Improvement In Forecast Accuracy Research Articles

Dynamic functional time-series forecasts of foreign exchange implied volatility surfaces

This paper presents static and dynamic versions of univariate, multivariate, and multilevel functional time-series methods to forecast implied volatility surfaces in foreign exchange markets. We find that dynamic functional principal component analysis generally improves out-of-sample forecast accuracy. Specifically, the dynamic univariate functional time-series method shows the greatest improvement. Our models lead to multiple instances of statistically significant improvements in forecast accuracy for daily EUR–USD, EUR–GBP, and EUR–JPY implied volatility surfaces across various maturities, when benchmarked against established methods. A stylised trading strategy is also employed to demonstrate the potential economic benefits of our proposed approach.

The impact of mixed-frequency geopolitical risk on stock market returns

To timely track and evaluate the impact of dynamic geopolitical risk (GPR) shocks, we use mixed-frequency GPR to explain and forecast stock market returns. Our empirical findings confirm that the real-time GPR shock has a lasting negative impact on stock returns. Moreover, the overall model fits, and the reliability of forecasts is sensitive to the selected sample frequency. We show that the mixed-frequency GPR estimates can be considered more significant and robust than those from an analogous common-frequency approach. The out-of-sample return forecasting tests further suggest that the prediction models incorporating mixed-frequency GPR provide substantial and statistically significant improvements in stock return forecast accuracy.

Acceleration of hidden Markov model fitting using graphical processing units, with application to low-frequency tremor classification

Hidden Markov models (HMMs) are general purpose models for time-series data widely used across the sciences because of their flexibility and elegance. Fitting HMMs can often be computationally demanding and time consuming, particularly when the number of hidden states is large or the Markov chain itself is long. Here we introduce a new Graphical Processing Unit (GPU)-based algorithm designed to fit long-chain HMMs, applying our approach to a model for low-frequency tremor events. Even on a modest GPU, our implementation resulted in an increase in speed of several orders of magnitude compared to the standard single processor algorithm. This permitted a full Bayesian inference of uncertainty related to model parameters and forecasts based on posterior predictive distributions. Similar improvements would be expected for HMM models given large number of observations and moderate state spaces (<80 states with current hardware). We discuss the model, general GPU architecture and algorithms and report performance of the method on a tremor dataset from the Shikoku region, Japan. The new approach led to improvements in both computational performance and forecast accuracy, compared to existing frequentist methodology.

Forecast adjustments during post-promotional periods

Sales promotions are a key cause for judgmental adjustments to forecasts, especially in the world of Fast-Moving Consumer Goods (FMCG). Typically, three types of periods are relevant for sales promotions: a normal period as a comparison point, a promotional period, and a post-promotional period. Yet, research on forecasting with sales promotions has focussed specifically on promotional effects and their elevation versus a normal sales period, while generally disregarding adjustments for post-promotional periods. To investigate further into the effects of promotions after their occurrence, we employed an incentivized laboratory experiment focussing specifically on the role of human judgment in sales forecasting after a promotion has occurred. Our study shows that, without guidance, the post-promotional period is considered equal to a normal period in sales numbers. However, a post-promotion dip is commonly observed in reality. We therefore introduced information on the existence of post-promotional dips, in itself (treatment one) or with added average magnitude of such a dip (treatment two). Information provision resulted in increased forecasting accuracy in both treatments, but more so in treatment two. Thus, this study promotes the provision of guiding information. We find that structured support, allows an improvement in judgmental forecasting accuracy after a sales promotion.

A novel hybrid load forecasting framework with intelligent feature engineering and optimization algorithm in smart grid

Real-time, accurate, and stable forecasting plays a vital role in making strategic decisions in the smart grid (SG). This ensures economic savings, effective planning, and reliable and secure power system operation. However, accurate and stable forecasting is challenging due to the uncertain and intermittent electric load behavior. In this context, a rigid forecasting model with assertive stochastic and non-linear behavior capturing abilities is needed. Thus, a support vector regression (SVR) model emerged to cater the non-linear time-series predictions. However, it suffers from computational complexity and hard-to-tune appropriate parameters problem. Due to these problems, forecasting results of SVR are not as accurate as required. To solve such problems, a novel hybrid approach is developed by integrating feature engineering (FE) and modified fire-fly optimization (mFFO) algorithm with SVR, namely FE-SVR-mFFO forecasting framework. FE eliminates redundant and irrelevant features to ensure high computational efficiency. The mFFO algorithm obtains and tunes the SVR model’s appropriate parameters to effectively avoid trapping into local optimum and returns accurate forecasting results. Besides, most literature studies are focused on forecast accuracy improvement. However, the forecasting model’s effectiveness and productiveness are determined equally by its stability and convergence rate. Considering only one objective (accuracy or stability or convergence rate) is inadequate; thus, the proposed FE-SVR-mFFO forecasting framework achieves these three relatively independent objectives simultaneously. To evaluate the effectiveness and applicability of the proposed framework, real half-hourly load data of five states of Australia (New South Wales (NSW), Queensland (QLD), South Australia (SA), Tasmania (TAS), and Victoria (VIC)) are employed as a case study. Experimental results show that the proposed framework outperforms benchmark frameworks like EMD-SVR-PSO, FS-TSFE-CBSSO, VMD-FFT-IOSVR, and DCP-SVM-WO in terms of accuracy, stability, and convergence rate.

The Best of Both Worlds: Forecasting US Equity Market Returns Using a Hybrid Machine Learning–Time Series Approach

Predicting long-term equity market returns is of great importance for investors to strategically allocate their assets. The authors explore machine learning (ML) methods to forecast 10-year-ahead US stock returns and compare the results with the traditional Shiller regression-based forecasts more commonly used in the asset-management industry. The authors find that ML techniques can only modestly improve the forecast accuracy of a traditional Shiller cyclically adjusted price-to-earnings ratio model, and they actually result in worse performance than the vector autoregressive model (VAR)–based two-step approach. The authors then implement this approach with ML techniques and allow for unspecified nonlinear relationships (a hybrid ML-VAR approach). They find about 50% improvement in real-time forecast accuracy for 10-year annualized US stock returns. <b>TOPICS:</b>Security analysis and valuation, big data/machine learning, quantitative methods, statistical methods, performance measurement <b>Key Findings</b> ▪ Applying machine learning (ML) techniques within a robust economic framework such as Davis et al.’s (2018) two-step approach is superior than applying such techniques in isolation (directly forecasting equity returns). ▪ Using the two-step approach, integrating ML with the vector autoregressive model (ML-VAR) to dynamically forecast earning yields reduces dramatically out-of-sample forecast errors, yielding an improvement of about 50% in forecast accuracy for long-horizon U.S. stock market returns. ▪ Among the ML algorithms tested, the ensemble method, which averages all other model forecasts, consistently provides improved predictive power.

Controlling Shareholders’ Equity Pledges, CSR Information Disclosure, and the Accuracy of Analysts’ Earnings Forecasts

We identify and predict circumstances where equity pledging by controlling shareholders reduces the accuracy of analysts’ earnings forecasts in the Chinese market. First, we predict that controlling shareholders opportunistically manipulate firm disclosure to avoid or alleviate margin call costs of pledging, increasing non-diversifiable information risk and impairing the accuracy of analysts’ earnings forecasts. Consistent with predictions, analyst forecast errors are positively associated with the proportion of shares pledged and this association strengthens with the severity of margin call pressure. Results demonstrate that the consequences of share pledging are widespread, with increases in analyst forecast error continuing to occur even in low information asymmetry contexts such as large firms and stable earnings. Second, CSR disclosures and the quality of CSR reporting moderates the relation between analysts’ forecast errors and share pledging by controlling shareholders. The adverse effects of share pledging significantly decline when firms disclose CSR reports with improvements in analyst forecast accuracy in excess of 50 percent on average. Further, equity pledging by controlling shareholders represents a financing transaction and we find that capital stakeholder disclosures in the CSR report are generally most useful in improving analyst forecast accuracy.

A new paradigm for short-range forecasting of severe weather over the Indian region

While destruction associated with floods during the monsoon season and cyclones receives wide attention, the extreme weather in the form of hail, lightning and high winds have also caused widespread devastation over India on a small spatial scale in recent years, especially during the period of March to June. India Meteorological Department (IMD) organized a special forecast improvement campaign during the period March to June of 2017–2019 when the weather forecasts at all offices of IMD were targeted towards an accurate forecast of the extreme form of thunderstorms and their associated impact in short range to nowcasting timescale and their dissemination. The purpose of this study is to quantify the improvement in operational thunderstorm forecast accuracy, in short range (24 h Severe Weather Guidance at subdivisional level) and nowcast scale (nowcasts for individual stations valid for 3 h and issued every three hours) during March to June of 2017 to 2019 and compare the same with the accuracy of previous years. As a result of these efforts, there has been a significant jump in forecast accuracy in the 24-h thunderstorm forecast as well as 3-h nowcast guidance for thunderstorms across the country. Probability of Detection (POD) scores for India as a whole for the 24-h thunderstorm forecast has doubled, while the false alarms (FAR) have remained at the same level as before the start of the forecast campaign. The results indicate that since a thunderstorm is a disastrous weather event, the forecasters generally tend towards spatial over-forecasting. However, this is not uniform across the months. There is systematic lower accuracy in the season transition months of March (winter to summer) and June (dry summer to wet summer). While POD decreases in both March and June, FAR decrease throughout the season. The significant evolution of atmospheric parameters (moisture in particular) as the season changes, favours the maturation of thunderstorms to cumulonimbus stage as the season progresses, and the problem of over forecasting in March becomes a problem of under forecasting of thunderstorms in June. Another reason for false alarms is the unconscious linkage of the thunderstorm with the pattern of rainfall occurrence. However, since all rain-giving clouds over India do not necessarily mature to the cumulonimbus stage, and vice versa, the two are not always related. This is particularly true for the more arid regions of the country, especially in March, where false alarms are higher. The poor density of reporting observatories compared to the mesoscale nature of the events may also increase false alarms, especially over the small maritime islands and the arid regions of the mainland. The accuracy of the All India 3 hourly station level nowcast also improved systematically since 2017. Despite these constraints, the improvements at all scales were possible due to (a) augmentation of observation network by the rapid expansion of Doppler radars network throughout the Indian mainland as well as the installation of a ground-based lightning detection network, (b) numerical modeling products introduced in 2019 to provide short-range forecasts for all aspects of convection; both of which are incorporated into the forecast framework through Standard Operating Procedures (SOP) to standardize the forecast procedure throughout the Indian region. A more objective forecast strategy, using data generated from a denser network of DWRs and crowdsourcing methods as well as more accurate mesoscale models will go a long way to further improve the thunderstorm forecasts.

Backcasting, Nowcasting, and Forecasting Residential Repeat-Sales Returns: Big Data meets Mixed Frequency

The Case-Shiller is the reference repeat-sales index for the U.S. residential real estate market, yet it is released with a two-month delay. We find that incorporating recent information from 71 financial and macro predictors improves backcasts, now-casts, and short-term forecasts of the index returns. Combining individual forecasts with recently-proposed weighting schemes delivers large improvements in forecast accuracy at all horizons. Additional gains obtain with mixed-data sampling methods that exploit the daily frequency of financial variables, reducing the mean square forecast error by as much as 13% compared to a simple autoregressive benchmark. The forecast improvements are largest during economic turmoils, throughout the 2020 COVID-19 pandemic period, and in more populous metropolitan areas.

Accounting for Smoking in Forecasting Mortality and Life Expectancy.

Smoking is one of the main risk factors that has affected human mortality and life expectancy over the past century. Smoking accounts for a large part of the nonlinearities in the growth of life expectancy and of the geographic and sex differences in mortality. As Bongaarts (2006) and Janssen (2018) suggested, accounting for smoking could improve the quality of mortality forecasts due to the predictable nature of the smoking epidemic. We propose a new Bayesian hierarchical model to forecast life expectancy at birth for both sexes and for 69 countries with good data on smoking-related mortality. The main idea is to convert the forecast of the non-smoking life expectancy at birth (i.e., life expectancy at birth removing the smoking effect) into life expectancy forecast through the use of the age-specific smoking attributable fraction (ASSAF). We introduce a new age-cohort model for the ASSAF and a Bayesian hierarchical model for non-smoking life expectancy at birth. The forecast performance of the proposed method is evaluated by out-of-sample validation compared with four other commonly used methods for life expectancy forecasting. Improvements in forecast accuracy and model calibration based on the new method are observed.

A Forecast-Skill-Based Dynamic Pre-Storm Level Control for Reservoir Flood-Control Operation

The design and operation of reservoirs based on conventional flood-limited water levels (FLWL) implicitly adopts the assumption of hydrological stationarity. As such, historical-record-based FLWL may not be the best choice for flood-control operations due to the inherent non-stationarity of rainfall inputs. With maturing flood forecasts, this study focuses on establishing linkage between FLWL and skill of forecast, thus developing a “dynamic pre-storm level” approach for reservoir flood-control operations. The approach utilizes forecast flood magnitude, forecast skill and exceedance probability of forecast error to determine the pre-storm reservoir storage for each flood event. The exceedance probability of forecast error for each incoming flood is used as the reservoir flood control standard instead of the probability of a static return interval flood. This approach is demonstrated in a hypothetical situation in the Three Gorges Reservoir in China. The results show that under zero-forecast-skill conditions, the proposed dynamic pre-storm level matches well with the Three Gorges Reservoir-designed FLWL; and, as the forecast accuracy/skill increase, the proposed approach can make better use of the increased forecast accuracy, thereby maximizing floodwater utilization and reservoir storage. In this way, coupling the new approach with FLWL allows for more efficient and economic day-to-day reservoir operations without adding any flood risk. This study validates the usefulness of dynamic water level control during flood season, considering the improvement of flood forecast accuracy.

A feature reconstruction-based multi-task regression model for cyanobacterial distribution forecasting along the water column

Cyanobacterial water pollution has been threatening the cleaner ecosystem and urban sustainability due to the harmfulness to aquatic ecosystems and human health, which triggers the development of an effective forecasting tool for cyanobacterial blooms. Along the water column, the variations in cyanobacteria cell densities show various distribution patterns and are influenced by multiple environmental factors. Most data-driven models treat cyanobacteria forecasting at a specific water depth as a single task, which fails to share knowledge amongst water depths, resulting in unfavourable forecasting accuracy. This is why an increasing number of nonlinear black-box models have been built for cyanobacteria forecasting but at the expense of model interpretability. This study aims to investigate whether forecasting accuracy and model interpretability can be enhanced by (i) using easily accessible predictors and (ii) developing a feature reconstruction-based multi-task regression model with knowledge sharing amongst water depths. Real-world data from a tropical lake are used to evaluate the effectiveness of the model. For the studied lake, the highest average cyanobacteria cell density occurs at 1.0 m, after which it decreases by over 30% at 5.5 m. The correlation coefficients of time-serial cyanobacteria cell densities between adjacent water depths are greater than 0.95 (P < 0.001). The forecasting results indicate that, compared to single-task nonlinear models, 20.59%, 16.25%, and 22.70% error reductions, measured by the mean square error, are achieved for one-day-ahead, two-day-ahead, and three-day-ahead cyanobacterial bloom forecasts. The accurate bloom and non-bloom signals under the proposed model are up to 94.81% and 98.28%. Based on the proposed model, the relative importance of predictors, the sparsity of regression coefficients, and the covariance relationship of regression coefficients can interpret the model adequately and elucidate the mechanism of knowledge sharing and forecasting accuracy improvement.

Understanding forecast reconciliation

A series of recent papers introduce the concept of Forecast Reconciliation, a process by which independently generated forecasts of a collection of linearly related time series are reconciled via the introduction of accounting aggregations that naturally apply to the data. Aside from its clear presentational and operational virtues, the reconciliation approach generally improves the accuracy of the combined forecasts. In this paper, we examine the mechanisms by which this improvement is generated by re-formulating the reconciliation problem as a combination of direct forecasts of each time series with additional indirect forecasts derived from the linear constraints. Our work establishes a direct link between the nascent Forecast Reconciliation literature and the extensive work on Forecast Combination. In the original hierarchical setting, our approach clarifies for the first time how unbiased forecasts for the entire collection can be generated from base forecasts made at any level of the hierarchy, and we illustrate more generally how simple robust combined forecasts can be generated in any multivariate setting subject to linear constraints. In an empirical example, we show that simple combinations of such forecasts generate significant improvements in forecast accuracy where it matters most: where noise levels are highest and the forecasting task is at its most challenging.

Forecast combination puzzle in the HAR model

The Heterogeneous Autoregressive (HAR) model of Corsi (2009) has become the benchmark model for predicting realized volatility given its simplicity and consistent empirical performance. Many modifications and extensions to the original model have been proposed that often only provide incremental forecast improvements. In this paper, we take a step back and view the HAR model as a forecast combination that combines three predictors: previous day realization (or random walk forecast), previous week average, and previous month average. When apply- ing the Ordinary Least Squares (OLS) to combine the predictors, the HAR model uses optimal weights that are known to be problematic in the forecast combination literature. In fact, the simple average forecast often outperforms the optimal combination in many empirical applications. We investigate the performance of the simple average forecast for the realized volatility of the Dow Jones Industrial Average equity index. We find dramatic improvements in forecast accuracy across all horizons and different time periods. This is the first time the forecast combination puzzle is identified in this context.

Techniques to evaluate the modifier process of National Weather Service flood forecasts

The operational hydrologists of the United States’ National Weather Service (NWS) develop river forecasts as guidance for those at risk of flood damage and update those flood forecasts in real-time as more information becomes available. To do so they rely on experience and intuition to adjust the inputs, state variables, and parameters of hydrologic models. NWS hydrologists use the term “modifiers” to refer collectively to these adjustments. This paper demonstrates the development and application of tools (statistical and graphical) to aid operational hydrologists in the achievement of accurate flood forecasts. Analysis of variance (ANOVA) identifies the relative contribution to forecast uncertainty of each modifier. Heat map visualizations illustrate for operational hydrologists the basin, lead-time, and season-specific effects of their modifiers choices. The tools provide operational hydrologists with insight into which of three commonly applied modifiers (precipitation, soil moisture, and unit hydrograph shape) are most likely to provide improvement in flood forecast accuracy. The tools are demonstrated for a case study of four watersheds within in the Ohio River Valley, using data for flood events sampled from 1990 to 2018. The findings of this research show that operational hydrologists in the Ohio River Basin would do well apply no modifiers in the winter (leaving hydrologic input variables and parameters at baseline values). And though the forecast might be improved by real-time adjustments to the unit hydrograph in summer months, recommendations for particular unit hydrograph modification levels cannot be made with confidence. These findings call into question the modifier adjustment program as a standard process. In the evaluated cases, modifiers do not systematically improve flood forecasts. Improvement may be more efficiently achieved through better calibration of hydrologic models or techniques for reduction of precipitation uncertainty.

The Impact of Data Filtration on the Accuracy of Multiple Time-Domain Forecasting for Photovoltaic Power Plants Generation

The paper reports the forecasting model for multiple time-domain photovoltaic power plants, developed in response to the necessity of bad weather days’ accurate and robust power generation forecasting. We provide a brief description of the piloted short-term forecasting system and place under close scrutiny the main sources of photovoltaic power plants’ generation forecasting errors. The effectiveness of the empirical approach versus unsupervised learning was investigated in application to source data filtration in order to improve the power generation forecasting accuracy for unstable weather conditions. The k-nearest neighbors’ methodology was justified to be optimal for initial data filtration, based on the clusterization results, associated with peculiar weather and seasonal conditions. The photovoltaic power plants’ forecasting accuracy improvement was further investigated for a one hour-ahead time-domain. It was proved that operational forecasting could be implemented based on the results of short-term day-ahead forecast mismatches predictions, which form the basis for multiple time-domain integrated forecasting tools. After a comparison of multiple time series forecasting approaches, operational forecasting was realized based on the second-order autoregression function and applied to short-term forecasting errors with the resulting accuracy of 87%. In the concluding part of the article the authors from the points of view of computational efficiency and scalability proposed the hardware system composition.

Forecasting Swiss exports using Bayesian forecast reconciliation

This paper proposes a novel forecast reconciliation framework using Bayesian state-space methods. It allows for the joint reconciliation at all forecast horizons and uses predictive distributions rather than past variation of forecast errors. Informative priors are used to assign weights to specific predictions, which makes it possible to reconcile forecasts such that they accommodate specific judgmental predictions or managerial decisions. The reconciled forecasts adhere to hierarchical constraints, which facilitates communication and supports aligned decision-making at all levels of complex hierarchical structures. An extensive forecasting study is conducted on a large collection of 13,118 time series that measure Swiss merchandise exports, grouped hierarchically by export destination and product category. We find strong evidence that in addition to producing coherent forecasts, reconciliation also leads to substantial improvements in forecast accuracy. The use of state-space methods is particularly promising for optimal decision-making under conditions with increased model uncertainty and data volatility.

Technical Indicators and Realized Volatility: What is the Source of Predictability?

Forecasting Realized Volatility (RV) is of paramount importance for both academics and practitioners. During recent decades, academic literature has made substantial progress both in terms of methods and predictors under consideration. Despite the popularity of technical indicators, there has been only scarce reference to the effectiveness of this group of predictors in forecasting RV. This paper examines the out-of-sample forecasting performance of technical indicators for S&P500 RV relative to macroeconomic predictors. We contend that these sets of predictors impact volatility at different frequencies. We demonstrate that by generating economically motivated amalgamation forecasts from these predictors taking into account the frequency dimension leads to substantial improvements in forecast accuracy. Technical indicators perform especially strongly for forecasting the short frequency component which complements macroeconomic variables which perform strongly at longer frequencies. We extend our experiment by examining timing effects in the positive performance. We demonstrate that using both sets of predictors leads to consistent and robust outperformance of the benchmark during recessions and expansions, during high and low VIX periods and during different out-of-sample periods. The results indicate that the complementary nature of technical indicators and macroeconomic variables in RV forecasting should be taken into consideration by practitioners and policymakers.

GA-based design of optimal discrete wavelet filters for efficient wind speed forecasting

Wind energy is getting more and more integrated into power grids, giving rise to some challenges because of its inherent intermittent and irregular nature. Wind speed forecasting plays a fundamental role in overcoming such challenging issues and, thus, assisting the power utility manager in optimizing the supply–demand balancing through wind energy generation. This paper suggests a new hybrid scheme WNN, based on discrete wavelet transform (DWT) combined with artificial neural network (ANN), for wind speed forecasting. More specifically, this work aims at designing the most appropriate discrete wavelet filters, best adapted to a one day ahead wind speed forecasting. The optimized DWT filters are intended to effectively preprocess the wind speed time series data in order to enhance the prediction accuracy. Using wind speed data collected from three different locations in the Magherbian region, the obtained simulation results indicate that the proposed approach outperforms other conventional wavelet-based forecasting structures regarding the wind speed prediction precision. Moreover, compared to the standard wavelet ‘db4’ based approach, the optimized wavelet filter-based structure leads to a forecasting accuracy improvement, in terms of RMSE and MAPE index errors, that amounts to nearly 13% and 19%, respectively.

The Value of PV Power Forecast and the Paradox of the “Single Pricing” Scheme: The Italian Case Study

One of the major problem of photovoltaic grid integration is limiting the solar-induced imbalances since these can undermine the security and stability of the electrical system. Improving the forecast accuracy of photovoltaic generation is becoming essential to allow a massive solar penetration. In particular, improving the forecast accuracy of large solar farms’ generation is important both for the producers/traders to minimize the imbalance costs and for the transmission system operators to ensure stability. In this article, we provide a benchmark for the day-ahead forecast accuracy of utility scale photovoltaic (PV) plants in 1325 locations spanning the country of Italy. We then use these benchmarked forecasts and real energy prices to compute the economic value of the forecast accuracy and accuracy improvement in the context of the Italian energy market’s regulatory framework. Through this study, we further point out several important criticisms of the Italian “single pricing” system that brings paradoxical and counterproductive effects regarding the need to reduce the imbalance volumes. Finally, we propose a new market-pricing rule and innovative actions to overcome the undesired effects of the current dispatching regulations.