This collection of five lectures I will teach at the Indian School of Business covers a small selection of topics in FinTech. I will be using slides, readings, referring to various URLs (web sources) and programming in Python. I will use mostly open source tools, but some in class demos will also use proprietary products. I will host all the materials in github at the following repository, from which you may download the needed files (or clone and pull) the repository if you are familiar with using github. See: http://github.com/srdas/IDTF.

The five lectures are described below. The numbering begins from Lecture 6, as there are going to be five lectures preceding this one.

This lecture covers: (i) An introduction to Python in Jupyter notebooks; (ii) Blockchains and Cryptocurrencies; (iii) Payment systems. We will begin with an introduction to Python for Financial Modeling, so that you are able to come up to speed with the main language being used for many applications in FinTech. In order to get a job in the FinTech space, Python is the lingua franca. In preparation for this class, please undertake the following.

• Install anaconda, which includes Python and many useful libraries. Download the single install file from here: https://www.anaconda.com/download/. Once you have installed it, use the launcher to start up the Jupyter notebook. I will assume you know how to initiate and use Jupyter notebooks. (By the way, Jupyter is an acronym that the dev team chose as it includes three popular programming languages, Julia, Python, and R.)
• Visit the Jupyter project web site (http://jupyter.org/) and check out the documentation there. See: https://jupyter.readthedocs.io/en/latest/content-quickstart.html.
• Try out the starting notebooks from here: https://try.jupyter.org/
• Especially, if you have not use Python before, then please go through the "Welcome to Python" notebook. Open it in your browser and then execute each cell block, you need to use the command Shift+Enter (as stated there). Look at the output and then compare it with the code.
• Read the PDF file "Python Tutorial for Absolute Beginners.pdf" in the folder for Lecture 6.
• A good book to get started with Python is "Python for Data Analysis", which I prefer to call p4da - it an excellent introduction, and is especially apt because the author WesMcKinney built the Pandas package (Chapter 5) for use at one of the top hedge funds (a good example of a firm that uses FinTech). An older edition is available online: http://www.cin.ufpe.br/~embat/Python%20for%20Data%20Analysis.pdf. I strongly recommend skimming through and learning the first four chapters, and then spending time on chapter 5, giving it a much more careful read. In addition, the best way to learn is to try out lines of code in a Jupyter notebook alongside, as you read.
• I will be using the following notebook in class, which will be provided in advance: GetStarted_Python_in_Finance.ipynb. You will find this in the course folder. You do not need to work through this before class, but load it up and bring it on your laptop for class.
• Python is the lingua franca of FinTech, see for example: https://stxnext.com/blog/2017/09/19/why-python-should-be-technology-choice-your-fintech/ and https://www.cleveroad.com/blog/see-why-python-is-the-best-programming-language-for-fintech-software.

• Next, we will move on to a technical exposition of blockchains and related cryptocurrencies. You will have already obtained some familiarity with this topic from earlier lectures.
• The slides for this lectures are in: 6_BlockChains_and_Cryptocurrencies. Within these slides there will also be links to more readings. This is to incentivize you to go through the slides in advance and follow the links and do some more reading to facilitate discussion in class. This is expected of you.
• Coverage will be on both the business and technical aspects of: distributed ledgers; blockchain; cryptocurrency; bitcoin; transaction flow; hash functions; digital signatures; double spending; proof of work; mining; security; malicious attack; Ethereum; smart contracts; merkle tree; gas; difficulty; Ethereum virtual machine; DAO; initial coin offerings; hyperledger; futures; blockchain infrastructure; payment systems; crypto portfolios (see CryptoPortfolios.ipynb).
• Please read all the PDF files in the lecture folder.
• The two Python notebooks for this segment are: SimpleCryptocurrencyCode.ipynb and CryptoPortfolios.ipynb.

• In this lecture, we will undertake a somewhat detailed look at why and how text analytics is becoming an important part of the FinTech revolution.
• We will use Python and learn many Python tools for text analytics.
• In the first Python notebook titled TextMining we introduce basic text analysis. Please work through this before class.
• The package that is useful is known as NLTK, and this stands for "Natural Language Tool Kit". See http://www.nltk.org/book/.
• We will learn how to import textual data from the web using the requests package, and also clean it up using the Beautiful Soup package. (Note: when you see the "import" command in Python, it is importing a package, and this means that the package needs to be installed. Sometimes the package is already part of the base installation. But it may not be. In such cases when you import the package it will throw an error. Then go ahead and install the package using the "pip install" command followed by the package name.)
• We will learn how to sentiment score text, and discuss how this may be used for financial modeling. For a detailed exposition of text mining in finance, please read the book titled "Text and Context: Language Analytics for Finance" (Das 2014): http://srdas.github.io/Papers/Das_TextAnalyticsInFinance.pdf.
• In the TwitterAnalysis.ipynb notebook, we will see how to handle a large collection of tweets. This is a popular input into automated trading strategies. We will also learn how to pull tweets online using the Twitter API, and then sentiment score these tweets.
• In the IndiaNewsExtractor.ipynb notebook, we will see how to pull news topics from the Economic Times. Here we will learn how to use CSS selectors. We will also learn various text analytics operations such as removing punctuation, numbers, stopwords, stemming, etc. We will consider the notion of a Corpus. The basic text data structure is a "Term Document Matrix" which we will construct from real data, and see why it is a fundamental starting point for textual analysis. We will also use this to construct Wordclouds.
• We will use a large Reuters data set of news articles to demonstrate advanced topics in financial text analysis, such as: topic modeling; word2vec; t-SNE representations. Again, this will be done using the NLTK package.

• In RegTech we discuss how technology enables new opportunities in financial regulation.
• As a example we will look at how emails may be used as an early warning indicator of financial failure. The paper titled "Zero-Revelation RegTech: Detecting Risk through Linguistic Analysis of Corporate Emails and News" (Das, Kim, Kothari 2017) will be discussed, and you may download the paper here: http://srdas.github.io/Papers/EnronZeroRev.pdf. The slides to accompany this part of the class are in 7_Enron_Text_Analytics.
• Also come prepared to discuss how RegTech may be implemented in India.

This entire lecture deals with financial automation in three separate areas of finance. As we will see, automation that is supported by theoretical modeling is a powerful disruptive force in FinTech.

• A discussion of goal-based wealth management products that are now being widely used by "robo" advisors.
• The theoretical modeling for goal based wealth management will be discussed. Read the paper titled "Portfolio Optimization with Mental Accounts," (Sanjiv Das, Harry Markowitz, Jonathan Scheid, and Meir Statman, 2010), PDF in the lecture folder. The companion practitioner version of the paper is also provided in Das, Markowitz, Scheid, and Statman, 2011.
• The Python program implementation for this is in the following Jupyter notebook: Markowitz_MeanVariance_Optimization_BPT.ipynb.
• The slides for this segment are in 8_GoalBasedAssetMgmt_BPT_RoboAdvising.
• We also examine how derivatives are used in this portfolio approach. See the paper JEDC_OptionsStructuredProducts_BPT.pdf.
• The original work by Markowitz is also available in the following PDFs in the folder: markowitz_JF_1952.pdf, markowitz1959.pdf.
• We analyze solutions when portfolios are underperforming their goals (Das, Kim, Statman 2014); see DasKimStatman_JPM_2014_ComingUpShort.pdf.
• Tax optimization is an important part of the process, and we examine this in Das, Ding, Newell, and Ostrov (2017); see: DasDingNewellOstrov_JOIS_2017.pdf.
• We examine longevity risk, an emerging issue that needs addressing through automation assisted planning.

• A quick introduction to network theory and how it may be used to think about systemic risk. Read Barabasi_ScaleFree_Networks.pdf.
• An application to the Indian financial system. See: 8_SysRiskIndia.

• An interesting new area of FinTech is democratized trading. In this new paradigm, individual investors can create their own trading algorithms, and share them on automated trading and backtesting platforms, like Quantopian: https://www.quantopian.com/home. We will examine the use of Quantopian briefly.
• The slides for this part of the lecture are in 8_Democratized_Trading.
• We will also review an entire range of metrics for trading algorithms here.
• See tutorials here: https://www.quantopian.com/tutorials.
• Another platform for automated trading is Numerai: https://numer.ai/. See also: https://hackernoon.com/numerai-walkthrough-quantitative-analysis-machine-learning-for-fun-and-profit-3dcdccabd920.

We continue with various examples of automation in FinTech. Here we use credit analysis as a exemplar for data robots. The slides for this lecture are in 9_FinTech_ML_Credit. All Python notebooks, data, and readings are also in the folder.

• We begin to look at AI and machine learning in this lecture.
• We will use Python to analyze a data set for credit card fraud detection, which is a huge problem being handled by modern technologies. Credit card companies that do not have good fraud detection algorithms are going to be at a massive disadvantage compared to those that do. The data set we will analyze is here: https://www.kaggle.com/dalpozz/creditcardfraud. We will construct a model for detecting fraud. This is a "needle in a haystack" problem, because fraud constitutes around 0.2% of all transactions. Please read this web site before class.
• The Python notebook with this analysis is in CC_Fraud_RF.ipynb.
• We will examine how to solve the sample imbalance problem using SMOTE, an oversampling technique.

• Here, we see how modern data analysis robots may be used to eliminate human modelers. The fact that we can just feed the data into a tool which then finds the best model is astonishing, and indicates how even white-collar jobs are now at risk. We will use https://www.datarobot.com/. This is a licensed software that is inexpensive for students in case you wish to try it out.
• As a second example, we will also use the same technology to let a machine learn about credit scoring from a database of small loans. Please review the following problem before class: https://www.kaggle.com/wendykan/lending-club-loan-data.

• Chatbots are a different kind of robot, now being widely used in finance for investor interfacing.
• We will build a simple chatbot using Lex, which is Amazon's chatbot constructor. See: https://aws.amazon.com/lex/. Review this site before class.
• The brief notes are in the following PDF: Chatbot using Lex.pdf.

• We examine the role of deep learning in finance. Slides for this lecture are in 10_AI_in_Finance. Our exploration will lead to a full analysis of the tensorflow package for deep learning.
• We will go through the backpropagation algorithm that drives the speed up of deep learning. And we will understand the role of hardware and data in making this new paradigm so successful.
• The Python notebook for deep learning is in FinTech_AI_DeepLearning_InFinance.ipynb. In this notebook you will learn about Keras and Tensorflow. See: https://keras.io/, https://www.tensorflow.org/.
• We will examine how deep learning is used for option pricing. See Das and Culkin (2017): CulkinDas_JOIM_2017.pdf.
• We will also see how deep learning may be used in an attempt to predict the direction of the stock index.
• Read How AI Will Invade Every Corner of Wall Street - Bloomberg.pdf for an examination of the effect this branch of FinTech on Wall Street. Think about how AI will impact finance in India. Come prepared to offer suggestions.
• A skeptical view is offered in Is AI Riding a One-Trick Pony_ - MIT Technology Review.pdf.
• Will AI take over humanity? The very interesting article by Francois Chollet, The impossibility of intelligence explosion – François Chollet – Medium.pdf, explores this issue. This is a fantastic read. Chollet is also the author of the widely used Keras package that is the front end we will use for running tensorflow in class.
• Please see the other PDF files in the lecture folder for other readings that we will discuss in class.