Predicting Financial Prices of Stock Market using Recurrent Convolutional Neural Networks

Financial time-series prediction has been long and the most challenging issues in financial market analysis. The deep neural networks is one of the excellent data mining approach has received great attention by researchers in several areas of time-series prediction since last 10 years. “Convolutional neural network (CNN) and recurrent neural network (RNN) models have become the mainstream methods for financial predictions. In this paper, we proposed to combine architectures, which exploit the advantages of CNN and RNN simultaneously, for the prediction of trading signals. Our model is essentially presented to financial time series predicting signals through a CNN layer, and directly fed into a gated recurrent unit (GRU) layer to capture long-term signals dependencies. GRU model perform better in sequential learning tasks and solve the vanishing gradients and exploding issue in standard RNNs. We evaluate our model on three datasets for stock indexes of the Hang Seng Indexes (HSI), the Deutscher Aktienindex (DAX) and the S&P 500 Index range 2008 to 2016, and associate the GRU-CNN based approaches with the existing deep learning models. Experimental results present that the proposed GRU-CNN model obtained the best prediction accuracy 56.2% on HIS dataset, 56.1% on DAX dataset and 56.3% on S&P500 dataset respectively.

Looking at Extremes without Going to Extremes: A New Self-Exciting Probability Model for Extreme Losses in Financial Markets

Forecasting market risk lies at the core of modern empirical finance. We propose a new self-exciting probability peaks-over-threshold (SEP-POT) model for forecasting the extreme loss probability and the value at risk. The model draws from the point-process approach to the POT methodology but is built under a discrete-time framework. Thus, time is treated as an integer value and the days of extreme loss could occur upon a sequence of indivisible time units. The SEP-POT model can capture the self-exciting nature of extreme event arrival, and hence, the strong clustering of large drops in financial prices. The triggering effect of recent events on the probability of extreme losses is specified using a discrete weighting function based on the at-zero-truncated Negative Binomial (NegBin) distribution. The serial correlation in the magnitudes of extreme losses is also taken into consideration using the generalized Pareto distribution enriched with the time-varying scale parameter. In this way, recent events affect the size of extreme losses more than distant events. The accuracy of SEP-POT value at risk (VaR) forecasts is backtested on seven stock indexes and three currency pairs and is compared with existing well-recognized methods. The results remain in favor of our model, showing that it constitutes a real alternative for forecasting extreme quantiles of financial returns.

Forecasting Bitcoin closing price series using linear regression and neural networks models

In this article we forecast daily closing price series of Bitcoin, Litecoin and Ethereum cryptocurrencies, using data on prices and volumes of prior days. Cryptocurrencies price behaviour is still largely unexplored, presenting new opportunities for researchers and economists to highlight similarities and differences with standard financial prices. We compared our results with various benchmarks: one recent work on Bitcoin prices forecasting that follows different approaches, a well-known paper that uses Intel, National Bank shares and Microsoft daily NASDAQ closing prices spanning a 3-year interval and another, more recent paper which gives quantitative results on stock market index predictions. We followed different approaches in parallel, implementing both statistical techniques and machine learning algorithms: the Simple Linear Regression (SLR) model for uni-variate series forecast using only closing prices, and the Multiple Linear Regression (MLR) model for multivariate series using both price and volume data. We used two artificial neural networks as well: Multilayer Perceptron (MLP) and Long short-term memory (LSTM). While the entire time series resulted to be indistinguishable from a random walk, the partitioning of datasets into shorter sequences, representing different price “regimes”, allows to obtain precise forecast as evaluated in terms of Mean Absolute Percentage Error(MAPE) and relative Root Mean Square Error (relativeRMSE). In this case the best results are obtained using more than one previous price, thus confirming the existence of time regimes different from random walks. Our models perform well also in terms of time complexity, and provide overall results better than those obtained in the benchmark studies, improving the state-of-the-art.

On the predictability of high and low prices: The case of Bitcoin

<p>Bitcoin has attracted the attention of investors lately due to its significant market capitalization and high volatility. This work considers the modeling and forecasting of daily high and low Bitcoin prices using a fractionally cointegrated vector autoregressive (FCVAR) model. As a flexible framework, FCVAR is able to account for two fundamental patterns of high and low financial prices: their cointegrating relationship and the long memory of their difference (i.e., the range), which is a measure of realized volatility. The analysis comprises the period from January 2012 to February 2018. Empirical findings indicate a significant cointegration relationship between daily high and low Bitcoin prices, which are integrated on an order close to the unity, and the evidence of long memory for the range. Results also indicate that high and low Bitcoin prices are predictable, and the fractionally cointegrated approach appears as a potential forecasting tool for<br />cryptocurrencies market practitioners.</p>

Financial Pricing in Property and Liability Insurances, Evidence From the Egyptian Insurance Market

The use of pricing a model's insurance derivatives in corporate risk management, particularity in insurance has grown rapidly recently. Financial prices for insurance reflects equilibrium relationships between risk and return or, minimally, avoid the creation of arbitrage opportunities. The major objective of this article is to provide evidence that in the Egyptian insurance market during the period 2002-2013, using Black-Scholes model, there was a transfer of wealth from policyholders to insurance companies via overvaluation of insurance premiums. This contribution may have some crucial implications in terms of the “fairness” of pricing insurance contracts.

An Essay on Financial Information in the Era of Computerization

This article deals with the increasing computerization of the financial markets and the consequences of such process on our ability to collect information about financial prices. The concept of information is at the heart of financial economics simply because this notion is a precondition for all investments. Since financial prices characterize an agreement on a transaction between two counterparties, they understandably became a key informational indicator for decision. This article will analyse the increasing computerization of financial sphere by discussing the recent emergence of what is called a “flash crash” and its impact on the traditional ways of collecting information in finance (technical analysis, fundamental analysis and statistical approach). I argue that the growing computerization of financial markets generated a “hyper-reality” in which financial prices do not refer to “something” anymore implying a revision of our usual way of defining/using the notion of information.

Asset price formation and behavioral biases

Purpose The purpose of this paper is to examine the debate on whether psychology affects asset prices using agent-based modeling. Design/methodology/approach The authors set up three simulation regimes where the first regime contains fundamental investors who invest based on the mean-variance framework. The second regime includes purely irrational investors who invest based on behavioral biases. The third regime combines the two types of investors. The authors test whether the return properties from regime 3 converge to that of regime 1 or 2. Findings Results suggest that the type of irrationality affects return properties in different ways. Irrational investors who are introspective in their irrationality, only examining their performance and deficiencies, do not have much of a systematic effect on stock returns when combined with rational investors. However, irrational investors that aggregate information in an irrational manner have a systematic effect when combined with rational investors. Research limitations/implications Research implication of using simulation analysis is that the results need to be verified via other methods such as empirical and/or experimental analysis. Practical implications Practical implications of the research is that policy makers can look for factors that investors use to aggregate to better understand the movement of financial prices and ignore other factors. Social implications Social implication is that mass psychology impacts financial prices. Originality/value No other paper has used agent-based/behavioral analysis to better understand how different types of behavior may impact financial prices in different ways.