Detecting financial statement fraud using dynamic ensemble machine learning

Our study uses Machine learning to develop an advanced fraud detection model that can detect fraudulent firms. We build our model using raw financial and non-financial variables following prior literature. In addition, we introduce the Dynamic Ensemble Selection algorithm to the fraud detection literature, which combines individual classifiers dynamically to make a final prediction. Using several performance evaluation metrics, we find that our model can outperform several machine learning models used in recent studies.     

Insurance fraud detection: A statistically validated network approach

Fraud is a social phenomenon, and fraudsters often collaborate with other fraudsters, taking on different roles. The challenge for insurance companies is to implement claim assessment and improve fraud detection accuracy. We developed an investigative system based on bipartite networks, highlighting the relationships between subjects and accidents or vehicles and accidents. We formalize filtering rules through probability models and test specific methods to assess the existence of communities in extensive networks and propose new alert metrics for suspicious structures. We apply the methodology to a real database—the Italian Antifraud Integrated Archive—and compare the results to out-of-sample fraud scams under investigation by the judicial authorities.     

Detecting accounting fraud in companies reporting under US GAAP through data mining

The accounting fraud detection models developed on financial data prepared under US Generally Accepted Accounting Principles (GAAP) in the current literature achieve significantly weaker performance than models based on financial data prepared under different accounting standards. This study contributes to the US GAAP accounting fraud data mining literature through the attainment of higher model performance than that reported in the prior literature. Financial data from the 10-K forms of 320 fraudulent financial statements (80 fraudulent companies) and 1,200 nonfraudulent financial statements (240 nonfraudulent companies) were collected from the US Security and Exchange Commission. The eight most commonly used data mining techniques were applied to develop prediction models. The results were cross-validated on a testing dataset and then compared according to parameters of accuracy, F-measure, and type I and II errors with existing studies from the US, China, Greece, and Taiwan. As a result, the developed predictive models for accounting fraud achieved performance comparable to those achieved by models built on data from other accounting standards. Moreover, the developed models also significantly outperformed (accuracy 10.5%, F-measure 16.1%, type I error 12.2% and type II error 15.2%) existing studies based on US GAAP financial data. Furthermore, this study provides an extensive literature review encompassing recent accounting fraud theory. It enhances the existing US fraud data mining literature with a performance comparison of studies based on other accounting standards.     

Can alert models for fraud protect the elderly clients of a financial institution?

ABSTRACT

Using account-level transaction data at a major financial institution, we predict the incidence of suspicious activity that can be related to the external financial fraud of its elderly clients. The data consists of over 5 million accounts of clients aged 70 years and older, and over 250 million transactions extending from January 2015 to August 2016. Our main focus is to improve the detection of alerts within a proprietorial transaction monitoring system. Using logistic regression, random forest and support vector machine learning techniques, together with corrections for imbalanced alert samples, we provide a new alert model for the protection of elderly clients at a financial institution, with out-of-sample predictive accuracy. Our findings show the relative influence of client traits and account activity in our select external fraud alert models.     

保险欺诈防范研究与思考

在我国保险欺诈问题日趋严重的背景下,本文通过博弈论理论以及matlab等软件进行回归预测模型的构建,对国际上在反保险欺诈领域领先的国家采取的措施及其经验进行了定量分析。同时,结合我国国情,提出解决我国保险欺诈问题的建议。     

Enhancing claim classification with feature extraction from anomaly-detection-derived routine and peculiarity profiles

Usage-based insurance is becoming the new standard in vehicle insurance; it is therefore relevant to find efficient ways of using insureds' driving data. Applying anomaly detection (AD) to vehicles' trip summaries, we develop a method allowing to derive a “routine” and a “peculiarity” anomaly profile for each vehicle. To this end, AD algorithms are used to compute a routine and a peculiarity anomaly score for each trip a vehicle makes. The former measures the anomaly degree of the trip compared with the other trips made by the concerned vehicle, while the latter measures its anomaly degree compared with trips made by any vehicle. The resulting anomaly scores vectors are used as routine and peculiarity profiles. Features are then extracted from these profiles, for which we investigate the predictive power in the claim classification framework. Using real data, we find that features extracted from the vehicles' peculiarity profile improve the classification.     

Detection of false investment strategies using unsupervised learning methods

In this paper we address the problem of selection bias under multiple testing in the context of investment strategies. We introduce an unsupervised learning algorithm that determines the number of effectively uncorrelated trials carried out in the context of a discovery. This estimate is critical for computing the familywise false positive probability, and for filtering out false investment strategies.     

Detecting Accounting Fraud in Publicly Traded U.S. Firms Using a Machine Learning Approach

We develop a state-of-the-art fraud prediction model using a machine learning approach. We demonstrate the value of combining domain knowledge and machine learning methods in model building. We select our model input based on existing accounting theories, but we differ from prior accounting research by using raw accounting numbers rather than financial ratios. We employ one of the most powerful machine learning methods, ensemble learning, rather than the commonly used method of logistic regression. To assess the performance of fraud prediction models, we introduce a new performance evaluation metric commonly used in ranking problems that is more appropriate for the fraud prediction task. Starting with an identical set of theory-motivated raw accounting numbers, we show that our new fraud prediction model outperforms two benchmark models by a large margin: the Dechow et al. logistic regression model based on financial ratios, and the Cecchini et al. support-vector-machine model with a financial kernel that maps raw accounting numbers into a broader set of ratios.     

Evaluating interpretable machine learning predictions for cryptocurrencies

This study explores various machine learning and deep learning applications on financial data modelling, analysis and prediction processes. The main focus is to test the prediction accuracy of cryptocurrency hourly returns and to explore, analyse and showcase the various interpretability features of the ML models. The study considers the six most dominant cryptocurrencies in the market: Bitcoin, Ethereum, Binance Coin, Cardano, Ripple and Litecoin. The experimental settings explore the formation of the corresponding datasets from technical, fundamental and statistical analysis. The paper compares various existing and enhanced algorithms and explains their results, features and limitations. The algorithms include decision trees, random forests and ensemble methods, SVM, neural networks, single and multiple features N-BEATS, ARIMA and Google AutoML. From experimental results, we see that predicting cryptocurrency returns is possible. However, prediction algorithms may not generalise for different assets and markets over long periods. There is no clear winner that satisfies all requirements, and the main choice of algorithm will be tied to the user needs and provided resources.     

Anti-money laundering and financial fraud detection: A systematic literature review

Money laundering has affected the global economy for many years, and there are several methods of solving it presented in the literature. However, when tackling money laundering and financial fraud together there are few methods for solving them. Thus, this study aims to identify methods for anti-money laundering (AML) and financial fraud detection (FFD). A systematic literature review was performed for analysis and research of the methods used, utilizing the SCOPUS and Web of Science databases. Of the 48 articles that aligned with the research theme, 20 used quantitative methods for AML and FFD solution, 13 were literature reviews, 7 used qualitative methods, and 8 used mixed methods. This study contributes by presenting a systematic literature review that fills two research gaps: lack of studies on AML and FFD, and the methods used to solve them. This will assist researchers in identifying gaps and related research.     

大数据背景下的农险反欺诈检测:国际经验与技术选择

中国农业保险市场中欺诈骗保等违法行为不容忽视,亟需运用数据挖掘技术提高农业保险发展质量。首先,本文对反欺诈检测的常用方法进行了梳理,包括异常值检测、聚类法、线性回归法、社会关系网络分析法等方法。其次,本文总结美国运用数据挖掘技术开展农业保险反欺诈检测的基本经验。美国利用以政府主导、研究机构参与的模式,开发出多种欺诈检测项目,为美国农业保险节约巨额资金。再者,基于国际经验,本文提出适用于中国农业保险反欺诈检测的相关性异常值检测法、合谋关系检测法和机器学习法。最后,为进一步推动数据挖掘技术在中国农业保险反欺诈检测中的运用,本文提出建立农业保险大数据库、建立数据挖掘合作平台、建立常态化的数据利用机制、以及培育和激励数据挖掘人才等建议。     

Forecasting high-frequency excess stock returns via data analytics and machine learning

Borsa Istanbul introduced data analytics to present additional information about its market conditions. We examine whether this product can be utilized via various machine learning methods to predict intraday excess returns. Accordingly, these analytics provide significant prediction ratios above 50% with ideal profit ratios that can reach up to 33%. Among all the methods considered, XGBoost (logistic regression) performs better in predicting excess returns in the long-term analysis (short-term analysis). Results provide evidence for the benefits of both the analytics and the machine learning methods and raise further discussion on the semistrong market efficiency.     

Corporate governance performance ratings with machine learning

We use machine learning with a cross-sectional research design to predict governance controversies and to develop a measure of the governance component of the environmental, social, governance (ESG) metrics. Based on comprehensive governance data from 2,517 companies over a period of 10 years and investigating nine machine-learning algorithms, we find that governance controversies can be predicted with high predictive performance. Our proposed governance rating methodology has two unique advantages compared with traditional ESG ratings: it rates companies' compliance with governance responsibilities and it has predictive validity. Our study demonstrates a solution to what is likely the greatest challenge for the finance industry today: how to assess a company's sustainability with validity and accuracy. Prior to this study, the ESG rating industry and the literature have not provided evidence that widely adopted governance ratings are valid. This study describes the only methodology for developing governance performance ratings based on companies' compliance with governance responsibilities and for which there is evidence of predictive validity.     

Stock price prediction using DEEP learning algorithm and its comparison with machine learning algorithms

Security indices are the main tools for evaluation of the status of financial markets. Moreover, a main part of the economy of any country is constituted of investment in stock markets. Therefore, investors could maximize the return of investment if it becomes possible to predict the future trend of stock market with appropriate methods. The nonlinearity and nonstationarity of financial series make their prediction complicated. This study seeks to evaluate the prediction power of machine‐learning models in a stock market. The data used in this study include the daily close price data of iShares MSCI United Kingdom exchange‐traded fund from January 2015 to June 2018. The prediction process is done through four models of machine‐learning algorithms. The results indicate that the deep learning method is better in prediction than the other methods, and the support vector regression method is in the next rank with respect to neural network and random forest methods with less error.     

Stock investment strategy combining earnings power index and machine learning

We propose an intermediate-term stock investment strategy based on fundamental analysis and machine learning. The approach uses predictors from the Earnings Power Index (EPI) as input variables derived from cross-sectional and time-series data from a company’s financial statements. The analytical methods of machine learning allow us to validate the link between financial factors and excess returns directly. We then select stocks for which returns are likely to increase at the time of the next disclosed financial statement. To verify the proposed approach’s usefulness, we use company data listed publicly on the Korean stock market from 2013 to 2019. We examine the profitability of trading strategy based on ten machine-learning techniques by forming long, short, and hedge portfolios with three different measures. As a result, most portfolios, including EPI-related variables, present positive returns regardless of the period. Especially, the neural network of the two layers with sigmoid function presents the best performance for the period of 3 months and 6 months, respectively. Our results show that incorporating machine learning is useful for mid-term stock investment. Further research into the possible convergence of financial statement analysis and machine-learning techniques is warranted.     

Machine learning methods in finance: Recent applications and prospects

We study how researchers can apply machine learning (ML) methods in finance. We first establish that the two major categories of ML (supervised and unsupervised learning) address fundamentally different problems than traditional econometric approaches. Then, we review the current state of research on ML in finance and identify three archetypes of applications: (i) the construction of superior and novel measures, (ii) the reduction of prediction error, and (iii) the extension of the standard econometric toolset. With this taxonomy, we give an outlook on potential future directions for both researchers and practitioners. Our results suggest many benefits of ML methods compared to traditional approaches and indicate that ML holds great potential for future research in finance.     

Enhancing the government accounting information systems using social media information: An application of text mining and machine learning

This study demonstrates a way of bringing an innovative data source, social media information, to the government accounting information systems to support accountability to stakeholders and managerial decision-making. Future accounting and auditing processes will heavily rely on multiple forms of exogenous data. As an example of the techniques that could be used to generate this needed information, the study applies text mining techniques and machine learning algorithms to Twitter data. The information is developed as an alternative performance measure for NYC street cleanliness. It utilizes Naïve Bayes, Random Forest, and XGBoost to classify the tweets, illustrates how to use the sampling method to solve the imbalanced class distribution issue, and uses VADER sentiment to derive the public opinion about street cleanliness. This study also extends the research to another social media platform, Facebook, and finds that the incremental value is different between the two social media platforms. This data can then be linked to government accounting information systems to evaluate costs and provide a better understanding of the efficiency and effectiveness of operations.     

Drivers of economic and financial integration: A machine learning approach

We propose a new approach to identifying drivers of economic and financial integration, separately, and across emerging and developed countries. Our advanced machine learning technique allows for nonlinear relationships, corrects for over-fitting, and is less prone to noise. It also can tackle a large number of highly correlated explanatory variables and controls for multicollinearity. Results suggest that general economic growth, increasing international trade, and contained population growth have helped emerging countries catch up to the level of the economic integration of developed countries. However, slow financial development and a high level of investment riskiness have hindered the speed of emerging countries’ financial integration. Furthermore, the results suggest that integration is a gradual process and is not driven by cyclical or transitory events.     

Analysing and forecasting co-movement between innovative and traditional financial assets based on complex network and machine learning

We study the co-movement between innovative financial assets (i.e., FinTech-related stocks, green bonds and cryptocurrencies) and traditional assets. We construct a co-movement mode transmission network and discuss the network topology during the pre-COVID-19 and COVID-19 periods. We extract network topology information to predict the co-movement mode by machine learning algorithms. We further propose dynamic trading strategies based on the co-movement mode prediction. The empirical results show that (i) the evolution of co-movement is dominated by some key modes, and the mode transmission relies on intermediate modes and shows certain periodicity; (ii) the co-movement relationships are influenced by the ongoing COVID-19 outbreak; and (iii) the novel approach, which combines complex network and machine learning, is superior in co-movement mode prediction and can effectively bring diversification benefits. Our work provides valuable insights for market participants.