Machine Learning For Stock Broking
Download as PPTX, PDF1 like766 views
The document outlines the capabilities of AlgoAnalytics, highlighting its expertise in applying machine learning and advanced mathematical techniques in various domains including finance, healthcare, and retail. It details the company's predictive analytics solutions for stock broking houses, emphasizing tools like dormancy analysis, recommender systems, and credit scoring. Additionally, it presents indicative case studies showcasing real-time applications and the effectiveness of these solutions in enhancing client engagement and decision-making.
Machine Learning For Stock Broking
- 1. Machine Learning for Stock Broking Houses
- 2. Page 2 © AlgoAnalytics All rights reserved Overview AlgoAnalytics: Company Profile One Stop AI Shop Solutions for BFSI Segment Indicative Case Studies
- 3. Page 3 © AlgoAnalytics All rights reserved CEO and Company Profile Aniruddha Pant CEO and Founder of AlgoAnalytics PhD, Control systems, University of California at Berkeley, USA 2001 • 20+ years in application of advanced mathematical techniques to academic and enterprise problems. • Experience in application of machine learning to various business problems. • Experience in financial markets trading; Indian as well as global markets. Highlights • Experience in cross-domain application of basic scientific process. • Research in areas ranging from biology to financial markets to military applications. • Close collaboration with premier educational institutes in India, USA & Europe. • Active involvement in startup ecosystem in India. Expertise • Vice President, Capital Metrics and Risk Solutions • Head of Analytics Competency Center, Persistent Systems • Scientist and Group Leader, Tata Consultancy Services Prior Experience • Work at the intersection of mathematics and other domains • Harness data to provide insight and solutions to our clients Analytics Consultancy • +30 data scientists with experience in mathematics and engineering • Team strengths include ability to deal with structured/ unstructured data, classical ML as well as deep learning using cutting edge methodologies Led by Aniruddha Pant • Develop advanced mathematical models or solutions for a wide range of industries: • Financial services, Retail, economics, healthcare, BFSI, telecom, … Expertise in Mathematics and Computer Science • Work closely with domain experts – either from the clients side or our own – to effectively model the problem to be solved Working with Domain Specialists About AlgoAnalytics
- 4. Page 4 AlgoAnalytics - One Stop AI Shop Aniruddha Pant CEO and Founder of AlgoAnalytics •We use structured data to design our predictive analytics solutions like churn, recommender sys •We use techniques like clustering, Recurrent Neural Networks, Structured Data •We use text data analytics for designing solutions like sentiment analysis, news summarization and many more •We use techniques like natural language processing, word2vec, deep learning, TF-IDF Text Data •Image data is used for predicting existence of particular pathology, image recognition and many others •We use techniques like deep learning – convolutional neural network, artificial neural networks and technologies like TensorFlow Image Data •We use sound data to design factory solutions like air leakage detection, identification of empty and loaded strokes from press data, engine-compressor fault detection •We use techniques like deep learning Sound Data BFSI •Dormancy Analysis •Recommender System •Credit/Collection Score Retail •Churn Analysis •Recommender System •Image Analytics Healthcare •Medical Image Diagnostics •Work flow optimization •Cash flow forecasting Legal •Contracts Management •Structured Document decomposition •Document similarity in text analytics Internet of Things •Predictive in ovens •Air leakage detection •Engine/compressor fault detection Others •Algorithmic trading strategies •Risk sensing – network theory •Network failure model © AlgoAnalytics All rights reserved
- 5. Page 5 © AlgoAnalytics All rights reserved Predict Dormancy – Finding which clients are unlikely to transact and take action Recommender System – Suggesting products likely to increase chance of action for a particular customer, cross-up sell Credit score – Application, behavior and collection scores, estimation of default Analytics and loans – Origination, pricing and valuation of loans Channel adoption and preference – Use demographics and trading data to build a classification model Whole Gamut of Solutions Automatic RFP Responses – Developed Machine Learning Based Question Answer System to respond to RFPs. Signature Verification – extract the signature thru OCR and validate the signature based on type of the document . Document Processing – Contract decomposition, document similarity and others Automated News Download and Summarization – Automatic download of relevant news items, News summarization Smart Inbox + Smart Reply – Routing emails to appropriate inbox and responding automatically to client email queries. Virtual Relationship Manager/ Customer Support Assistant – Assistant to increase accessibility for clients. This was developed using Microsoft framework and CNTK Text Analytics, Image Analytics, Time Series Modelling, Intent Analytics BOT Apps Text Analytics, Image Analytics, Client Analytics
- 6. Page 6 © AlgoAnalytics All rights reserved Indicative Case Studies Client Analytics
- 7. Page 7 © AlgoAnalytics All rights reserved Client Analytics - Dormancy, Stock & Mutual Fund Recommendation The relationship manager connects with the predicted dormant clients with the recommendation on stocks and Mutual Funds real time on the existing Apps and devices. Applications Real-Time Applications Mobile Apps Web Applications Last 5 stocks bought Recommended Stock Probability Intraday INFOSYS_TECHNOLOG IES Intraday State Bank of India 0.1629 Intraday MOTHERSON_SUMI_ SYSTEM Intraday L&T 0.137 Intraday UTI BANK LTD Longterm ICICI Banking Copora 0.124 Intraday ASIAN PAINT Intraday ICICI Banking Copora 0.0709 Intraday LIC HOUSING FIN Intraday Bharat Forge 0.061 Dormancy Prediction • RM wise list of predicted dormant clients will be published on the RM Dashboard. • This input can be feed into the stock and Mutual Fund Recommendation systems Stock Recommender • The system is designed to recommend stocks for the selected clients by RM or for the entire list of clients Mutual Fund Recommender • The system is designed to recommend Mutual funds types (Equity, Hybrid and Debt) for the selected clients by RM or for the entire list of clients Recommended MF Probability Equity MF 0.467 Hybrid MF 0.237 Debt MF 0.164 Integration with existing Applications on various devices
- 8. Page 8 © AlgoAnalytics All rights reserved Q5 Customers with no trades were marked as DORMANT Test data Label Q1 Q2 Q3 Q4 Q2 Q3 Q4 Q5 Modelling (Machine learning Algorithms) Result Evaluation Prediction • Train data • Data aggregation quarter-wise Trades data Roughly 6% of the clients responsible for ~80% of the loss Past Brokerage Number of Trades Margin Amount Exchange Ledger Amount Examples of features used Client profiles in terms of attributes computed from past trading. - Active clients = 1.03Mn - Active clients for which trade data is available = 346K - Average count of active clients who traded at least once during train period = 254K Prediction for quarter Jul – Sep 2015 Oct – Dec 2015 Accuracy 81.10% 78.30% Sensitivity 88.35% 75.42% Specificity 72.78% 81.90% Prevalence 53.4% 55.57% AUC 89.56% 88.21% Total clients 252845 255873 % Growth in Nifty -5.01% -0.03% Dormancy Prediction: predicts customers likely to stop trading • INR 1.6 M brokerage from 2,200 (11% of 20K - CRM assigned ) reactivated clients. • INR 309 K brokerage from 1,881 (4.8% of 39K – CRM not assigned ) reactivated clients
- 9. Page 9 © AlgoAnalytics All rights reserved Stock Based Recommender System Data Filtering •Discard Short Lived Sessions •Remove Rare Items •Consider only top ‘n’ most popular items Training and Testing •Training, Validation and Testing set •Deep learning •Final Recommendations Evaluation •Recall: Number of times actual next item in the sequence is in top ‘k’ recommendations Observed Recall@5 – 30.39% Last 5 stocks bought Recommended Stock Probability Intraday INFOSYS_TECHNOLOGIES Intraday State Bank of India 0.1629 Intraday MOTHERSON_SUMI_SYST EM Intraday L&T 0.137 Intraday UTI BANK LTD Longterm ICICI Banking Copora 0.124 Intraday ASIAN PAINT Intraday ICICI Banking Copora 0.0709 Intraday LIC HOUSING FIN Intraday Bharat Forge 0.061
- 10. Page 10 © AlgoAnalytics All rights reserved Cross Selling Mutual Funds Approach Inclination to Invest in MF •The inclination can be defined differently as suitable for business. •MF Buys/ (MF Buys + Equity TO) •MF Buys/ (MF Buys + Intraday TO) •Count of MF invested in. •Even a custom objective required by business. •Median can be used as threshold to label Features •Trading ratios {intraday, positional, short term, mid term, long term, open}. •% of TO in NIFTY-50. •Equity PnL, •% of Equity TO in total TO. •MF features like MF Buys. Value of MF objective in the previous period can be used as a feature. Modeling • Consecutive Period Setup: Example • Train Features: Jul – Dec 15 • Train Labels: Jan – Jun 16 • Test Features: Jan - Jun 16 • Test Labels: Jul - Dec 16 • Same Period Setup: Example, •Train Features, Label: Jul – Dec 15. •Test Features, Labels: Jan – Jun 16 Client universe has MF Buys > 0 in label period and Equity TO > 0 in feature period. Median is used as threshold to label. The same threshold used for train data is used as threshold for the test data. The approach is applied independently to each type of MF such as Debt, Equity & Hybrid using the same set of features. Problem Statement: Cross selling mutual fund given the equity portfolio and buy sells for clients
- 11. Page 11 © AlgoAnalytics All rights reserved MF Prediction: Predicts high inclination customers to buy MF MF Buys/ (MF Buys + Equity TO) for Debt MF MF Buys/ (MF Buys + Equity TO) for Equity MF MF Buys/ (MF Buys + Equity TO) for Hybrid MF 10 – fold CV Out of Sample 10 – fold CV Out of Sample 10 – fold CV Out of Sample Accuracy 70.23% 61.29% 74.43% 68.17% 70.59% 63.13% Kappa 40.46% 25.20% 48.86% 37.00% 41.18% 26.83% Sensitivity 75.77% 82.19% 75.30% 76.61% 78.55% 70.20% Specificity 64.69% 44.14% 73.56% 62.90% 62.63% 56.95% PPV 68.21% 54.68% 74.01% 56.28% 67.76% 58.77% NPV 72.75% 75.14% 74.86% 81.18% 74.49% 68.62% Prevalence 50.00% 45.06% 50.00% 38.40% 50.00% 46.64% AUC 68.01% 76.90% 67.01% Performances for Consecutive Period Model Setup
- 12. Page 12 © AlgoAnalytics All rights reserved Further Areas Where ML can be implemented First notice of loss in insurance Claim processing Fraud checking Policy Renewal Underwriting long term care and life insurance applications Matching records across various platforms while onboarding customer Monthly review of all accounts with zero balance, nil activity in lockers etc and automatic emails to accountholders Credit card operations Credit initiation: banks screening review Wire transfers checking for beneficiary details and against negative lists (AML/ frauds) Portfolio management Mortgage procession Post trade operations – payment record processing Trading record keeping compliance Monitoring trade performance INSURANCE BANKING BROKING
- 13. Page 13 © AlgoAnalytics All rights reserved Our Capabilities Achieve competitive advantage by leveraging analytics to improve decisions, enhance marketing, and influence consumer behavior. Predictive analytics to turbo-charge decisions across lines of business for marketing and risk management Strong Predictive modeling techniques using advanced vizualization Deep knowledge of wholesale banking & market place; and customer behavior Statistical skills Advanced Expertise in techniques like time series, decision trees clustering, linear regression, ensemble models Advanced AI techniques to derive interesting & unexpected insights from data for customer retention Help address the right questions, find the strategic answers to leverage your business
- 14. Interested in knowing more? July 12, 2017 Contact: info@algoanalytics.com