Applied Machine Learning with R (Trading Use Case) - 2020

Define trading, explain trading instruments like futures, options, and currencies, and describe the roles of hedge funds, banks, and retail traders in global markets.

Explore trading instruments across stocks, futures, forex, and CFDs, focusing on crude oil futures (WTI). Learn about leverage, margins, and how a machine learning trading algorithm applies across markets.

Compare discretionary and systematic trading, showing how rule-based algorithms, tested strategies, and emotional discipline enable robust, objective, and scalable trading compared to manual decisions.

Explore quantitative trading fundamentals and backtesting with walk-forward testing. Apply machine learning with decision trees and random forests to data from Yahoo Finance and Bloomberg, considering risks.

Explore online sources for quantitative trading ideas, test and backtest ideas from papers, and use forums, screener tools, and community sites to inspire your own trading strategies.

Explore a trading hypothesis using crude oil futures daily open high low close data and features like moving-average position and candle type to predict next-day price direction with classification models.

Apply simple moving averages and exponential moving averages as features in a quantitative trading model, using daily crude oil data, and identify doji candles as potential signals.

Install the R programming language and the RStudio IDE by downloading the latest installer and using the free/open source version.

Navigate the interface, run simple expressions in the console, and define vectors by type (character, numeric, integer, logical) using the c function.

Learn to create matrices with matrix(), bind columns and rows with cbind and rbind, build a price data frame with return and price, and apply ifelse to derive kind column.

Compare free data sources with premium data, highlighting options like Yahoo Finance and Google Finance, brokers, and NinjaTrader, and show how to access data via Excel, CSV, or API.

Explore Yahoo finance market data across asset classes, view Apple and crude oil price charts, and learn to pull data into R with a finance package.

Export daily price data from NinjaTrader to a text file on your computer using the provided indicator, with crude oil futures data available in the course folder.

Identify and account for issues in free datasets, such as noisy, incomplete, and biased data. Learn to guard against look ahead bias, survivorship bias, and overfitting to ensure backtesting reliability.

Learn to use the quantmod package to retrieve Apple stock data from Yahoo Finance, plot price series, and add a 20-day exponential moving average.

Prepare and visualize financial data for machine learning in R. Clean and format price data into a time series with ohlc columns for trading models.

Load and format trading data for a time series, create price and moving average charts, and compare exponential moving averages to spot price trends in 2017-2018.

Create feature engineering pipelines by transforming daily open, high, low, and close data into feature columns. Derive moving averages, returns, and next-day targets to train and test predictive models.

Explore additional courses on profitable stock trading and investment strategies, cryptocurrency trading with unsupervised machine learning, and technical analysis to develop optimized trading strategies in finance.

Build a trading feature set in R by creating candle-based variables, previous-day data, dodgy candle rules, ema filters, and daily to next-day returns, all integrated into one data frame.

Explore feature engineering in R code, including removing the first 20 rows with empty values, cleaning data with moving averages, and building training/testing ranges, predictors, and target formulas for modeling.

Define machine learning and its problem types; explain supervised learning, classification and regression, data importance, feature engineering, and relate to the quantitative trading workflow.

Explore linear regression as a regression algorithm for predicting y from x using a linear relationship. Learn ordinary least squares, the best-fit line, and multivariate cases with multiple predictors.

Explore classification in machine learning by examining algorithms like logistic regression, decision trees, and random forests, and apply them to a trading strategy that predicts next-day price direction.

Explore the naive Bayes classifier, applying Bayes' theorem under a conditional independence assumption to predict the posterior probability of a class from features, with R examples for trading use.

Explore the course GitHub repository to access codes and algorithms, including data loading, feature engineering, machine learning training, testing, performance comparison, and portfolio generation for crude oil futures.

Train and evaluate a naive bayes algorithm on trading data, splitting into training and testing sets and using feature engineering to predict next-day candle direction and equity curves.

Explore building and evaluating a trading model using a confusion matrix to assess accuracy and misclassification error, discuss equity curves, and emphasize fixed-window training, testing, and validation.

Learn how support vector machines classify data into two classes by a hyperplane that maximizes the margin, using linear and nonlinear classifiers with support vectors for trading data.

Demonstrate training and testing a support vector machine in R using the e1071 package, exploring kernels to affect predictive performance and trading outcomes.

Learn how random forest, a supervised ensemble of decision trees, uses a packing method to boost prediction by combining models and voting for classification or regression.

Train and test a random forest model on the prepared trading data. Explore prediction performance, equity curves, and feature importance to understand model behavior.

Explore how random forests are trained and tested, examine error rates and accuracy across classes, and assess how the number of trees and feature importance influence model performance and overfitting.