Technical Analysis for Short-Term Forecasting of Financial Data and Turn of the Year Effect

Thumbnail Image
Paul, Parmod Kumar
Vijay, Vivek
Journal Title
Journal ISSN
Volume Title
Indian Institute of Technology Jodhpur
Stock market always attact investors to invest money according to their choice form which large profits can be earned. The fundamental drive behind maximizing this profit is strategy of buying and selling of stocks. Prediction of buying and selling patterns of stocks, or the whole market has always been a challenging task. It is due to the complexity,high volatility and non-inearity in the data. The rate of variation of financial time seres depends on several factors, such as fluctuations, interest rates and volume of transactions. Several statistical and machine learning techniques have been developed to forecast the movement of financial time series. Here we first discuss the trading band approach tob predict buy or sell patterns of a particular stock. These bands suggest buy or sell signals based on historical movements. Originally developed by J.H Hurst, these bands became more popular when a trading band was defined by using Moving Average (MA). The most popular trading band is the Bollinger Band, develoed by John Bollinger in 1980. These are volatility bands placed below and above the moving average of given financial time series. Al-though, the dynamic nature of these band makes them useful for different secrities with standard settings but due to the low decision time they arev unable to capture sudden peaks. We develop a new trading band ( Optimal Band ) which is based on absolute extrema (maxima and minima) and local extrema. We also develop an approach of predicting the buy / sell pattern using Hidden Markov Models. On the other hand, if trading bands and technical indicators exhibit similar partterns for two or more stocks, the decision is made on the basis of return and association with parttern. We first classify the historic data as per their pattern by using the Optimal Band. For each of the categories of pattern, we further divide the whole data into different categories of returns. If the interest lies in the interested in forecasting the returns then the historic value of pattern are used to predict the same but if one is interested in forecasting the returns then the historic value of pattern becomes more useful. Therefore, it becomes important to analyze the strength of dependence between the two variable, returns and patterns. We use historic data to see buying and selling pattern by using the Optimal Band. The pattern data is tham divided into there categories, namely, sell, neutral and buy. This is further used to estimate the future category of returns, high, moderate and low. The whole data is then presented in the form of a 2-dimensional contingency table by using the variables, returns and pattern. In techincal analysis, one of the fundamental drivers is volmue of transactios. We include volume as the third variable with its two categories, namely up and down. This division of volume is parimarily based on the range of historic returns. This creates a 3-dimensional contingency table. there are two possible sets of partial tables corresponding to the variable volume. We test different hypotheses for these tables. turn of the year effect, also known January effect, refers to a phenomenon of changing behaviour of stocks during some trading days of the January month. The presence of this effect is well investigated on high returns of the January month for small capital companies. We provide an evidence of the effect by using buying selling ratio and logistic regression. We predict the probability of next pattern from the given state of pattern ( buy, sell, neutral). The reslts are demonstrated for the data od Cipla Pharmaceutical Pvt. Ltd., Tata Motors and Maruti Suzuki.
Paul, Parmod Kumar. (2018). Technical Analysis for Short-Term Forecasting of Financial Data and Turn of the Year Effect (Doctor's thesis). Indian Institute of Technology Jodhpur, Jodhpur.