Companies survey their customers to measure their satisfaction levels with the company and its services. The received responses are crucial as they allow companies to assess their respective performances and find ways to make needed improvements. This study focuses on the non-systematic bias that arises when customers assign numerical values in ordinal surveys. Using real customer satisfaction survey data of a large retail bank, we show that the common practice of segmenting ordinal survey responses into uneven segments limit the value that can be extracted from the data. We then show that it is possible to assess the magnitude of the irreducible error under simple assumptions, even in real surveys, and place the achievable modeling goal in perspective. We finish the study by suggesting that a thoughtful survey design, which uses either a careful binning strategy or proper calibration, can reduce the compounding non-systematic error even in elaborated ordinal surveys. A possible application of the calibration method we propose is efficiently conducting targeted surveys using active learning.
{"title":"What can be learned from satisfaction assessments?","authors":"N. Cohen, Simran Lamba, P. Reddy","doi":"10.1145/3383455.3422535","DOIUrl":"https://doi.org/10.1145/3383455.3422535","url":null,"abstract":"Companies survey their customers to measure their satisfaction levels with the company and its services. The received responses are crucial as they allow companies to assess their respective performances and find ways to make needed improvements. This study focuses on the non-systematic bias that arises when customers assign numerical values in ordinal surveys. Using real customer satisfaction survey data of a large retail bank, we show that the common practice of segmenting ordinal survey responses into uneven segments limit the value that can be extracted from the data. We then show that it is possible to assess the magnitude of the irreducible error under simple assumptions, even in real surveys, and place the achievable modeling goal in perspective. We finish the study by suggesting that a thoughtful survey design, which uses either a careful binning strategy or proper calibration, can reduce the compounding non-systematic error even in elaborated ordinal surveys. A possible application of the calibration method we propose is efficiently conducting targeted surveys using active learning.","PeriodicalId":447950,"journal":{"name":"Proceedings of the First ACM International Conference on AI in Finance","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129339557","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Neeti Pokhriyal, A. Dara, B. Valentino, Soroush Vosoughi
Researchers have used social media data to estimate various macroeconomic indicators about public behaviors, mostly as a way to reduce surveying costs. One of the most widely cited economic indicator is consumer confidence index (CCI). Numerous studies in the past have focused on using social media, especially Twitter data, to predict CCI. However, the strong correlations disappeared when those models were tested with newer data according to a recent comprehensive survey. In this work, we revisit this problem of assessing the true potential of using social media data to measure CCI, by proposing a robust non-parametric Bayesian modeling framework grounded in Gaussian Process Regression (which provides both an estimate and an uncertainty associated with it). Integral to our framework is a principled experimentation methodology that demonstrates how digital data can be employed to reduce the frequency of surveys, and thus periodic polling would be needed only to calibrate our model. Via extensive experimentation we show how the choice of different micro-decisions, such as the smoothing interval, various types of lags etc. have an important bearing on the results. By using decadal data (2008--2019) from Reddit, we show that both monthly and daily estimates of CCI can, indeed, be reliably estimated at least several months in advance, and that our model estimates are far superior to those generated by the existing methods.
{"title":"Social media data reveals signal for public consumer perceptions","authors":"Neeti Pokhriyal, A. Dara, B. Valentino, Soroush Vosoughi","doi":"10.1145/3383455.3422556","DOIUrl":"https://doi.org/10.1145/3383455.3422556","url":null,"abstract":"Researchers have used social media data to estimate various macroeconomic indicators about public behaviors, mostly as a way to reduce surveying costs. One of the most widely cited economic indicator is consumer confidence index (CCI). Numerous studies in the past have focused on using social media, especially Twitter data, to predict CCI. However, the strong correlations disappeared when those models were tested with newer data according to a recent comprehensive survey. In this work, we revisit this problem of assessing the true potential of using social media data to measure CCI, by proposing a robust non-parametric Bayesian modeling framework grounded in Gaussian Process Regression (which provides both an estimate and an uncertainty associated with it). Integral to our framework is a principled experimentation methodology that demonstrates how digital data can be employed to reduce the frequency of surveys, and thus periodic polling would be needed only to calibrate our model. Via extensive experimentation we show how the choice of different micro-decisions, such as the smoothing interval, various types of lags etc. have an important bearing on the results. By using decadal data (2008--2019) from Reddit, we show that both monthly and daily estimates of CCI can, indeed, be reliably estimated at least several months in advance, and that our model estimates are far superior to those generated by the existing methods.","PeriodicalId":447950,"journal":{"name":"Proceedings of the First ACM International Conference on AI in Finance","volume":"110 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124446802","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Table extraction has long been a pervasive problem in financial services. This is more challenging in the image domain, where content is locked behind cumbersome pixel format. Luckily, advances in deep learning for image segmentation, OCR, and sequence modeling provides the necessary heavy lifting to achieve impressive results. This paper presents an end-to-end pipeline for identifying, extracting and transcribing tabular content in image documents, while retaining the original spatial relations with high fidelity.
{"title":"Financial table extraction in image documents","authors":"W. Watson, Bo Liu","doi":"10.1145/3383455.3422520","DOIUrl":"https://doi.org/10.1145/3383455.3422520","url":null,"abstract":"Table extraction has long been a pervasive problem in financial services. This is more challenging in the image domain, where content is locked behind cumbersome pixel format. Luckily, advances in deep learning for image segmentation, OCR, and sequence modeling provides the necessary heavy lifting to achieve impressive results. This paper presents an end-to-end pipeline for identifying, extracting and transcribing tabular content in image documents, while retaining the original spatial relations with high fidelity.","PeriodicalId":447950,"journal":{"name":"Proceedings of the First ACM International Conference on AI in Finance","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115838820","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Renato A. de Oliveira, Heitor Soares Ramos Filho, D. H. Dalip, A. Pereira
Automated stock trading is now the de-facto way that investors have chosen to obtain high profits in the stock market while keeping risk under control. One of the approaches is to create agents employing Reinforcement Learning (RL) algorithms to learn and decide whether or not to operate in the market in order to achieve maximum profit. Automated financial trading systems can learn how to trade optimally while interacting with the market pretty much like a human investor learns how to trade. In this research, a simple RL agent was implemented using the SARSA algorithm. Next, it was tested against 10 stocks from Brazilian stock market B3 (Bolsa, Brasil, Balcão). Results from experiments showed that the agent was able to provide high profits with less risk when compared to a supervised learning agent that used a LSTM neural network.
自动化股票交易现在是投资者在控制风险的同时获得高额利润的一种事实上的方式。其中一种方法是创建使用强化学习(RL)算法的智能体，以学习和决定是否在市场中操作以获得最大利润。自动金融交易系统可以学习如何在与市场互动的同时进行最佳交易，就像人类投资者学习如何交易一样。在本研究中，使用SARSA算法实现了一个简单的RL代理。接下来，对巴西B3股票市场(Bolsa, Brasil, balc)的10只股票进行测试。实验结果表明，与使用LSTM神经网络的监督学习代理相比，该代理能够提供高利润和低风险。
{"title":"A tabular sarsa-based stock market agent","authors":"Renato A. de Oliveira, Heitor Soares Ramos Filho, D. H. Dalip, A. Pereira","doi":"10.1145/3383455.3422559","DOIUrl":"https://doi.org/10.1145/3383455.3422559","url":null,"abstract":"Automated stock trading is now the de-facto way that investors have chosen to obtain high profits in the stock market while keeping risk under control. One of the approaches is to create agents employing Reinforcement Learning (RL) algorithms to learn and decide whether or not to operate in the market in order to achieve maximum profit. Automated financial trading systems can learn how to trade optimally while interacting with the market pretty much like a human investor learns how to trade. In this research, a simple RL agent was implemented using the SARSA algorithm. Next, it was tested against 10 stocks from Brazilian stock market B3 (Bolsa, Brasil, Balcão). Results from experiments showed that the agent was able to provide high profits with less risk when compared to a supervised learning agent that used a LSTM neural network.","PeriodicalId":447950,"journal":{"name":"Proceedings of the First ACM International Conference on AI in Finance","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131715976","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
T. Balch, Benjamin E. Diamond, Antigoni Polychroniadou
Inventory matching is a process by which a broker or bank pairs buyers and sellers, without revealing their respective orders in a public exchange. Banks often undertake to match their clients, so that these clients can trade securities without incurring adverse price movements. If a bank finds matches between clients, it may execute them at reduced rates; if no matches are found, the clients must trade in a public market, which introduces costs for both parties. This problem is distinct from that solved by dark pools or public exchanges, which implement Continuous Double Auctions (CDAs). CDAs incorporate both price and volume. Inventory matching incorporates volume alone, and extracts price from an external source (such as a public market). As it is currently conducted, inventory matching requires that clients share their intentions to buy or sell certain securities---along with the sizes of their positions---with the bank. Clients worry that if this information were to "leak" in some way, other market participants could become aware of their intentions, and cause the price to move adversely against them before they trade. A solution to this problem promises to enable more clients to match their orders more efficiently---with reduced market impact---while also eliminating the risk of information leakage. We present a cryptographic approach to multi-client inventory matching, which preserves the privacy of clients. Our central tool is threshold fully homomorphic encryption; in particular, we introduce an efficient, fully-homomorphic integer library which combines GPU-level parallelism with insights from digital circuit design. Our solution is also post-quantum secure. We report on an implementation of our protocol, and describe its performance.
{"title":"SecretMatch","authors":"T. Balch, Benjamin E. Diamond, Antigoni Polychroniadou","doi":"10.1145/3383455.3422569","DOIUrl":"https://doi.org/10.1145/3383455.3422569","url":null,"abstract":"Inventory matching is a process by which a broker or bank pairs buyers and sellers, without revealing their respective orders in a public exchange. Banks often undertake to match their clients, so that these clients can trade securities without incurring adverse price movements. If a bank finds matches between clients, it may execute them at reduced rates; if no matches are found, the clients must trade in a public market, which introduces costs for both parties. This problem is distinct from that solved by dark pools or public exchanges, which implement Continuous Double Auctions (CDAs). CDAs incorporate both price and volume. Inventory matching incorporates volume alone, and extracts price from an external source (such as a public market). As it is currently conducted, inventory matching requires that clients share their intentions to buy or sell certain securities---along with the sizes of their positions---with the bank. Clients worry that if this information were to \"leak\" in some way, other market participants could become aware of their intentions, and cause the price to move adversely against them before they trade. A solution to this problem promises to enable more clients to match their orders more efficiently---with reduced market impact---while also eliminating the risk of information leakage. We present a cryptographic approach to multi-client inventory matching, which preserves the privacy of clients. Our central tool is threshold fully homomorphic encryption; in particular, we introduce an efficient, fully-homomorphic integer library which combines GPU-level parallelism with insights from digital circuit design. Our solution is also post-quantum secure. We report on an implementation of our protocol, and describe its performance.","PeriodicalId":447950,"journal":{"name":"Proceedings of the First ACM International Conference on AI in Finance","volume":"37 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130839844","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Recently, most stock exchanges in the U.S. employ maker-taker fees, in which an exchange pays rebates to traders placing orders in the order book and charges fees to traders taking orders from the order book. Maker-taker fees encourage traders to place many orders that provide market liquidity to the exchange. However, it is not clear how maker-taker fees affect the total cost of a taking order, including all the charged fees and the market impact. In this study, we investigated the effect of maker-taker fees on the total cost of a taking order with our artificial market model, which is an agent-based model for financial markets. We found that maker-taker fees encourage market efficiency but increase the total costs of taking orders.
{"title":"Analysis of the impact of maker-taker fees on the stock market using agent-based simulation","authors":"Isao Yagi, Mahiro Hoshino, T. Mizuta","doi":"10.1145/3383455.3422523","DOIUrl":"https://doi.org/10.1145/3383455.3422523","url":null,"abstract":"Recently, most stock exchanges in the U.S. employ maker-taker fees, in which an exchange pays rebates to traders placing orders in the order book and charges fees to traders taking orders from the order book. Maker-taker fees encourage traders to place many orders that provide market liquidity to the exchange. However, it is not clear how maker-taker fees affect the total cost of a taking order, including all the charged fees and the market impact. In this study, we investigated the effect of maker-taker fees on the total cost of a taking order with our artificial market model, which is an agent-based model for financial markets. We found that maker-taker fees encourage market efficiency but increase the total costs of taking orders.","PeriodicalId":447950,"journal":{"name":"Proceedings of the First ACM International Conference on AI in Finance","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133373273","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
L. Bisi, P. Liotet, Luca Sabbioni, Gianmarco Reho, N. Montali, Marcello Restelli, Cristiana Corno
Automated Trading Systems' impact on financial markets is ever growing, particularly on the intraday Foreign Exchange market. Historically, the FX trading systems are based on advanced statistical methods and technical analysis able to extract trading signals from financial data. In this work, we explore how to find a trading strategy via Reinforcement Learning by means of a state-of-the-art batch algorithm, Fitted Q-Iteration. Furthermore, we include a Multi-Objective formulation of the problem to keep the risk of noisy profits under control. We show that the algorithm is able to detect favorable temporal patterns, which are used by the agent to maximize the return. Finally, we show that as risk aversion increases, the resulting policies become smoother, as the portfolio positions are held for longer periods.
{"title":"Foreign exchange trading: a risk-averse batch reinforcement learning approach","authors":"L. Bisi, P. Liotet, Luca Sabbioni, Gianmarco Reho, N. Montali, Marcello Restelli, Cristiana Corno","doi":"10.1145/3383455.3422571","DOIUrl":"https://doi.org/10.1145/3383455.3422571","url":null,"abstract":"Automated Trading Systems' impact on financial markets is ever growing, particularly on the intraday Foreign Exchange market. Historically, the FX trading systems are based on advanced statistical methods and technical analysis able to extract trading signals from financial data. In this work, we explore how to find a trading strategy via Reinforcement Learning by means of a state-of-the-art batch algorithm, Fitted Q-Iteration. Furthermore, we include a Multi-Objective formulation of the problem to keep the risk of noisy profits under control. We show that the algorithm is able to detect favorable temporal patterns, which are used by the agent to maximize the return. Finally, we show that as risk aversion increases, the resulting policies become smoother, as the portfolio positions are held for longer periods.","PeriodicalId":447950,"journal":{"name":"Proceedings of the First ACM International Conference on AI in Finance","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130457397","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We examine a variety of graphical models to construct optimal portfolios. Graphical models such as PCA-KMeans, autoencoders, dynamic clustering, and structural learning can capture the time varying patterns in the covariance matrix and allow the creation of an optimal and robust portfolio. We compared the resulting portfolios from the different models with baseline methods. In many cases our graphical strategies generated steadily increasing returns with low risk and outgrew the S&P 500 index. This work suggests that graphical models can effectively learn the temporal dependencies in time series data and are proved useful in asset management.
{"title":"Graphical models for financial time series and portfolio selection","authors":"Ni Zhan, Yijia Sun, Aman Jakhar, Hening Liu","doi":"10.1145/3383455.3422566","DOIUrl":"https://doi.org/10.1145/3383455.3422566","url":null,"abstract":"We examine a variety of graphical models to construct optimal portfolios. Graphical models such as PCA-KMeans, autoencoders, dynamic clustering, and structural learning can capture the time varying patterns in the covariance matrix and allow the creation of an optimal and robust portfolio. We compared the resulting portfolios from the different models with baseline methods. In many cases our graphical strategies generated steadily increasing returns with low risk and outgrew the S&P 500 index. This work suggests that graphical models can effectively learn the temporal dependencies in time series data and are proved useful in asset management.","PeriodicalId":447950,"journal":{"name":"Proceedings of the First ACM International Conference on AI in Finance","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128722761","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Youngjin Park, Deokjun Eom, Byoung Ki Seo, Jaesik Choi
Forecasting with multivariate time series, which aims to predict future values given previous and current several univariate time series data, has been studied for decades, with one example being ARIMA. Because it is difficult to measure the extent to which noise is mixed with informative signals within rapidly fluctuating financial time series data, designing a good predictive model is not a simple task. Recently, many researchers have become interested in recurrent neural networks and attention-based neural networks, applying them in financial forecasting. There have been many attempts to utilize these methods for the capturing of long-term temporal dependencies and to select more important features in multivariate time series data in order to make accurate predictions. In this paper, we propose a new prediction framework based on deep neural networks and a trend filtering, which converts noisy time series data into a piecewise linear fashion. We reveal that the predictive performance of deep temporal neural networks improves when the training data is temporally processed by a trend filtering. To verify the effect of our framework, three deep temporal neural networks, state of the art models for predictions in time series finance data, are used and compared with models that contain trend filtering as an input feature. Extensive experiments on real-world multivariate time series data show that the proposed method is effective and significantly better than existing baseline methods.
{"title":"Improved predictive deep temporal neural networks with trend filtering","authors":"Youngjin Park, Deokjun Eom, Byoung Ki Seo, Jaesik Choi","doi":"10.1145/3383455.3422565","DOIUrl":"https://doi.org/10.1145/3383455.3422565","url":null,"abstract":"Forecasting with multivariate time series, which aims to predict future values given previous and current several univariate time series data, has been studied for decades, with one example being ARIMA. Because it is difficult to measure the extent to which noise is mixed with informative signals within rapidly fluctuating financial time series data, designing a good predictive model is not a simple task. Recently, many researchers have become interested in recurrent neural networks and attention-based neural networks, applying them in financial forecasting. There have been many attempts to utilize these methods for the capturing of long-term temporal dependencies and to select more important features in multivariate time series data in order to make accurate predictions. In this paper, we propose a new prediction framework based on deep neural networks and a trend filtering, which converts noisy time series data into a piecewise linear fashion. We reveal that the predictive performance of deep temporal neural networks improves when the training data is temporally processed by a trend filtering. To verify the effect of our framework, three deep temporal neural networks, state of the art models for predictions in time series finance data, are used and compared with models that contain trend filtering as an input feature. Extensive experiments on real-world multivariate time series data show that the proposed method is effective and significantly better than existing baseline methods.","PeriodicalId":447950,"journal":{"name":"Proceedings of the First ACM International Conference on AI in Finance","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127241487","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Predicting market volatility is a critical issue in financial market research and practice. Moreover, in natural language processing, how to effectively leverage long- and short-term event sequences to predict market volatility is still a challenge. Especially, applying traditional recurrent neural networks (RNNs) on an extremely long event sequence is infeasible due to the high time complexity and the limited capability of the memory units in RNNs. In this paper, we propose a new deep neural network-based architecture named Long- and Short-term Memory Retrieval (LSMR) architecture to forecast short-term and mid-term volatility. LSMR architecture consists of three separate encoders, a query extractor, a long-term memory retriever, and a volatility predictor. The query extractor and the long-term memory retriever compose a long-term memory retrieval mechanism that enables the LSMR to handle the extremely long event sequences. Experiments on our novel news dataset demonstrate the superior performance of our proposed models in predicting highly volatile scenarios, compared to existing methods in the literature.
{"title":"Market volatility prediction based on long- and short-term memory retrieval architectures","authors":"Jie Yuan, Zhu Zhang","doi":"10.1145/3383455.3422545","DOIUrl":"https://doi.org/10.1145/3383455.3422545","url":null,"abstract":"Predicting market volatility is a critical issue in financial market research and practice. Moreover, in natural language processing, how to effectively leverage long- and short-term event sequences to predict market volatility is still a challenge. Especially, applying traditional recurrent neural networks (RNNs) on an extremely long event sequence is infeasible due to the high time complexity and the limited capability of the memory units in RNNs. In this paper, we propose a new deep neural network-based architecture named Long- and Short-term Memory Retrieval (LSMR) architecture to forecast short-term and mid-term volatility. LSMR architecture consists of three separate encoders, a query extractor, a long-term memory retriever, and a volatility predictor. The query extractor and the long-term memory retriever compose a long-term memory retrieval mechanism that enables the LSMR to handle the extremely long event sequences. Experiments on our novel news dataset demonstrate the superior performance of our proposed models in predicting highly volatile scenarios, compared to existing methods in the literature.","PeriodicalId":447950,"journal":{"name":"Proceedings of the First ACM International Conference on AI in Finance","volume":"56 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-10-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124626191","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：