In this study, the statistics of price change patterns is investigated firstly in Speculation Game (an agent-based market model characterized with round-trip trades), then in several real financial instruments, and finally in two other representative market models. The occurrences of historical patterns from Speculation Game suggest that the speculative spirit of the market may be demonstrated as the significant deviation from the uniform frequency patterns. This implication can also be verified from the statistical results of those highly speculative assets, such as gold price and foreign exchange rates. Furthermore, it is found that the reproduction of such historical patterns requires a bottom-up modeling of the markets, as the price change patterns can hardly be achieved in stochastic process models.

The increase of high-frequency trading (HFT) in stock market increases the importance of the analysis of its behaviors. In this study, we analyze it based on the previous research which analyzed it from the viewpoint of "Make and Take orders". The characteristics of our analysis is "analysis of tick distance", and our findings include: 1) the upgrade of stock trading system enables the handling of current trading volume that increased 4 times from 2010 to 2018, 2) Although HFT is making "make orders" away from the best quotes, such orders will not be executed.

This paper investigates the stock of Terilogy Co., Ltd.(code 3356, JASDAQ) using VPIN (Volume-Synchronized Probability of Informed Trading) proposed in Easley, et al. [1]. Especially, the stock is attracting attention in SNS etc., stock price is fluctuating. We examine the investor behavior with VPIN, and also consider how to use this theory is applied to the stock trading.

There are many studies predicting a stock market using social media. Suwa et al. (2017) proposed a VI index prediction model. They assumed that changes sentiments of investors are topics change posted on social media. However, in their prediction model, the data to use verification is included in the data to use to develop their topic model. Hence, their model might be overfitting. Therefore, we propose a prediction model of VI index avoiding overfitting. We developed a program that applies new posting messages to topic models of a learning period. We created data for verification using this program. As a result, we found that a logistic regression using time series topics on the past seven trading days may predict a rise in the VI index.

For financial institutions, it is important to monitor the performance and performance factors of corporate customers. Financial institutions accumulate large amounts of data to monitor customer information including their performance. It is relatively easy to extract information on performance from structured data, while it is difficult for non-structured data like text data. In this research, we tried to extract sentences that represent customer's performance factors from text data created by financial institutions. Our proposed method using deep learning extracts sentences similar to sentences prepared as correct examples in advance. Due to the difference in properties of text data, our method showed better performance than the prior research.

In this paper, we propose a methodology to forecast the direction and extent of volatility in mid-to-long term excess return of stock price by applying natural language processing and neural networks on the context of analyst reports. Analyst reports are prepared by analysts in research department in stock brokerage firms and we consider the content of reports include usefull information to forecast movements in stock prices. First, our method extracts opinion sentences from analyst reports, while the remaining parts correspond to non-opinion sentences. Second, our method predicts stock price movements by inputting opinion sentences and non-opinion sentences to neural networks separately.

In this paper, we proposed a new approach taking causal relationship into consideration with text mining for analyst report and news in automatic summarization. This approach can be used for reducing work load to read analyst report for institutional investors and gathering important economic information for investment decision for investment analysts. First, we analyzed the validity of the method in extracting causal relationship which can be evaluated from the textual data of Nomura Securities Co., Ltd. As a result, the using method could extract basis information of analyst's opinion from analyst report in higher precision, and we could confirm the style of analyst in expression of opinion and basis.

In this paper, we try to analyze relationships between analyst reports and corporate performances. The analyst reports are documents written about markets forecasts and they are useful for investment judgment. As analyst reports are written in natural language and data to be analyzed becomes complicated, a neural computational method called `potential learning' which can interpret internal representations was used. As a result, we found that a generalization performance of the model was 0.6773 (accuracy) and words related to `word category: abstraction' may affect corporate performances.

In this research, we introduce a system for searching causal relationships of events related to economics and finance in a chain-like manner from a database extracted from economic text data. This system also introduces service application ideas such as inferring economic ripple effects from events entered by users and presenting related stocks tracing the causal chain from news articles.

In recent years, investment strategies using artificial intelligence have attracted a significant amount of research attention. However, it is difficult to construct an efficient investment strategy using artificial intelligence owing to the variable factors in market prices. Therefore, this study aims to focus on a trading method called the NT ratio transaction to reduce the number of price-variable factors. This transaction is an arbitrage transaction, which utilizes the difference in the price movements between Nikkei 225 futures and TOPIX futures. These futures generally exhibit similar price movements and even if the price differences expand, they tend to return to their original separation. Using this transaction, we can target profits from this price difference while offsetting a considerable number of price-variable factors. Therefore, in this study, we construct a model to acquire an investment strategy based on NT ratio transactions via deep reinforcement learning and confirm the effectiveness of this model.

Researchs for financial time series in stock or foreign exchange markets, have been one of traditional themes of financial market analysis. Statistical model approaches such as ARMA and GARCH were mainstream of conventional analysis. However, it is difficult to understand and predict financial time series structures, which are generally characterized by high noise level and low autocorrelation. Meanwhile, researchs to capture the structure by artificial intelligence has been increasing in recent years. In particular, Long Short-Term Memory (LSTM), which can capture time series structure, is already widely used in the field of natural language processing and speech recognition. Therefore, in this study, we investigated the model performance in each TOPIX core30 constituent stock by using logistics regression (LOG), random forest (RAF), gradient boosting (GBT), support vector machine (SVM), and LSTM. The performance was evaluated by metrics such as prediction accuracy, F1 measure, AUC, and return. As a result, LSTM showed the best performance in the models. Moreover, we discussed the effectiveness of the stock market neutral strategy by applying the above prediction models. 10-quantile portfolios using the predicted probability outputted by the model, remarks higher accuracy and returns than individual stock trading in all models. Furthermore, LSTM outperformed the others and it is consistent with the result of S&P500 constituent stocks analysis.

In recent years, many researchers have studied stock trading using technical analysis. However, it is necessary to have deep knowledge to use such technical analysis and it is difficult to make a profit using such techniques. Therefore, we construct an evolutionary model to create a profitable investment strategy using technical indicators.

The goal of our research is to build default prediction models on the basis of machine learning models and to obtain useful information for corporate credit risk evaluation. The novelty of this work is twofold. The first point is on how to use time-series information of macroeconomic indexes for the default prediction model for small and medium-sized companies. Since macroeconomic indexes and financial data are different in frequency of being obtained, we considered how to combine these two kinds of data, as input of the default prediction model. In order to combine these data, we summarized time-series information of macroeconomic indexes in the form of mean, percentage change, and volatility. Regarding percentage change, some periods were adopted for the purpose of summarizing both of macrotrends and microtrends. The summarized forms and corporate financial indicators were used as input of the default prediction model in this research. As a result, the default prediction model with inputs of the financial indicators and the macroeconomic indexes outperformed the model with inputs of only financial indicators. Furthermore, the model, to the inputs of which the percentage changes in the fine periods summarizing microtrends were added, outperformed the model not considering the percentage changes in the fine periods. Therefore, considering macroeconomic indexes, especially our proposed method summarizing macrotrends and microtrends, has been found effective for default prediction. The second point is regarding which financial indicators are important in default prediction for small and medium-sized companies by industry sectors. We divided companies into eight industry sectors and investigated which financial indicators are important in each industry sector on the basis of variable importance evaluated with random forest.

Cash management is important to avoid cash-shortage, especially for Small and Medium Enterprises(SME) due to lack of excess cash. Whereas, cash management at SME is not sufficient because of its complexity and high cost. Therefore, we investigate a convenient method to predict a time to cash-shortage and an amount of shortage by Option Approach. In this study, We calculates a probability of default by using Option Approach, regarding its " asset " and " debt " as cash amount and any amount respectively. We evaluated Area Under Curve (AUC) by changing a prediction term, debt amount and calculation term of asset. We found several facts: max AUC is 0.81, short prediction term or large cash decrease makes high AUC, and calculation term is very small relation to AUC. As a result, Option Approach can predict a cash of SME in short term or large change. This approach is useful for cash management of SME to decide an acceptable debt amount and terms.

Recent globalism of economic makes the effect of a certain small bankruptcy enhance internationally beyond of the area and country. So, we propose the model in which the interaction between banks and companies is focused and simulate and consider the effect of whether banks invest the funds or not to the growth of companies. In particular, by varying the width of change of the growth rate of companies according to whether they are invested, we researched the difference in the distribution of the final size of companies. The model of banks is based on actual data such as financial statement analysis edited by Japanese Bankers Association. As the results, the investment of banks did not raise the mean of growth rate but widened the disparity in size among companies. As future plans, we'd like to reproduce and analyze the cause of large chain bankruptcy using the extended model.

In this paper, we modeled stock markets to investigate the effect of index investing on stock price formation. We showed that index investing has little effect on stock price formation in our stock markets model by analyzing results from experiments with various market settings.

Asset Management Industry may experience a drastic change. The added value of the industry might be lower due to the improvement of technology (Fintech), Fiduciary duty and passive trend. Under such situation, the differentiation of traditional investment will be more and more difficult, and investor will favor alternative products. However, Asset Management companies face the big hurdle in providing alternative products to individual investors due to the Liquidity. Japan regulator tend to restrict the illiquidity assets from the perspective of investor protection. This study will focus on the Real Estate, one of the major alternative products. Real Estate itself is illiquidity asset, while equity Real Estate Investment Trusts (REIT), one of the Real Estate investment products, is liquidity asset and every investor can trade it every day, every time as equity. And regulator tend to restrict other less liquidity Real Estate investment products from the perspective of investor protection as many investors had painful experience in 90s. There were many empirical researches for the liquidity issue of equity REIT. However, all there are using the historical data, and it is difficult the analysis without the market environmental change. This study uses a financial market simulation (artificial market) constructed virtually on computer to assess the relationship between liquidity and market price. Artificial market method has been using for short selling rule and other regulation rule in Japan, but no trade frequency simulation has been conducted before. Analysis results showed that 1. High liquidity might heighten the volatility of the marketable securities. And it will case the price divergence between market price and original value. 2. Fundamental investors and Technical investors should have the same trade frequency. 3. For real estate assets, due to the information disclosure frequency, the trade frequency which can make market efficiency might be once per monthly.

The financial authorities around the world have been trying to prevent the occurrence of the financial crisis and to reduce the impact by creating a common framework for finance. In response to the Lehman shock that occurred in 2008, many financial institutions were forced to fail due to lack of liquidity, and from the reflection that not only the capital adequacy ratio regulation but also the liquidity ratio regulation is necessary, Basel III regulation consisting of the capital adequacy ratio regulation and liquidity ratio regulation was formulated in 2010. However, some problems are pointed out in the liquidity ratio regulation established under the Basel III regulation. Liquidity ratio regulation formulated under the Basel III regulation is trying to encourage banks to prepare for the financial crisis accompanying liquidity shock by ensuring adequate liquidity but this will lower the bank's asset management yield and it is feared that it may be a factor that puts pressure on management and, on the contrary, makes the financial system unstable. Therefore, in this research, we evaluate the influence of the liquidity ratio regulation on the Basel III regulation on the stability of the financial system by simulation. There are Liquidity Coverage Ratio (LCR) regulation and Net Stable Funding Ratio (NSFR) regulation in the liquidity ratio regulation on the Basel III regulation. In this research, we will examine the merits and demerits of the impact of the LCR regulation and the NSFR regulation, which is the liquidity ratio regulation of Basel III regulation, on the financial system.