1. Mukesh Patel School of Technology Management and
Engineering
EXECUTIVE SUMMARY
ON
BIG DATA ANALYTICS AND RESEARCH IN
INSTITUTIONAL EQUITIES
BY
PRATIK NAWANI
Roll Number: J025
SAP Number: 71112110004
Faculty Mentor: Prof. Siba Panda
Industry Mentor: Mr. Abhijit Tare
ORGANISATION:
MOTILAL OSWAL FINANCIAL SERVICES LTD.
2. PROJECT INSIGHT
Big data is no longer confined to the realm of technology. Today it is a business imperative
and is providing solutions to long-standing business challenges for banking and financial
markets companies around the world especially NBFCs. Big data is promising and
differentiating for financial services companies. With no physical products to manufacture,
data – being the sole source of information – is one of arguably their most important assets.
This big data is structured and unstructured and requires certain logics to be built on them in
order to draw intelligence from them. This is exactly what my first project entailed. This
eclectic data is in huge amounts even for a particular department. One such department was
Institutional Equities which was responsible for trading, sales and research in all sectors with
their major focus being on facilitating block and bulk deals which earn them the maximum
brokerage. These large deals were greater than 75 lakhs each which involved finding a buyer
and seller so that brokerage can be gained on both sides. These clients were Mutual Funds,
Hedge Funds, Pension Funds, Foreign Investors and foreign funds, HNI clients having a
significant stake in companies. A system which could predict their buying or selling behavior
looking at historical patterns, live data, trends, their positions in stocks and previous trades
done with MOSL was the aim. Thus, a predictive system was built based on logics and
validations to identify counters and potential buyers and sellers. Thus, it enabled MOSL to
gain a competitive advantage over others vying for the same clients. This system also made
the sales trading efficient in many ways as it standardized certain terminologies, names and
other data which were present in different formats.
A trading analytics system called “INSIGHT” was built using data from stock exchanges,
data vendors, Bloomberg terminal etc. to predict fund behaviour for making block and bulk
deals during the day. This means which fund has a likelihood of selling the stock and who
can actually match the buyers need by selling an equivalent large amount in a single trade and
vice versa needs to be found out based on intelligence built on raw data. Therefore, in order
to find such matches an enormous amount of data- both structured and unstructured needs to
be aggregated. This Big data analysis is of a multitude of reports and live feeds which contain
humungous amount of data. They are:
End of Day Block and Bulk deals reported on NSE and BSE
Insider Trading Data – Both Live and Historical
Shareholding Patterns of Companies and Holders List from Bloomberg
3. Historical Holders and their position increase/decrease in a particular stock over a
period of time
Pledge Data
Live Market Condition/ News
Domestic Asset management Company Reports
The Business Requirement Specification was documented which forms the basis for any
analytics system. This document is used by the technical team to code the system. All the
intelligence and equations are part of the BRS which was formed by me after interacting with
Sales traders, sales team and research department. During this process, the financial
implications of the reports analyzed and their impact on intraday block deals etc. were
studied and how can this information be leveraged to predict these deals. Furthermore, I
learnt how to use the Bloomberg Terminal and its various functions within it for fund
analysis, shareholders list and monitoring intraday block deals, the Bloomberg Excel Add-in
function and how to automate the process for downloading shareholders list for more than
1000 stocks.
QUANTITATIVE RESEARCH
The other half of my internship entailed working with the Quantitative Research team which
involved doing technical analysis, building pair trading strategies and Value-at-risk models.
TECHNICAL ANALYSIS
I learnt the basics of technical analysis and then started doing stock predictions for a few
stocks and certain indices along with the research team and send these reports to clients and
fund managers (both foreign and domestic). These stock predictions were done using GPC
function on Bloomberg.
The basis of technical analysis relies on the 3 facets:
1. Price discounts everything: Price of the stock assumes everything that the investor
needs to know about the company/index
2. History repeats itself: Prices tend to move in trends and history is an excellent
indicator of the future.
4. 3. “What” is more important than “Why”: The reason behind the stock going up or down
is not important. The fact that it has risen or fallen and what might happen in the
future is purely dependent on the trend and the price of the stock.
The concepts of Candlestick theory was studied along with concepts of support, resistance,
how to draw trend lines, rising and falling channels, observing breakouts on significant
volumes, moving averages (simple as well as exponential), symmetrical, ascending and
descending triangles, indicators like Relative Strength Index (RSI), Moving Average
Convergence and Divergence (MACD), plotting Bollinger Bands, performing regression
analysis for long term view and application of other Stochastic indicators.
PAIR TRADING
A pair trade is a statistical arbitrage method in which it entails taking a long position in one
security together with an equal short position in another that is strongly correlated and co-
integrated with it. It is sometimes used to refer to multiple long and short positions that are
similarly matched. The pair trade is similar to a switch trade. In this type of trade, client
usually buys/sells a specified number of shares of a particular stock and sells/buys a specified
number of shares of some other stock, belonging to the same sector. Pair Trading is a market
neutral strategy and is meant to profit regardless of whether equities rise or fall. A profit or
loss on a pair trade depends on whether the spread between paired positions widens or
narrows. It is a mean reversion strategy in which the profit is realised when the pair’s price
ratio reverts back to normal.
The pair trading involves plotting ±2 Standard deviation of price ratios of highly correlated
pairs which are also co-integrated. Whenever the price ratio deviates from the 65 day
Displaced Moving average, the pair is entered by going long on the under performer and
shorting the over performer. When the pair reverts back to the mean, the positions are
squared off which gives the profit. Thus, the model was built which predicted the profit
potential and the stop loss was kept at ±3 Standard Deviation.
VALUE-AT-RISK
VaR calculates the worst expected loss over a given horizon at a given confidence level under
normal market conditions. A predictive parametric VaR model was built for portfolio
managers and HNI clients from exceeding risk tolerances and their risk-appetite. The
parametric method is also known as the variance-covariance method. The variance-
5. covariance method to calculate the value at risk calculates the mean, or expected value, and
standard deviation of an investment portfolio. The variance-covariance looks at the price
movements of investments over a historical period and uses probability theory to compute a
portfolio's maximum loss. The model was based on entirely on this. The clients required to
minimize losses in the stocks they held and find appropriate times when they can exit a stock
and diversify their portfolio when an opportunity arises by buying a stock which is not
correlated to its portfolio reducing the systematic risk Beta and eventually the Value at risk
value over a particular holding period.
CONCLUSION
The first project undertaken by me helped me to understand the working of an NBFC
especially its Institutional Equities department. I acquired an in-depth knowledge of Block
and Bulk deals and how are they executed by sales traders and dealers. The plethora of
learnings included:
1. Basic use of Bloomberg Professional analytics software and its various functions
2. Interpretation of the implications of various reports which assist traders in making
decisions for making block and bulk deals
3. Understanding different types of funds and their analysis based on various
parameters
The second half of my experience entailed working with the Quantitative Research team and
the knowledge gained was immense. An in-depth application of technical analysis of various
stocks and indices was performed by doing predictions using technical indicators. A
statistical arbitrage trading model was built using pair trading strategies.