Data Science in Finance

Tags

Introduction

Introduction
Financial institutions are the pillars of the modern society, and have supported people in different walks of life. It is important to understand the applications of data science, if we are to predict the actions of these institutions and their evolution in the upcoming years.

In 2001, American energy company – Enron along with its auditing firm Arthur Andersen, were caught in an accounting fraud that eventually led to their bankruptcy. Shareholders of Enron ended up losing $74 billion and over 20,000 employees lost their jobs. Imagine if this fraud was detected at an early stage, or if the US. Securities and Exchange Commission could monitor or regulate this better?

Greece is still recovering from what is known as the worst economic crisis in the whole of Euro Zone. One of the main reasons for Greek Government Debt Crisis was that Eurostat(EU statistical institute) could not predict accurate GDP growth, deficit or public debt because of unreliable and inaccurate reporting of fiscal statistics. This debacle led to financial investors losing faith in the Greek economy and since then has had a ripple effect across the Eurozone.  If Eurostat could detect this at an early stage, this entire crisis could have been avoided perhaps?

Examples above show how economies of countries have been affected by frauds or unreliable and inaccurate data. Large amount of data can be leveraged to model this complex ecosystem of Finance and help predict frauds and avoid such faults in future.

Data science has assumed an increasingly important role in financial markets post financial crisis in 2011. It has exposed the weakness and limitations of existing models. Data science is also being used extensively for forecasting stock markets, currency exchange rate and bankruptcies [1]. It has evolved further and has helped us to understand and manage complex processes like price optimization, customer profiling and credit risk through historical data and strong statistical methods.

What is it?

Data Analytics in the field of Finance can be broadly classified into 3 separate categories, which are Descriptive Analytics, Predictive Analytics and Prescriptive Analytics.

 

Descriptive Analytics

Descriptive Analytics is the branch of analytics, which deals with understanding and analyzing historical data. It helps organizations understand what has happened in the past [2]. In finance descriptive Analytics usually involve:

• Measuring of Volatility of a specific stock in terms of its market price (Volatility of a stock price is nothing but the Standard deviation of the stock price over a given time frame) [3].
• Comparing two stocks on the basis of its cumulative returns [3].
• Analyzing specific chart and determining patterns on them by drawing trend lines. These patterns help an investor in making a decision based on the movement of the trend line. (Some of these chart patterns are Symmetric Triangle, Ascending/Descending triangle (Bull and bear approach), Head and Shoulder, Triple and Double Bottoms and Tops etc.) [4].

Predictive Analytics

Predictive Analytics uses a variety of techniques like data mining, machine learning, modeling and game theory to estimate the likelihood of a future outcome. Some of its applications in finance usually involve predicting the price of a given stock, determining credit scoring of a customer, determining risk involved for an investment etc. Considering the example of credit scoring, a financial institution can determine whether a given customer will be able to pay in the future depending on their credit history.

Prescriptive Analytics

Prescriptive Analytics, considered as the future of finance, informs the user about various possible actions that can be undertaken and suggests him the most optimal action. Prescriptive Analytics is the big brother of Predictive analytics and uses complex modeling algorithms like the Monte Carlo Simulation model. This model is one of the most popular prescriptive models in finance and is used by many companies and institutions for financial planning, making decisions and mitigating financial risks [5].

Conclusion

In the recent years, the application of data science to finance has produced interesting results. In addition to academia, several Wall Street firms like Morgan Stanley and JPMorgan have contributed significantly to the development of new methods and approaches [6]. In order to successfully implement these methods, the industry demands mature analytical processes, extensive mathematical skills and sophisticated statistics.

The need for data science is well recognized, but using it to analyses the humongous financial data to draw meaningful insights, is a challenging problem that requires sound domain expertise and interdisciplinary collaboration. Historically, in many industries, deep insights have been obtained only after accumulating enough empirical regularities [7].The future of data science in finance will be based on generating enough regularities and combine them with the prior functional knowledge via generic machine learning techniques. The “power” of data science in finance, if leveraged properly, has a crucial role to play in future.

References

 1. http://www.nag.com/IndustryArticles/DMinFinancialApps.pdf -Data Mining in Finance

2 http://www.bigdata-startups.com/understanding-business-descriptive-predictive-prescriptive-analytics/ — determining credit scoring

3 http://faculty.washington.edu/ezivot/econ424/descriptiveStatisticsPowerPoint.pdf – Descriptive statistics in finance (Measuring volatality (SD), Comparing Cumulative Returns on stock etc.)

4 http://www.chartadvisor.com/freereport/free_report_pg7.aspx – Analyzing Chart Patterns Descriptive Analytics

5 http://www.b-eye-network.com/view/17224 – Monte Carlo Simulation and Prescriptive analytics

6. Tze Leung Lai,Haipeng Xing – Statistical Models and Methods for Financial Markets

7. Mitchell T., Machine Learning. Tata McGraw Hill, 1997