead bias in their in-sample forecasts. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.07992 |
| By: | Wang Yi; Takashi Hasuike |
| Abstract: | Improvements in return forecast accuracy do not always lead to proportional improvements in portfolio decision quality, especially under realistic trading frictions and constraints. This paper adopts the Smart Predict--then--Optimize (SPO) paradigm for portfolio optimization in real markets, which explicitly aligns the learning objective with downstream portfolio decision quality rather than pointwise prediction accuracy. Within this paradigm, predictive models are trained using an SPO-based surrogate loss that directly reflects the performance of the resulting investment decisions. To preserve interpretability and robustness, we employ linear predictors built on return-based and technical-indicator features and integrate them with portfolio optimization models that incorporate transaction costs, turnover control, and regularization. We evaluate the proposed approach on U.S. ETF data (2015--2025) using a rolling-window backtest with monthly rebalancing. Empirical results show that decision-focused training consistently improves risk-adjusted performance over predict--then--optimize baselines and classical optimization benchmarks, and yields strong robustness during adverse market regimes (e.g., the 2020 COVID-19). These findings highlight the practical value of the Smart Predict--then--Optimize paradigm for portfolio optimization in realistic and non-stationary financial environments. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.04062 |
| By: | Anthony M. Diercks; Jared Dean Katz; Jonathan H. Wright |
| Abstract: | Prediction markets offer a new market-based approach to measuring macroeconomic expectations in real-time. We evaluate the accuracy of prediction market-implied forecasts from Kalshi, the largest federally regulated prediction market overseen by the CFTC. We compare Kalshi with more traditional survey and market-implied forecasts, examine how expectations respond to macroeconomic and financial news, and how policy signals are interpreted by market participants. Our results suggest that Kalshi markets provide a high-frequency, continuously updated, distributionally rich benchmark that is valuable to both researchers and policymakers. |
| JEL: | C5 E3 G1 |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:34702 |
| By: | Danila Ovechkin (Bank of Russia, Russian Federation) |
| Abstract: | This study investigates the hypothesis of a nonlinear relationship between aggregate demand and inflation in the Russian economy. To detect the nonlinear effect, the aggregated Consumer Price Index was decomposed into cyclical (more sensitive to aggregate demand) and acyclical (less sensitive to aggregate demand) components. The decomposition methodology employed in the paper reveals a stable nonlinear link between aggregate demand and inflation. It is shown that the slope of the Phillips curve becomes significantly steeper, i.e., the sensitivity of inflation to economic activity increases, when two conditions are met simultaneously: 1) current general price growth rates exceed long-term inflation expectations; 2) the output gap is positive. Furthermore, it is established that the use of a nonlinear Phillips curve can significantly improve forecast accuracy if a preliminary decomposition of the CPI into cyclical and acyclical components is performed. The forecasting accuracy is asymmetric: inflation forecasts derived from Phillips curves (both linear and nonlinear) demonstrate higher precision during crisis periods. The obtained result proves robust to changes in the trend estimation method, alterations in the nonlinearity condition (using only a positive output gap), the exclusion of sharp CPI changes from the sample, and shifts in the left and right boundaries of the sample. The robustness of the result is also demonstrated with respect to the shock control procedure used in CPI decomposition: even without this procedure, the ability to detect the nonlinear relationship and the improved forecast accuracy (at least at the 9- to 12-month horizon) are preserved. |
| Keywords: | Phillips curve, inflation, business cycle, nonlinearity |
| JEL: | C22 C53 E31 E47 |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:bkr:wpaper:wps161 |
| By: | Zefeng Chen; Darcy Pu |
| Abstract: | Can fully agentic AI nowcast stock returns? We deploy a state-of-the-art Large Language Model to evaluate the attractiveness of each Russell 1000 stock daily, starting from April 2025 when AI web interfaces enabled real-time search. Our data contribution is unique along three dimensions. First, the nowcasting framework is completely out-of-sample and free of look-ahead bias by construction: predictions are collected at the current edge of time, ensuring the AI has no knowledge of future outcomes. Second, this temporal design is irreproducible -- once the information environment passes, it can never be recreated. Third, our framework is 100% agentic: we do not feed the model news, disclosures, or curated text; it autonomously searches the web, filters sources, and synthesises information into quantitative predictions. We find that AI possesses genuine stock selection ability, but only for identifying top winners. Longing the 20 highest-ranked stocks generates a daily Fama-French five-factor plus momentum alpha of 18.4 basis points and an annualised Sharpe ratio of 2.43. Critically, these returns derive from an implementable strategy trading highly liquid Russell 1000 constituents, with transaction costs representing less than 10\% of gross alpha. However, this predictability is highly concentrated: expanding beyond the top tier rapidly dilutes alpha, and bottom-ranked stocks exhibit returns statistically indistinguishable from the market. We hypothesise that this asymmetry reflects online information structure: genuinely positive news generates coherent signals, while negative news is contaminated by strategic corporate obfuscation and social media noise. |
| Date: | 2026–01 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2601.11958 |