AI for Investment Strategies

• Weekdays – 8 weeks
• Weekend – 10 weeks
• FastTrack – 6 weeks

• Weekdays – 8 weeks
• Weekend – 10 weeks
• FastTrack – 6 weeks

• Introduction
• Stock Prices
• Market Mechanics
• Data Processing
• Stock Returns
• Momentum Trading

• Outliers and Filtering Signals
• Regression
• Time Series Modeling
• Volatility
• Pairs Trading and Mean Reversion

• Stocks, Indices and Funds
• ETFs
• Portfolio Risk and Return
• Portfolio Optimization

• Factors Models of Returns
• Risk Factor Models
• Alpha Factors
• Advanced Portfolio Optimization with Risk and Alpha Factors Models

• Setting Up Hardware & Software
• Choose PC or Virtual Private Server
• Note on the trading restrictions for Retail Traders
• Install Trading Terminal
• Test and Production Terminals
• Install version control software: Git Desktop
• Organize your Version Control Repositories
• Install R
• Decision Support with R Statistical Software
• All functions in R package 'Lazytrade'
• Launch MT4 terminal with options

• Understanding Programming Environment
• Getting to know our Programming Environment MQ1.4 Editor
• Robot main structure
• Types of Algo trading systems: Rule, Model, Hybrid based

• R package Lazytrade
• Our Robot Template
• Logging trading results to file
• Reading commands from Decision Support System
• How to understand and modify this robot?
• Adding option to close all positions on Friday evening
• Using MetaTrader Terminal Profiles to manage different setups
• Keeping MT Terminals Profiles under Version Control

• FALCON_D - simple trading robot
• Why Periodic Optimization?
• Optimization Method P1. Settings overview
• Optimization Method P2. Collecting data during Trades Simulation
• Analyse simulated trades
• Evaluation of results
• Activity to practice

• Introduction to Statistically Control Trades and Basic Principles
• Link to Big Strategy and Profit Factor Monitoring
• R package 'Lazytrade'
• Statistical Trading Analysis and Control
• Statistical Analysis of Trading Results and Control of Trading Robots
• Why to write Functions. Deep Dive

• Reinforcement Learning - Generic Example
• Reinforcement Learning - Adapting Example to Trading problem
• Reinforcement Learning -Trade Trigger
• Reinforcement Learning - Function Get RL Policy
• Reinforcement Learning - Function Record Policy
• Simulate performance of Reinforcement Learning
• Adaptive Reinforcement Learning control parameters
• Adapted Trade Trigger script. Reading R data files into R
• Writing best control parameters.
• Writing Optimal control parameters. P1 (Making 'for' loop)
• Writing Optimal control parameters. P2 (Log data within a function)
• Writing Optimal control parameters. P3 (Data manipulation using 'pipes)

• Introduction to Social Media
• Building a Sustainable Social Media Strategy
• Converged Social Media
• Content Marketing in a Social Media World
• Visual Social Media
• Understanding and Sparking Social Sharing
• Storytelling an Essential Part of Your Social Narrative
• Influencer Marketing and Online Reputation Management
• Social Media Selling
• Social Media Measurement

• Intro to Natural Language Processing
• Text Processing
• Feature Extraction
• Financial Statements
• Basic NLP Analysis
• Analyze corporate filings and generate sentiment-based trading signals

• Introduction
• Web scrapping and Sentiment Analysis. Basic Principles
• What is Sentiment analysis? Polarity Scoring
• Reading Macroeconomic news and limit trades
• Code Overview: read news from Forex Factory
• Reading news on the Schedule (automate it!)
• Reading News and Derive Country Sentiment
• Any good News? Idea of the Forex Social Research!
• Code review: News Sentiment for Decision Support System
• Analysing Twitter Sentiment Predictive Model
• Setup Developer Account in Twitter
• Create an App in twitter developer
• Getting code for this section
• Encrypt Tokens in R Script
• Getting twitter data into R
• Extract Knowledge from Text: Searching and - manipulating relevant Twitter content
• Knowledge from Text
• Searching tweets by term methods
• Manipulate and visualize the tweeter content
• More deep Sentiment Analysis of text
• Predict future price change based on sentiment pattern
• Correlating Sentiment Pattern to price change
• Script to collect sentiment score and price data
• Script to train the deep learning model
• Script to test the deep learning model
• Script to predict price change using model and sentiment scores

• Introduction to Idea of Market Status Detection with Artificial - Intelligence?
• Why and How to detect Market Status with Artificial Intelligence
• Deep Learning architecture in R [h2o.ai]
• Practical development of Deep Learning Model capable to detect 6 market types
• R package 'Lazytrade'
• Logging data from financial Assets
• Prepare Data for Deep Learning Model
• Install h2o to your computer
• Solving Supervised Regression problems with Deep Learning
• Solving Supervised Classification problems with Deep Learning

• Introduction to the Practical Section
• Predicting using the model: Productionize Deep Learning application
• Evaluating results of Market Type prediction
• Automate deep learning model scoring
• Continuous improvement of Deep Learning Model
• Script to Train the model first time
• Script to Score [and collect more] data
• Script to Update the model

• Script to Select Initial Data
• Script to Train the model first time
• Script to Score [and collect more] data
• Script to Update the model

• Introduction to Neural Networks
• Training Neural Networks
• Deep Learning with Tensorflow
• Recurrent Neural Networks
• Embeddings & Word2Vec
• Sentiment Prediction RNN & LSTM

• Overview
• Decision Trees
• Model Testing and Evaluation
• Random Forests
• Feature Engineering
• Overlapping Labels
• Feature Importance

• Intro to Backtesting
• Optimization with Transaction Costs
• Attribution

Problem Statement: - Optimizing portfolio using Beta methodology

Description: - Here we need to create multiple portfolios using smart beta methodology and optimization. The portfolio performances will also be tracked by error tracking. Calculation of turnover of portfolios will let us understand the best timings to rebalance. We will figure out portfolio weights on analysing fundamental data and by quadratic programming.

Problem Statement: - Apply NLP on corporate filings to analyse sentiment

Description: - This project leads us to work with Natural Language Processing (NLP) on corporate filings and statements for cleaning data and text processing to feature extraction and modelling. We can use bag-of-words and TF-IDF to figure out company specific sentiments. Based upon sentiments the investment decision can be taken like where to invest or better time to buy or sell.

Problem Statement: - Configure a trading robot to make decision on analysing different factors.

Description: - Falcon_S or Falcon_A trding robot can be utilized to analyse news sentiments and define threshold and update the same. This is project to work with deep neural networks to process and interpret news data. We will apply backtests with model over news data for generating the signal.