nep-cmp New Economics Papers
on Computational Economics
Issue of 2022‒11‒07
eighteen papers chosen by
Stan Miles
Thompson Rivers University

  1. DNN-ForwardTesting: A New Trading Strategy Validation using Statistical Timeseries Analysis and Deep Neural Networks By Ivan Letteri; Giuseppe Della Penna; Giovanni De Gasperis; Abeer Dyoub
  2. What makes a satisfying life? Prediction and interpretation with machine-learning algorithms By Andrew E. Clark; Conchita D'Ambrosio; Niccolo Gentile; Alexandre Tkatchenko
  3. Algorithmic Trading Using Continuous Action Space Deep Reinforcement Learning By Naseh Majidi; Mahdi Shamsi; Farokh Marvasti
  4. Sentiment Analysis on Inflation after Covid-19 By Xinyu Li; Zihan Tang
  5. Embedding-based neural network for investment return prediction By Jianlong Zhu; Dan Xian; Fengxiao; Yichen Nie
  6. MetaTrader: An Reinforcement Learning Approach Integrating Diverse Policies for Portfolio Optimization By Hui Niu; Siyuan Li; Jian Li
  7. Human wellbeing and machine learning By Ekaterina Oparina; Caspar Kaiser; Niccolo Gentile; Alexandre Tkatchenko; Andrew E. Clark; Jan-Emmanuel De Neve; Conchita D'Ambrosio
  8. A Survey: Credit Sentiment Score Prediction By A. N. M. Sajedul Alam; Junaid Bin Kibria; Arnob Kumar Dey; Zawad Alam; Shifat Zaman; Motahar Mahtab; Mohammed Julfikar Ali Mahbub; Annajiat Alim Rasel
  9. Neural variance reduction for stochastic differential equations By P. D. Hinds; M. V. Tretyakov
  10. Where Are All the Jobs ? A Machine Learning Approach for High Resolution Urban Employment Prediction inDeveloping Countries By Barzin,Samira; Avner,Paolo; Maruyama Rentschler,Jun Erik; O’Clery,Neave
  11. Quantifying the role of interest rates, the Dollar and Covid in oil prices By Emanuel Kohlscheen
  12. "Non-Crossing Dual Neural Network: Joint Value at Risk and Conditional Tail Expectation estimations with non-crossing conditions". By Xenxo Vidal-Llana; Carlos Salort Sánchez; Vincenzo Coia; Montserrat Guillen
  13. Using Machine Learning to Promote Proactive Human Resources Management: A Case Study By Zakarya Laghzal; Lamya Temnati
  14. Preparation, Practice, and Beliefs : A Machine Learning Approach to Understanding Teacher Effectiveness By Filmer,Deon P.; Nahata,Vatsal; Sabarwal,Shwetlena
  15. Impact of the Rapid Expansion of Renewable Energy on Electricity Market Price: Using machine learning and shapley additive explanation By SHIMOMURA Mizue; KEELEY Alexander Ryota; MATSUMOTO Ken'ichi; TANAKA Kenta; MANAGI Shunsuke
  16. Towards Multi-Agent Reinforcement Learning driven Over-The-Counter Market Simulations By Nelson Vadori; Leo Ardon; Sumitra Ganesh; Thomas Spooner; Selim Amrouni; Jared Vann; Mengda Xu; Zeyu Zheng; Tucker Balch; Manuela Veloso
  17. Impact of the Rapid Expansion of Renewable Energy on Electricity Market Price: Using machine learning and shapley additive explanation By LI Chao; MANAGI Shunsuke
  18. Potential Applications of Quantum Computing for the Insurance Industry By Michael Adam

  1. By: Ivan Letteri; Giuseppe Della Penna; Giovanni De Gasperis; Abeer Dyoub
    Abstract: In general, traders test their trading strategies by applying them on the historical market data (backtesting), and then apply to the future trades the strategy that achieved the maximum profit on such past data. In this paper, we propose a new trading strategy, called DNN-forwardtesting, that determines the strategy to apply by testing it on the possible future predicted by a deep neural network that has been designed to perform stock price forecasts and trained with the market historical data. In order to generate such an historical dataset, we first perform an exploratory data analysis on a set of ten securities and, in particular, analize their volatility through a novel k-means-based procedure. Then, we restrict the dataset to a small number of assets with the same volatility coefficient and use such data to train a deep feed-forward neural network that forecasts the prices for the next 30 days of open stocks market. Finally, our trading system calculates the most effective technical indicator by applying it to the DNNs predictions and uses such indicator to guide its trades. The results confirm that neural networks outperform classical statistical techniques when performing such forecasts, and their predictions allow to select a trading strategy that, when applied to the real future, increases Expectancy, Sharpe, Sortino, and Calmar ratios with respect to the strategy selected through traditional backtesting.
    Date: 2022–10
  2. By: Andrew E. Clark; Conchita D'Ambrosio; Niccolo Gentile; Alexandre Tkatchenko
    Abstract: Machine Learning (ML) methods are increasingly being used across a variety of fields and have led to the discovery of intricate relationships between variables. We here apply ML methods to predict and interpret life satisfaction using data from the UK British Cohort Study. We discuss the application of first Penalized Linear Models and then one non-linear method, Random Forests. We present two key model-agnostic interpretative tools for the latter method: Permutation Importance and Shapley Values. With a parsimonious set of explanatory variables, neither Penalized Linear Models nor Random Forests produce major improvements over the standard Non-penalized Linear Model. However, once we consider a richer set of controls these methods do produce a non-negligible improvement in predictive accuracy. Although marital status, and emotional health continue to be the most important predictors of life satisfaction, as in the existing literature, gender becomes insignificant in the non-linear analysis.
    Keywords: life satisfaction, well-being, machine learning, British cohort study
    Date: 2022–06–07
  3. By: Naseh Majidi; Mahdi Shamsi; Farokh Marvasti
    Abstract: Price movement prediction has always been one of the traders' concerns in financial market trading. In order to increase their profit, they can analyze the historical data and predict the price movement. The large size of the data and complex relations between them lead us to use algorithmic trading and artificial intelligence. This paper aims to offer an approach using Twin-Delayed DDPG (TD3) and the daily close price in order to achieve a trading strategy in the stock and cryptocurrency markets. Unlike previous studies using a discrete action space reinforcement learning algorithm, the TD3 is continuous, offering both position and the number of trading shares. Both the stock (Amazon) and cryptocurrency (Bitcoin) markets are addressed in this research to evaluate the performance of the proposed algorithm. The achieved strategy using the TD3 is compared with some algorithms using technical analysis, reinforcement learning, stochastic, and deterministic strategies through two standard metrics, Return and Sharpe ratio. The results indicate that employing both position and the number of trading shares can improve the performance of a trading system based on the mentioned metrics.
    Date: 2022–10
  4. By: Xinyu Li; Zihan Tang
    Abstract: Based on global tweets from 2017 to 2022, we implement traditional machine learning and deep learning methods to build high-frequency measures of the public's sentiment index towards inflation and analyze the correlation with other online data sources such as google trend and market-oriented inflation index. First, we test out several machine learning approaches using manually labelled tri-grams and finally choose Bert model for our research. Second, we calculate inflation sentiment index through sentiment score of the tweets applying Bert model and analyse the regional and pre/post covid pattern. Lastly, we take other online data sources of inflation into consideration and prove that twitter-based inflation sentiment analysis method has an outstanding capability to predict inflation. The results suggest that Twitter combined with deep learning methods can be a novel and timely method to utilise existing abundant data sources on inflation expectations and provide daily and weekly indicators of consumers' perception on inflation.
    Date: 2022–09
  5. By: Jianlong Zhu; Dan Xian; Fengxiao; Yichen Nie
    Abstract: In addition to being familiar with policies, high investment returns also require extensive knowledge of relevant industry knowledge and news. In addition, it is necessary to leverage relevant theories for investment to make decisions, thereby amplifying investment returns. A effective investment return estimate can feedback the future rate of return of investment behavior. In recent years, deep learning are developing rapidly, and investment return prediction based on deep learning has become an emerging research topic. This paper proposes an embedding-based dual branch approach to predict an investment's return. This approach leverages embedding to encode the investment id into a low-dimensional dense vector, thereby mapping high-dimensional data to a low-dimensional manifold, so that highdimensional features can be represented competitively. In addition, the dual branch model realizes the decoupling of features by separately encoding different information in the two branches. In addition, the swish activation function further improves the model performance. Our approach are validated on the Ubiquant Market Prediction dataset. The results demonstrate the superiority of our approach compared to Xgboost, Lightgbm and Catboost.
    Date: 2022–09
  6. By: Hui Niu; Siyuan Li; Jian Li
    Abstract: Portfolio management is a fundamental problem in finance. It involves periodic reallocations of assets to maximize the expected returns within an appropriate level of risk exposure. Deep reinforcement learning (RL) has been considered a promising approach to solving this problem owing to its strong capability in sequential decision making. However, due to the non-stationary nature of financial markets, applying RL techniques to portfolio optimization remains a challenging problem. Extracting trading knowledge from various expert strategies could be helpful for agents to accommodate the changing markets. In this paper, we propose MetaTrader, a novel two-stage RL-based approach for portfolio management, which learns to integrate diverse trading policies to adapt to various market conditions. In the first stage, MetaTrader incorporates an imitation learning objective into the reinforcement learning framework. Through imitating different expert demonstrations, MetaTrader acquires a set of trading policies with great diversity. In the second stage, MetaTrader learns a meta-policy to recognize the market conditions and decide on the most proper learned policy to follow. We evaluate the proposed approach on three real-world index datasets and compare it to state-of-the-art baselines. The empirical results demonstrate that MetaTrader significantly outperforms those baselines in balancing profits and risks. Furthermore, thorough ablation studies validate the effectiveness of the components in the proposed approach.
    Date: 2022–09
  7. By: Ekaterina Oparina; Caspar Kaiser; Niccolo Gentile; Alexandre Tkatchenko; Andrew E. Clark; Jan-Emmanuel De Neve; Conchita D'Ambrosio
    Abstract: There is a vast literature on the determinants of subjective wellbeing. International organisations and statistical offices are now collecting such survey data at scale. However, standard regression models explain surprisingly little of the variation in wellbeing, limiting our ability to predict it. In response, we here assess the potential of Machine Learning (ML) to help us better understand wellbeing. We analyse wellbeing data on over a million respondents from Germany, the UK, and the United States. In terms of predictive power, our ML approaches perform better than traditional models. Although the size of the improvement is small in absolute terms, it is substantial when compared to that of key variables like health. We moreover find that drastically expanding the set of explanatory variables doubles the predictive power of both OLS and the ML approaches on unseen data. The variables identified as important by our ML algorithms - i.e. material conditions, health, and meaningful social relations - are similar to those that have already been identified in the literature. In that sense, our data-driven ML results validate the findings from conventional approaches.
    Keywords: subjective wellbeing, prediction methods, machine learning
    Date: 2022–07–20
  8. By: A. N. M. Sajedul Alam; Junaid Bin Kibria; Arnob Kumar Dey; Zawad Alam; Shifat Zaman; Motahar Mahtab; Mohammed Julfikar Ali Mahbub; Annajiat Alim Rasel
    Abstract: Manual approvals are still used by banks and other NGOs to approve loans. It takes time and is prone to mistakes because it is controlled by a bank employee. Several fields of machine learning mining technologies have been utilized to enhance various areas of credit rating forecast. A major goal of this research is to look at current sentiment analysis techniques that are being used to generate creditworthiness.
    Date: 2022–09
  9. By: P. D. Hinds; M. V. Tretyakov
    Abstract: Variance reduction techniques are of crucial importance for the efficiency of Monte Carlo simulations in finance applications. We propose the use of neural SDEs, with control variates parameterized by neural networks, in order to learn approximately optimal control variates and hence reduce variance as trajectories of the SDEs are being simulated. We consider SDEs driven by Brownian motion and, more generally, by L\'{e}vy processes including those with infinite activity. For the latter case, we prove optimality conditions for the variance reduction. Several numerical examples from option pricing are presented.
    Date: 2022–09
  10. By: Barzin,Samira; Avner,Paolo; Maruyama Rentschler,Jun Erik; O’Clery,Neave
    Abstract: Globally, both people and economic activity are increasingly concentrated in urban areas. Yet,for the vast majority of developing country cities, little is known about the granular spatial organization of such activity despite its key importance to policy and urbanplanning. This paper adapts a machine learning based algorithm to predict the spatial distribution of employmentusing input data from open access sources such as Open Street Map and Google Earth Engine. The algorithm is trainedon 14 test cities, ranging from Buenos Aires in Argentina to Dakar in Senegal. A spatial adaptation of the random forestalgorithm is used to predict within-city cells in the 14 test cities with extremely high accuracy (R- squared greaterthan 95 percent), and cells in out-of-sample ”unseen” cities with high accuracy (mean R-squared of 63 percent). Thisapproach uses open data to produce high resolution estimates of the distribution of urban employment for cities wheresuch information does not exist, making evidence-based planning more accessible than ever before.
    Date: 2022–03–22
  11. By: Emanuel Kohlscheen
    Abstract: This study analyses oil price movements through the lens of an agnostic random forest model, which is based on 1,000 regression trees. It shows that this highly disciplined, yet flexible computational model reduces in-sample root mean square errors (RMSEs) by 65% relative to a standard linear least square model that uses the same set of 11 explanatory factors. In forecasting exercises the RMSE reduction ranges between 51% and 68%, highlighting the relevance of non-linearities in oil markets. The results underscore the importance of incorporating financial factors into oil models: US interest rates, the dollar and the VIX together account for 39% of the models' RMSE reduction in the post-2010 sample, rising to 48% in the post-2020 sample. If Covid-19 is also considered as a risk factor, these shares become even larger.
    Keywords: dollar, forecasting, machine learning, oil, risk.
    JEL: C40 F30 Q40 Q41 Q47
    Date: 2022–09
  12. By: Xenxo Vidal-Llana (Universitat de Barcelona. Gran Via de les Corts Catalanes 585. 08007 Barcelona, Spain.); Carlos Salort Sánchez (Universitat de Barcelona. Gran Via de les Corts Catalanes 585. 08007 Barcelona, Spain.); Vincenzo Coia (University of British Columbia. West Mall 2329. Vancouver, BC Canada.); Montserrat Guillen (Gran Via de les Corts Catalanes 585. 08007 Barcelona, Spain.)
    Abstract: When datasets present long conditional tails on their response variables, algorithms based on Quantile Regression have been widely used to assess extreme quantile behaviors. Value at Risk (VaR) and Conditional Tail Expectation (CTE) allow the evaluation of extreme events to be easily interpretable. The state-of-the-art methodologies to estimate VaR and CTE controlled by covariates are mainly based on linear quantile regression, and usually do not have in consideration non-crossing conditions across VaRs and their associated CTEs. We implement a non-crossing neural network that estimates both statistics simultaneously, for several quantile levels and ensuring a list of non-crossing conditions. We illustrate our method with a household energy consumption dataset from 2015 for quantile levels 0.9, 0.925, 0.95, 0.975 and 0.99, and show its improvements against a Monotone Composite Quantile Regression Neural Network approximation.
    Keywords: Risk evaluation, Deep learning, Extreme quantiles. JEL classification: C31, C45, C52.
    Date: 2022–10
  13. By: Zakarya Laghzal (ENCG El Jadida, UCD - Université Chouaib Doukkali); Lamya Temnati (UCD - Université Chouaib Doukkali)
    Abstract: Since the Industrial Revolution the function of human resources (HR) has undergone several changes, today it is asked to adopt a long-term and proactive approach which serves to anticipate the needs of the company and the problems likely to impact its productivity and performance in order to implement long-term adaptation actions and be effective in its strategic approach. Otherwise Machine learning has been for some time, a trending technology that has seen massive use in many areas, according to a study conducted by IT decision makers from over 15 different business sectors in the UK, France, Germany and Spain, 87% of the samples have implemented this technology or plan to do so. This technology has allowed companies to improve their processes, increase their competitiveness and help decision-making in several management areas such as finance and marketing.This present work seeks to highlight the potential of this technology to promote proactive management of human resources by using Machine Learning algorithms in the analysis of turnover and the prediction of employees tending to leave their jobs using a IBM corporate database published as part of a competition for the development of an internal model used to identify employees intending to leave their jobs. The results of this study have shown that this technology can play a crucial role in the proactive management of human resources by providing information that makes it possible to pro-act and anticipate actions related to human resources management.
    Abstract: Depuis la révolution industrielle,la fonction des ressources humaines (RH) a subi plusieurs changements, aujourd'hui elle est sollicitée d'adopter une approche à long terme et proactive qui sert à anticiper les besoins de l'entreprise et les problèmes susceptibles d'impacter sa productivité et sa performance afin de mettre en place des actions d'adaptation à long terme et être efficace dans sa démarche stratégique. Par ailleurs la machine Learning (ou auto-apprentissage) est une technologie tendance depuis quelque temps, qui a connu une utilisation massive dans de nombreux domaines. Selon une étude menée par les décideurs IT issus de plus de 15 secteurs d'activités différents dans la Royaume uni, la France, l'Allemand et l'Espagne, 87% de l'échantillon ont implémenté cette technologie ou prévoient de le faire. Cette technologie a permis aux entreprises d'améliorer leur processus, augmenter leur compétitivité et aider à la prise de décision dans plusieurs domaines de gestion tels que la finance et le marketing. Ce présent travail cherche à mettre en exergue le potentiel de cette technologie pour favoriser une gestion proactive des ressources humaines en utilisant les algorithmes du Machine Learning dans l'analyse de turn-over et la prédiction des employés ayant tendance de quitter leurs emplois en se basant sur une base de données de l'entreprise IBM publiée dans le cadre d'une compétition pour le développement d'un modèle interne qui sert à identifier les employés ayant l'intention de quitter leurs emplois. Les résultats de cette étude ont montré que cette technologie peut jouer un rôle crucial dans la gestion proactive des ressources humaines en offrant des informations qui permettent de pro-agir et anticiper les actions liées à la gestion des ressources humaines.
    Keywords: Machine learning,proactive management of human resources,Turn-Over,Turn-over,Machine Learning,La gestion proactive des ressources humaines
    Date: 2022–05–31
  14. By: Filmer,Deon P.; Nahata,Vatsal; Sabarwal,Shwetlena
    Abstract: This paper uses machine learning methods to identify key predictors of teacher effectiveness,proxied by student learning gains linked to a teacher over an academic year. Conditional inference forests and theleast absolute shrinkage and selection operator are applied to matched student-teacher data for math and Kiswahili fromgrades 2 and 3 in 392 schools across Tanzania. These two machine learning methods produce consistent results andoutperform standard ordinary least squares in out-of-sample prediction by 14–24 percent. As in previous research,commonly used teacher covariates like teacher gender, education, experience, and so forth are not good predictorsof teacher effectiveness. Instead, teacher practice (what teachers do, measured through classroom observations andstudent surveys) and teacher beliefs (measured through teacher surveys) emerge as much more important. Overall,teacher covariates are stronger predictors of teacher effectiveness in math than in Kiswahili. Teacher beliefsthat they can help disadvantaged and struggling studentslearn (for math) and they have good relationships within schools (for Kiswahili), teacher practice of providingwritten feedback and reviewing key concepts at the end of class (for math), and spending extra time with strugglingstudents (for Kiswahili) are highly predictive of teacher effectiveness. As is teacher preparation on how to teachfoundational topics (for both Math and Kiswahili). These results demonstrate the need to pay more systematicattention to teacher preparation, practice, and beliefs in teacher research and policy.
    Date: 2021–11–15
  15. By: SHIMOMURA Mizue; KEELEY Alexander Ryota; MATSUMOTO Ken'ichi; TANAKA Kenta; MANAGI Shunsuke
    Abstract: The increase in variable renewable energy (VRE) has brought significant changes in the power system, including a decrease in the average electricity market price owing to the merit order effect (MOE). In this study, we use machine learning and Shapley additive explanation (SHAP) to comprehensively examine the drivers of market price volatility, including the interaction between VRE and demand, fuel prices, and operation capacity in the Japanese electricity market which solar power installation is expanding rapidly. The results of SHAP reveal that there is a large decline effect for market price in solar power during daytime; however, the effect varies depending on the time of day, season, and demand. In addition, the results suggest that the market price increases when demand is high and solar generation is low, such as during summer evenings, which may be because of natural gas generation with higher marginal costs. The study reveals that impact of expanded VRE will not only have the MOE which decreasing average market prices, but may also prompt structural changes in electricity supply, causing market instability and price spikes in the transition process.
    Date: 2022–09
  16. By: Nelson Vadori; Leo Ardon; Sumitra Ganesh; Thomas Spooner; Selim Amrouni; Jared Vann; Mengda Xu; Zeyu Zheng; Tucker Balch; Manuela Veloso
    Abstract: We study a game between liquidity provider and liquidity taker agents interacting in an over-the-counter market, for which the typical example is foreign exchange. We show how a suitable design of parameterized families of reward functions coupled with associated shared policy learning constitutes an efficient solution to this problem. Precisely, we show that our deep-reinforcement-learning-driven agents learn emergent behaviors relative to a wide spectrum of incentives encompassing profit-and-loss, optimal execution and market share, by playing against each other. In particular, we find that liquidity providers naturally learn to balance hedging and skewing as a function of their incentives, where the latter refers to setting their buy and sell prices asymmetrically as a function of their inventory. We further introduce a novel RL-based calibration algorithm which we found performed well at imposing constraints on the game equilibrium, both on toy and real market data.
    Date: 2022–10
  17. By: LI Chao; MANAGI Shunsuke
    Abstract: The positive effects of greenness in living environments on human well-being are known. As a widely used proxy, the nighttime light (NTL) indicates the regional socio-economic status and development level. Higher development levels and economic status are related to more opportunity and higher income, ultimately leading to greater human well-being. However, whether simple increases in greenness and NTL always produce positive results remains inconclusive. Here, we demonstrate the complex relationships between human well-being and greenness and NTL by employing the random forest method. The accuracy of this model is 81.83%, exceeding most previous studies. According to the analysis results, the recommended ranges of greenness and NTL in living environments are 10.91% - 32.99% and 0 – 17.92 nW/cm 2 ・sr , respectively. Moreover, the current average monetary values of greenness and NTL are 3351.96 USD/% and 658.11 USD/(nW/cm 2 ・sr) , respectively. The residential areas are far away from the abundant natural resources, which makes the main population desire more greenness in their living environments. Furthermore, high urban development density, represented by NTL, has caused adverse effects on human well-being in metropolitan areas. Therefore, retaining a moderate development intensity is an effective way to achieve a sustainable society and improve human well-being.
    Date: 2022–09
  18. By: Michael Adam
    Abstract: This paper is the documentation of a pre-study performed by AXA Konzern AG in collaboration with Fraunhofer ITWM to assess the relevance of quantum computing for the insurance industry. Beside a general overview of the status quo of quantum computing technologies, we investigate its applicability for the valuation of insurance contracts as a concrete use case. This valuation is a computationally intensive problem because the lack of closed pricing formulas requires the use of Monte Carlo methods. Therefore current technical capabilities force insurers to apply approximation methods for many subsequent tasks like economic capital calculation or optimization of strategic asset allocations. The business-criticality of these tasks combined with the existence of a quantum algorithm called Amplitude Estimation which promises a quadratic speed-up of Monte Carlo simulation makes this use case obvious. We provide a detailed explanation of Amplitude Estimation and present two quantum circuits which describe insurance-related payoff features in a quantum circuit model. An exemplary circuit that encodes dynamic lapse is evaluated both on a simulator and on real quantum hardware.
    Date: 2022–10

This nep-cmp issue is ©2022 by Stan Miles. It is provided as is without any express or implied warranty. It may be freely redistributed in whole or in part for any purpose. If distributed in part, please include this notice.
General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.