NOV 2, 2019
Research paper
APPLICATION OF TIME SERIES ANALYSIS IN FINANCIAL ECONOMICS
Tags: Statswork | Time series forecasting | Python expert | Linear Regression Models | Logistic Regression | Programmers | Statistical Data Analysis Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright © 2019 Statswork. All rights reserved
TIME SERIES ANALYSIS 1
Time series analysis is widely applied to forecast the pattern/trends in the financial and market data.
2
A time series is actually a sequence of data points recorded at regular intervals of time (yearly, quarterly, monthly, daily).
3
The main objective of a time series analysis is to develop a suitable model to describe the pattern or trend in data with more accuracy.
4
The performance of the time series models can be interpreted based on its error terms such as AIC, BIC, Mean Squared Error, etc. and it can be emphasized for forecasting.
5
The principle interest for every time series analysis is to split the original series into independent components.
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright Š 2019 Statswork. All rights reserved
TYPES OF TIME SERIES Time series includes two types: Univariate
Involves single variable.
Multi variate
Involves two or more variables.
Typically, time series are further splited into three main components: Trend Seasonality Cycle Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright Š 2019 Statswork. All rights reserved
EXAMPLES OF TIME SERIES Monthly or daily precipitation of a region.
01 Annual unemployment rate over a period of 10 years.
04
02
Daily stock prices (opening, closing) over a period of years/days.
03
Monthly bike sales over a period of 3 years. Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright Š 2019 Statswork. All rights reserved
FORECASTING TIME SERIES DATA Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright © 2019 Statswork. All rights reserved
Forecasting a time series data predicts the future outcomes based on the immediate past. Forecasting can be done for closing/opening rate of a stock in daily basis, quarterly revenues of a company, etc. Decomposition of time series is much easy to forecast the individual regular patterns produced than from the actual series.
Models to forecast time series data are: Autoregressive Integration Moving Average (ARIMA) Simple Moving Average (SMA) Exponential Smoothing (SES)
Neural Network (NN) Logistic Regression
Linear Regression Models Support Vector Machine
•Naive Bayes •VAR
•Hidden Markov •Gaussian Processes
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Contd... Copyright © 2019 Statswork. All rights reserved
Some of the complex models or techniques to forecast a time series data are: Neural Networks Autoregression (NNAR)
1 Generalized Autoregressive Conditional Heteroskedasticity (GARCH)
2
4
RNN (Recurrent Neural Network)
3 Bayesian-based models
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright Š 2019 Statswork. All rights reserved
TIME SERIES FORECAST MODELS Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright © 2019 Statswork. All rights reserved
APPLICATIONS OF TIME SERIES FORECASTING Time series models usually used to forecast the stock’s performance, interest rate, weather, etc. In this post, we will look at few situations where time series can be useful to forecast the future outcome.
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright © 2019 Statswork. All rights reserved
APPLICATION - 1 Daily values of the data with time factor available between 2010 and 2016 from Yahoo Finance repository. Forecasting is done for SERIES A values based on the most recent trend (lags) of SERIES A volumes and SERIES B values and volumes and market sentiment using ARIMA model.
STEP: 1 Entire time line for 4 related time series (SERIES A values and volumes, SERIES B values and volumes) are extracted. Other dataset from Kaggle to forecast the market sentiment are also used. The data involves top daily news headlines between 2008 and 2016.
STEP: 2 Checking whether the time series is approximately stationary and normalized. For this situation, RNN forecasting is used to predict the outcomes variable of interest that results in the inference of the future time evolution of the SERIES A values based on its past trends (including volumes) as well as the past trend of another SERIES B, and market sentiment. Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright Š 2019 Statswork. All rights reserved
RESULTS The mean absolute error (MAE) is used to understand the trend in this graph and it is 10, 24, 14, 15 for each sample respectively. The figure- 1 shows a comparison of 10-day forecast.
Figure- 1: Comparison of 10- day forecast.
In a traditional regression data, dependent or response variable is influenced by a set of independent variables. The degree of dependence on previous outcomes varies for each case, and can be explained by (ACF) Auto Correlation Factor as given in below figure- 2.
Figure- 2: ACF and PACF Plots of SERIES A value and volume.
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright Š 2019 Statswork. All rights reserved
CONCLUSION • Looking at the ACF and PACF plots, one can choose proper values of the parameters in the model. The results from ARIMA model will look like:
• Especially, from the results SERIES B values significantly influences the forecasting made for SERIES A. • The market sentiment (s) does too but to a lower extent, and volumes are relatively insignificant for forecasting the SERIES A values with ARIMA. Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright © 2019 Statswork. All rights reserved
APPLICATION - 2 Stock market data from 2016 from Kaggle and analyse the pattern of the data. • The data involves stocks of top companies such as Facebook, Apple, Amazon, etc. • Here is the trend of daily closing price of stocks for the month of January.
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright © 2019 Statswork. All rights reserved
RESULT The following graph depicts the trend of price change for a month of January. This is often referred as “momentum” in financial research.
CONCLUSION Time series in financial economics are highly important to analyse the trend or pattern of the variable of interest using an appropriate model. The above example clearly depicts the trend in price of the stock and this trend may be helpful in predicting the future stock values using suitable models as mentioned earlier.
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright © 2019 Statswork. All rights reserved
APPLICATION - 3 Trend of the sales and tractor demand in XYZ manufacturing company • The company is interested in understanding the impact of marketing efforts towards sales. • In such situation, finding the pattern of the sales and demand can be viewed using a well-known ARIMA model and predict the sales/demand for the upcoming years. • In addition, the impact of the marketing effort can be studied using exogenous variables under ARIMA model.
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright © 2019 Statswork. All rights reserved
SUMMARY There are various applications apart from these three in our day-to-day life. However, many financial organisation relies on time series forecasting to build their marketing strategy to meet the customer’s needs. Thus, a more proper model should be selected to analyse the pattern of financial data.
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright Š 2019 Statswork. All rights reserved
Statswork Lab @ Statswork.com www.statswork.com
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright Š 2019 Statswork. All rights reserved
PHONE NUMBER
UK
: +44-1143520021
INDIA
: +91-4448137070
EMAIL ADDRESS info@statswork.com
GET IN TOUCH WITH US
Research Planing | Data Collection | Semantic Annotation | Consumer & Retail Analytics | Econometrics
Copyright © 2019 Statswork. All rights reserved