Technical indicator selection and trading signal forecasting: varying input window length and forecast horizon for the Pakistan Stock Exchange

To improve the profitability and predictability of financial markets we highlight the issue of relevant technical indicators (TI) for different markets. In previous studies, not much attention has been given to feature selection problem by counting on the most popular technical indicators or following the footsteps of related literature. This research work is focusing on the importance of feature selection problem. There is no specific set of TI that is good enough to predict every market price movement. TI which are good predictors of developed markets are not suitable for emerging or frontier markets. The analysis is based on a well diversified sample of nine countries representing developed, emerging and frontier markets as per the categorization of Morgan Stanley Capital International (MSCI). We use random forest (RF) technique for feature selection from a group of 90 technical indicators. The results show that top five technical indicators are different for each market according to Gini index. Even within the three categories of market, these indicators and their ranking varies. Hence, appropriate feature selection according to market eventually improves the accuracy of predictive model and profitability.
 Purpose: Finding the relevant technical indicators for financial market is crucial for financial market prediction. There is no particular set of technical indicators that fits all financial market since every market has its own dynamics and it is important to first select the technical indicators that reflects the market conditions rather than using the most commonly employed technical indicators.
 Design/Methodology/Approach: Random forest technique is used to find the relevant technical indicator. And each indicator is ranked according to Gini index. Findings: The analysis suggest that the set of technical indicators is not the same for every financial market. Further with in the categories of developed markets, emerging markets and frontier markets the set of technical indicators changes.
 Implications/Originality/Value: Hence, it is summarized that to improve the forecasting ability of the predictive model it is important to find the market specific technical indicators that will end up in higher profit for investors.

There are two approaches to analyzing the value of a stock in financial markets: fundamental analysis and technical analysis. While fundamental analysis focuses on finding the intrinsic value of a stock based on a company's financial condition and current market conditions, technical analysis focuses on identifying trading signals in patterns by examining historical price behavior and statistics. Although technical analysis, which is based on the assumption that past price movements can be an indicator for future price movements, has a predefined set of rules, the interpretation of the results is closely related to the experience of the analyst. Therefore, the interpretive part of technical analysis has a subjective dimension. This subjective dimension and predefined set of rules indicate that machine learning methods with experience-based learning logic can be an important tool in identifying trading signals or predicting price movements. The aim of this study is to investigate the potential use of machine learning algorithms that use technical analysis indicators of stocks traded in Borsa Istanbul as input to predict trading signals and price movements. In the study, technical analysis indicators are analyzed with models based on machine learning methods and the results are compared. The findings show that the addition of machine learning methods to technical analysis strategies increases the predictive power of trading signals and price movements.

This study aims to retrieve useful knowledge from commonly adopted technical indicators, based on a soft computing model, to support investment decisions. Though the validity of technical analysis (TA) has been examined extensively by various statistical models in financial literature, a practical approach that may consider the inconsistency among various technical indicators and the down-side risk of an investment is still underexplored. As a result, the present study takes a twostage approach to construct a variable consistency dominancebased rough set approach (VC-DRSA) model, to retrieve the imprecise patterns and implicit knowledge from technical indicators. At the first stage, the trading signals indicated by various technical indicators are suggested by domain experts, and those signals were simulated by a trading strategy to examine the outcomes of each indicator. And the simulation results of each technical indicator are further processed by VC-DRSA model at the second stage for retrieving decision rules (i.e., knowledge). To illustrate the proposed model, the weighted average index of the Taiwan stock market was examined by using its historical data from mid/2002 to mid/2014, and a set of decision rules with more than 70% classification accuracy were inducted in this empirical case. The findings suggest that certain technical indicators should be considered simultaneously, and the obtained rules have practical implications for investors.

To support investment decision based on technical analysis (TA), this study aims to retrieve the knowledge or rules of various indicators by a hybrid soft computing model. Although the validity of TA has been examined extensively by various statistical methods in literature, previous studies mainly explored the effectiveness of each technical indicator separately; therefore, a practical approach that may consider the inconsistency of various technical indicators simultaneously and the down-side risk of an investment decision is still underexplored. Thus, a hybrid model—by constructing a variable consistency dominance-based rough set approach (VC-DRSA) information system with the fuzzy inference-enhanced discretization of signals—is proposed, to retrieve the imprecise patterns from commonly adopted technical indicators. At the first stage, the trading signals (i.e., buy, neutral, or sell) are preprocessed in two groups: straight-forward signals and complicated signals. The straight-forward technical indicators (i.e., for signals that are decided by precise rules) are suggested by domain experts, and the buy-in signals are simulated by several trading strategies to examine the outcomes of each indicator. As for those complicated signals (i.e., for signals that require imprecise judgments with perceived feeling of domain experts to identify patterns), a fuzzy inference technique is incorporated to enhance the discretization of signals; those signals are also simulated by the aforementioned trading strategies to obtain the corresponding results. At the second stage, the trading signals generated by each technical indicator and their pertinent results from the previous stage are combined for VC-DRSA modeling to gain decision rules. To illustrate the proposed model, the weighted average index of the Taiwan stock market was examined from mid/2002 to mid/2014, and a set of decision rules with nearly 80 % classification accuracy (both in the training and the testing sets) were obtained in this empirical case. The findings suggest that several technical indicators should be considered simultaneously, and the retrieved rules (knowledge) have practical implications for investors.

Technical analysis strategy optimization using a machine learning approach in stock market indices

Using the daily, weekly and monthly data of Morgan Stanley Taiwan Index’ Component Stocks from January 1998 to June 2004 as example, this paper try to improve the investment performance of technical analysis indices. First of all, this paper whitens original data through a grey model GM (1,1), and grey technical analysis indices are obtained. We use traditional technical analysis indices like RSI, BIAS, KD as agency indices, compare the investment performance between original and grey technical analysis indices. Based on the empirical result using frequency statistics and Sharpe and Treynor index, we find: 1.The investment performance using the original technical indices is better than that of the Taiwan stocks market weighed average index. 2.The portfolio investment performance using grey technical analysis strategy is better than that of buy-and-hold strategy and using traditional technical analysis strategy by Sharpe and Treynor index separately. 3.The investment performance of technical indices can be increased more effectively through a grey forecasting model GM (1,1). The investment performance of original technical analysis indices strategy increases significantly after using the grey technical analysis indices. The annually performance as geometric mean increases 278.5%. The weak-form market hypothesis in Taiwan stocks market could be rejected under some fact.

The creation of a predictive system that correctly forecasts future changes of a stock price is crucial for investment management and algorithmic trading. The use of technical analysis for financial forecasting has been successfully employed by many researchers. Input window length is a time frame parameter required to be set when calculating many technical indicators. This study explores how the performance of the predictive system depends on a combination of a forecast horizon and an input window length for forecasting variable horizons. Technical indicators are used as input features for machine learning algorithms to forecast future directions of stock price movements. The dataset consists of ten years daily price time series for fifty stocks. The highest prediction performance is observed when the input window length is approximately equal to the forecast horizon. This novel pattern is studied using multiple performance metrics: prediction accuracy, winning rate, return per trade and Sharpe ratio.

The paper researches on the effectiveness of technical indicator with the volume by choosing the volume- weighted moving average (VWMA) which consists of the simple moving average (MA) and volume based on the 2139 stocks from January 1, 2003 to January 1, 2013 in China's A- share market. There are four criteria to evaluate the volume- weighted moving average: sensitivity, reliability, risk and benefits. Comparing the performance of VWMA and MA in the sample stocks, there are several advantages by adding volume information to technical indicators: issue more trading signals, increase the proportion of profitable trading, reduce the average drawdown, and improve investment returns. Technical analysis indicators are applied extensively in stock market and frequently utilized as tools to forecast the trend of the stock price. But the results are often not satisfactory when investors use the technical analysis indicators which are only related to the stock price and not involve volume to forecast the trends. Volume is a number that investors willing to buy the stock at a price and shows the force of rose or fell. The volume is also on behalf of the support and pressure in stock market. Therefore the relationship between price and volume in stock market is always a hotspot in the financial sector and has been studied by a large number of scholars. In the paper, we research on the effectiveness of technical indicator with the volume on the basis of the conclusions about the relationship between price and volume and hope that the conclusions can be as the reference basis for the technical analyst when they study the stocks..

Technical indicators are algorithms that take financial time series data as input and predict price movement directions based on mathematical calculations. Technical analysts can predict the stock price trends by interpreting the results of different technical indicators. In this study, we investigate the prediction of price movement directions based on the use of technical indicators in learning algorithms. We explore the technical indicators that provide the most successful prediction when they are used together in learning algorithms. In this paper, we investigate the technical indicators that lead to the most successful price movement direction prediction. To do this, we explore all possible combinations of indicators with various machine learning algorithms. Here, a decision support system is proposed to predict price movement direction on financial time series data by using technical indicators in machine learning algorithms.

In this paper, a neural network-based stock price prediction and trading system using technical analysis indicators is presented. The model developed first converts the financial time series data into a series of buy-sell-hold trigger signals using the most commonly preferred technical analysis indicators. Then, a Multilayer Perceptron (MLP) artificial neural network (ANN) model is trained in the learning stage on the daily stock prices between 1997 and 2007 for all of the Dow30 stocks. Apache Spark big data framework is used in the training stage. The trained model is then tested with data from 2007 to 2017. The results indicate that by choosing the most appropriate technical indicators, the neural network model can achieve comparable results against the Buy and Hold strategy in most of the cases. Furthermore, fine tuning the technical indicators and/or optimization strategy can enhance the overall trading performance.

A trading strategy is an algorithm that provides decision support for a trader. An ideal system suggests which stocks to buy and sell at every moment. Limited but still very useful trading strategies suggest stocks to buy, but leave the sell decisions and the decision of proportions of different stocks to the trader, or to another automatic decision mechanism. In this paper we use a previously introduced method of predicting rank variables to produce both buy and sell decisions. The rank variables are predicted by neural networks, and provide an efficient way to produce daily buy (and also sell) suggestions. This should be seen in contrast to “ordinary” technical indicators that often give very few signals, or buy/sell signals for many stocks at the same time. The produced buy signals are further processed in a classification module that aims at identifying which of the numerous buy signals one should trust, and of which ones one should discard. The classification reduces the number of buy signals and also increases both hitrate and overall profit for a simulated trader. Data from the US stock market for 1992-2001 is used in the tests of the system, and the results show how a trading system’s performance can be significantly improved by adding a post-processing classification layer between the generation of trading signals and the actual decision making.

The Internet-of-Things (IoT) technologies are essential in deploying successful IoT-based services, especially in the financial services sector in recent years. Stock market prediction, which could also be an IoT-based service, is a very attractive topic that has inspired countless studies. Using financial news articles to forecast the effect of certain events, understand investors’ emotions, and react accordingly has been proved viable in the existing pieces of the literature. In this study, we utilized Chinese financial news in an attempt to predict the stock price movement and to derive a trading strategy based on news factors and technical indicators. First, the stock trend prediction (STP) approach is proposed. It first extracts keywords from the given articles. Then, the 2-word combination is employed to generate more meaningful keywords. The feature extraction and selection are followed to obtain important attributes for building a trading signal prediction model. Also, to make the trading signal more reliable, the technical indicators are considered to confirm the trading signal. Because the hyperparameters for the STP and technical indicators will have influenced the final results, an enhanced approach, namely, the genetic algorithm (GA)-based STP (GASTP) approach, is then proposed to find hyperparameters automatically for constructing a better prediction model. Experiments on real data sets were also made to show the effectiveness of the proposed algorithms. The results show that the GASTP performs better than the buy-and-hold strategy as well as the STP.

This paper presents a study and implementation of a stock trend prediction system based on Artificial Neural Network (ANN) and fuzzy logic rules. Technical analysis tools such as technical indicators and Elliott's wave theory were deployed in the presented prediction system. In this approach the neural network functions as a classifier, where the technical analysis indicators are its input features. The multilayer perceptron (MLP), Support Vector Machine (SVM) and Radial Bases Function (RBF) are tested as classification tools. Also, a fuzzy rule based system based on the Elliott's wave theory is developed to predict the short term stock trend. Finally, integration between these two modules is established using neural network. The System was trained and tested with real data from the Egyptian stock market. The obtained results are encouraging.

A dynamic threshold decision system for stock trading signal detection

In this paper, an expert system is developed using fuzzy genetic network programming with reinforcement learning (GNP-RL) in order to generate stock trading signals based on technical indices of the stock prices. In order to increase the accuracy and reliability of results, we applied Wavelet Transform to eliminate noises and irregularities in prices. Since choosing the most appropriate wavelet base is an important decision, the Energy to Shannon Entropy Ratio, as an objective method, is used in order to address this concern. For developing this system, we applied fuzzy node transition and decision making in both processing and judgment nodes of GNP-RL. Consequently, using these method not only did increase the accuracy of node transition and decision making in GNP's nodes, but also extended the GNP's binary signals to ternary trading signals. In other words, in our proposed Fuzzy GNP-RL model, a No Trade signal is added to conventional Buy or Sell signals. The proposed model has been used to generate trading signals for ten companies listed in Tehran Stock Exchange (TSE). The simulation results in testing time period shows that the developed system has more favorable performance in comparison with the simple GNP-RL with binary signals and Buy and Hold strategy.