24 Lessons in Quantitative Investing: Complete Syllabus
§ 24 Lessons in Quantitative Investing
Course Syllabus
I. Securities Fundamentals and Data Sources
This section covers the essential securities knowledge required for quantitative trading, such as price adjustment (adjusted prices). Adjusted prices are an unavoidable issue in quantitative trading and will permeate our entire course, yet many answers found online are incorrect. This section also reveals an unexpected issue: Python has problems with rounding.
1. Introduction
1.1. Development of Securities Investment and Quantitative Trading
1.2. Knowledge Framework for Quantitative Trading
1.3. Who is Suitable for Quantitative Trading?
1.4. Overview of Quantitative Strategies
1.4.1. Alpha Strategies
1.4.2. Market-Neutral Strategies
1.4.3. High-Frequency Arbitrage Strategies
1.4.4. Technical Analysis-Based Strategies
1.4.5. Strategy Research Methods
1.5. Course Content Overview
1.6. How to Study This Course
1.6.1. Prerequisites
1.6.2. Introduction to Online Quantitative Environments
2. Securities Fundamentals and Data Sources: Akshare
2.1. Exchanges and Security Codes
2.2. Knowledge of Price Adjustment (Adjusted Prices)
2.3. Akshare
2.3.1. Installing akshare in the Course Environment
2.3.2. Real-Time Stock Data
2.3.3. Historical Stock Data
2.3.4. Security Lists
2.3.5. Trading Calendars
2.3.6. Encapsulation and Improvement Suggestions
2.3.7. Exercises
3. Data Sources: Tushare, JqDataSdk
3.1. TUSHARE
3.1.1. Installation and Token Setup in Course Environment
3.1.2. Historical Stock Data
3.1.3. Security Lists
3.1.4. Trading Calendars
3.2. JoinQuant Local Data
3.2.1. Account Installation and Setup in Course Environment
3.2.2. Historical Stock Data
3.2.3. Security Lists
3.2.4. Trading Calendars
3.3. BAOSTOCK
3.3.1. Historical Stock Data
3.3.2. Security Lists
3.4. YFINANCE
The exercises in this section will highlight the linear transformation relationship between forward and backward adjusted prices.
4. Using Zillionare for Data Retrieval
4.1. Omicron
4.1.1. Initializing Omicron
4.1.2. Real-Time Stock Data
4.1.3. Historical Stock Data
4.1.4. Security Lists
4.1.5. Trading Calendars
4.1.6. Sector Data
4.2. Interpreting A-Share Data: Relationship Between Investor Count and Market Trends
5. Exercise Solutions (Video Only)
II. Introduction to Strategies
Lessons 6–8 constitute the second part of the course. We introduce three types of strategies from the perspectives of fundamentals, technicals, and trade execution, along with how to write strategy backtests and build a simple backtesting framework. In the examples, we also provide additional strategies, such as Connor's RSI strategy.
6. Small-Cap Strategy
6.1. Introduction to Small-Cap Strategy
6.2. Strategy Implementation - Manually Implementing the Simplest Backtest
6.2.1. Initialization
6.2.2. Plotting
6.2.3. Main Strategy Code
6.3. Strategy Optimization
6.3.1. Timing Optimization
6.3.2. Rule Optimization
6.3.3. Parameter Optimization
7. Bollinger Bands Strategy
7.1. Initialization Using Coursea
7.2. Bollinger Bands Strategy Based on Base Class - Frameworking the Backtest
7.3. Discussion on Strategy Optimization Directions
7.3.1. Parameter Optimization
7.3.2. Trend Judgment
8. Grid Trading
8.1. Grid Trading - A Strategy Without Timing
8.2. Code Implementation
8.2.1. Initialization
8.2.2. Evaluation Function
8.2.3. Strategy Behavior Analysis
8.3. Technical Implementation Issues
8.3.1. Order Price and Trading Timing
8.3.2. Stock Split and Bonus Issues
8.3.3. Trading Units
8.4. Strategy Optimization: From 1.37% to 79.8%!
8.4.1. Selecting Underlying Assets with "Bounds"
8.4.1.1. Large-Cap Stocks
8.4.1.2. Deeply Oversold Stocks
8.4.1.3. Convertible Bonds
8.4.2. Determining Grid Parameters Based on Historical Data
8.4.3. Improving Capital Efficiency
8.5. Trend-Following Grid
III. Data Analysis in Quantitative Trading: Theory and Implementation
Lessons 9–13 cover the fundamentals of quantitative analysis, focusing on the usage of libraries such as Numpy, Pandas, and Talib. The exercises in Chapter 9 provide many clever and commonly used Numpy exercises in quantitative trading.
9. Numpy and Pandas
9.1. Numpy
9.1.1. Creating Arrays
9.1.1.1. Vanilla Version
9.1.1.2. Pre-defined Special Arrays
9.1.1.3. Conversion from Existing Arrays
9.1.2. Inspecting Array Properties
9.1.3. Array Operations
9.1.3.1. Dimensionality Increase
9.1.3.2. Dimensionality Reduction
9.1.3.3. Transposition
9.1.3.4. Adding/Removing Elements
9.1.4. Logical Operations and Comparisons
9.1.5. Set Operations
9.1.6. Mathematical Operations
9.1.6.1. Dot Product
9.1.6.2. Aggregation Operations and Statistical Functions
9.1.7. Reading, Searching, and Lookup
9.1.7.1. Indexing and Slicing
9.1.7.2. Lookup, Replacement, and Filtering
9.1.8. Type Conversion and typing Module
9.1.9. Structured Arrays
9.1.10. IO
9.1.11. Common Functions in Quantitative Trading
9.1.11.1. REF(close, n)
9.1.11.2. EVERY(cond, n)
9.1.11.3. LAST(cond_list, n, m)
9.1.11.4. BARSLAST
9.1.11.5. CROSS
9.2. Pandas
9.2.1. Creation
9.2.2. Data Access
9.2.3. Iterating DataFrames
9.3. Pandas vs. Numpy
10. Ta-Lib
10.1. Installing Ta-Lib
10.1.1. Installing the Native Library
10.1.1.1. macOS
10.1.1.2. Linux
10.1.1.3. Windows
10.1.1.4. Using Conda
10.1.1.5. Third-Party Built Wheel Packages
10.1.2. Installing the Python Wrapper
10.2. Ta-Lib Overview
10.2.1. About the Documentation
10.2.2. Two Types of Interfaces
10.2.3. Method Overview
10.3. Common Indicator Functions
10.3.1. ATR
10.3.2. Moving Averages
10.3.2.1. SMA
10.3.2.2. EMA
10.3.2.3. WMA
10.3.3. Bollinger Bands
10.3.4. MACD
10.3.5. RSI
10.3.6. OBV (On-Balance Volume)
10.4. Pattern Recognition Functions
10.4.1. CDL3LINESTRIKE
10.4.2. CDL3WHITESOLDIERS
Statistics and probability play a crucial role in quantitative analysis. Lessons 11–12 review the most common statistical and probabilistic knowledge in quantitative trading, including moments from first to fourth, PDF/CDF, and covariance. The examples in this chapter solve problems such as whether to buy the dip when the Shanghai Composite Index drops by 4%.
11. Data Analysis and Python Implementation (1)
11.1. Examining Data Distribution
11.1.1. Finding the Center of Data
11.1.1.1. Mean and Centroid
11.1.1.2. Median
11.1.1.3. Mode
11.1.2. Measuring Data Dispersion
11.1.2.1. Quantiles
11.1.2.2. Variance and Standard Deviation
11.1.2.3. Frequency, PMF, PDF, CDF, PPF, and Histograms
11.1.2.4. Probability Density and Probability Density Function
11.1.2.5. Cumulative Probability and Cumulative Distribution Function (CDF)
11.1.2.6. CDF Estimation and Applications
11.1.2.7. Relationships Between Concepts
11.1.3. Distribution Shape of Data
11.1.4. Concept of Central Moments
11.1.5. Interpretation and Application of Skewness and Kurtosis in Investment
12. Data Analysis and Python Implementation (2)
12.1. Statistical Inference Methods
12.1.1. Quantile Plots
12.1.2. Hypothesis Testing Methods
12.2. Fitting, Regression, and Residuals
12.2.1. Residuals and Their Measurement
12.2.1.1. max_error
12.2.1.2. mean_absolute_error
12.2.1.3. mean_absolute_percentage_error
12.2.1.4. mean_squared_error
12.2.1.5. Rooted Mean Squared Error
12.2.2. Regression Analysis
12.3. Correlation
12.3.1. Covariance and Correlation Coefficient
12.3.2. Pearson and Spearman Correlation
12.3.3. Correlation Analysis Examples
12.4. Distance and Similarity
12.4.1. Listing Common Distance Definitions
12.4.2. How to Calculate Distance
12.5. Normalization
13. Practical Technical Analysis
Traditional technical analysis lacks strong theoretical support, but it is based on traders' experience and thus has its rationale. After learning Lessons 11–12, we apply our knowledge to technical analysis and discover new insights. Traditional technical analysis, empowered by statistical theory and adaptive parameters, significantly improves algorithm robustness and timing capabilities.
13.1. Box Detection
13.1.1. Statistical-Based Methods
13.1.2. Clustering-Based Algorithms
13.2. Finding Peaks and Valleys
13.2.1. Implementation in SciPy
13.2.2. Third-Party Library: Zigzag
13.2.3. Smoothing Curves
13.2.4. Double-Top Pattern Detection
13.2.5. Rounding Bottom Detection
13.3. Convexity/Concavity Detection
14. Factor Analysis
Factors are features with predictive power. Factor analysis and testing are rapid screening methods for their characteristics and are mandatory knowledge for entering quantitative institutions. We manually implement each step of factor analysis step-by-step, then introduce Alphalens, a common factor testing framework.
14.1. Factor Classification
14.2. Factor Analysis
14.2.1. Preprocessing
14.2.1.1. Outlier Clipping
14.2.1.2. Missing Values Handling
14.2.1.3. Distribution Adjustment
14.2.1.4. Standardization
14.2.1.5. Neutralization
14.3. Single-Factor Testing
14.3.1. Regression Method
14.3.1.1. Factor Evaluation via Regression
14.3.2. IC Analysis Method
14.3.2.1. Factor Evaluation via IC Analysis
14.3.3. Layered Backtest Method
14.3.4. Differences and Connections Among the Three Methods
14.4. Factor Evaluation System
15. Alphalens and Others
15.1. Alphalens
15.1.1. Alphalens Call Flow
15.1.2. Data Preprocessing
15.1.3. Factor Analysis
15.1.4. Common Alphalens Errors and Warnings
15.1.4.1. Timezone Issues
15.1.4.2. MaxLossExceedError
15.1.4.3. FutureWarning
15.2. JQFactor and jqfactor-analyzer
15.3. SymPy
15.4. Statistics
15.5. Statsmodels
15.5.1. OLS (Ordinary Least Squares) Estimation
15.5.2. Comparing OLS with RLM
15.5.3. ARIMA Models and Time Series Forecasting
15.6. Zipline
15.7. Pyfolio
15.8. TA
IV. Data Visualization
Plotting is not just for creating beautiful visualizations but also for unlocking the full potential of data and revealing hidden insights. This holds true in quantitative trading as well. We need the ability to plot candlestick charts and overlay backtest signals for strategy tuning, as well as generate backtest reports.
16. Matplotlib Plotting
16.1. Introduction to Matplotlib
16.2. How Plots Are Constructed
16.2.1. Top-Level Concepts
16.2.2. Relationships Between pyplot, Figure, and Axes
16.2.3. Layout
16.2.4. Figure Anatomy
16.3. High-Frequency Usage Objects
16.3.1. Axis
16.3.1.1. Spine Positioning and Hiding
16.3.1.2. Sharing X-Axis
16.3.1.3. Ticks
16.3.2. Text and Chinese Characters
16.3.3. Styles and Colors
16.3.3.1. Colormaps
17. Plotly Plotting
17.1. Basic Concepts in Plotly
17.2. Plotly Module Structure
17.2.1. Plotly Express
17.2.2. Graph Objects
17.2.3. Others
17.3. Comparing Plotly Express with go.Figure
17.4. Plotly Stock Analysis Chart Drawing
17.4.1. Candlestick Chart Drawing
17.4.2. Overlaying Technical Indicators
17.4.3. Subplots
17.4.4. Display Areas
17.4.5. Interactive Tooltips
17.5. Colors
17.5.1. Discrete Color Sequences
17.5.2. Continuous Color Scales
17.6. Themes and Templates
17.7. Introduction to Dash
17.7.1. Hello World
17.7.2. Connecting to Data
17.7.3. Adding Interactive Controls
17.7.4. Beautifying Applications
17.7.5. Deep Dive into Dash
18. Seaborn and PyEcharts Plotting
18.1. Seaborn
18.1.1. Seaborn Plotting Overview
18.1.1.1. Visualizing Statistical Relationships
18.1.1.2. Visualizing Data Distributions
18.1.1.3. Visualizing Bivariate Distributions
18.1.1.4. Joint and Marginal Distributions
18.1.1.5. Regression Fitting
18.1.2. Themes
18.1.3. Using Palettes
18.1.3.1. Qualitative Palettes
18.1.3.2. Continuous Palettes
18.1.3.3. Diverging Palettes
18.2. PyEcharts
18.2.1. Running in Notebook/JupyterLab
18.2.2. Calling Conventions
18.2.3. Using Options
18.2.4. Subplots and Layouts
18.2.4.1. Grid Layout
18.2.4.2. Page Layout
18.2.4.3. Tab Layout
18.2.4.4. Timeline
18.3. On Colors and Aesthetics
V. Backtesting Framework
Backtrader is the most famous open-source backtesting framework. Many institutions lack the R&D capability for proprietary quantitative investment research systems and often use backtrader internally for backtesting. We dedicate two lessons to explaining backtrader in depth.
19. Backtrader Backtesting Framework (1)
19.1. Quick Start
19.2. Backtrader Syntax Sugar
19.2.1. Time Series (Lines)
19.2.2. Operator Overloading
19.3. Data Feeds
19.3.1. GenericCSVData
19.3.2. Pandas Feed
19.3.3. Customizing a Feed
19.3.4. Adding New Data Columns
19.4. Multi-Timeframe Data
19.4.1. Comparing Multi-Timeframe Technical Indicators
19.5. Indicators
19.5.1. Built-In Indicator Library
19.5.2. Custom Indicators
19.5.2.1. Minimum Period
20. Backtrader Backtesting Framework (2)
20.1. Cerebro
20.1.1. Adding Loggers
20.1.2. Adding Observers
20.1.3. Execution and Plotting
20.2. Order
20.2.1. notify_order
20.3. Trading Agents
20.3.1. Querying Assets and Positions
20.3.2. Volume Limits
20.3.2.1. FixedSize
20.3.2.2. FixedBarPerc
20.3.2.3. BarPointPerc
20.3.3. Trading Timing - Cheat-On-Open
20.3.4. Trading Timing - Cheat-On-Close
20.3.5. Trading Functions
20.3.5.1. Standard Trading Functions
20.3.5.2. order_target Series
20.3.6. Portfolio Trading
20.3.7. OCO Orders
20.3.8. Slippage and Transaction Costs
20.3.8.1. Fixed Slippage
20.3.8.2. Percentage Slippage
20.3.9. Transaction Fees
20.4. Visualization
20.4.1. Observers
20.4.1.1. Broker Observer
20.4.1.2. BuySell Observer
20.4.1.3. Trade Observer
20.4.1.4. TimeReturn Observer
20.4.1.5. DrawDown Observer
20.4.1.6. Benchmark Observer
20.4.2. Custom Plotting
20.4.3. Collecting Backtest Data
20.5. Optimization
20.6. Summary
21. Strategy Backtest Evaluation
How to interpret backtest results? This is a strategy evaluation issue. Here we also answer a question: If your strategy goes live but performs below expectations, under what circumstances should you abort the strategy? This is a common interview question.
21.1. Return Rates
21.1.1. Simple Return Rate
21.1.2. Log Return Rate
21.1.3. Cumulative Returns
21.1.4. Aggregate Returns
21.1.5. Annual Return
21.2. Risk-Adjusted Returns
21.2.1. Sharpe Ratio
21.2.2. Relationship Between Sharpe Ratio and Asset Curve
21.2.3. Sortino Ratio
21.2.4. Max Drawdown
21.2.5. Relationship Between Sharpe and Max Drawdown
21.2.6. Annualized Volatility
21.2.7. Calmar Ratio
21.2.8. Omega Ratio
21.3. Benchmark-Related Metrics
21.3.1. Information Ratio
21.3.2. Alpha/Beta
21.4. Visualization of Strategy Evaluation
21.4.1. Metrics
21.4.2. Plots
21.4.3. Basic and Full
21.4.4. HTML
22. Backtesting Traps
Have you heard of the phenomenon where strategies "feast" during backtesting but "starve" in live trading? How is it caused? This section introduces backtesting traps, using rich practical experience to help you quickly compensate for insufficient live trading experience.
22.1. Survivorship Bias
22.2. Look-Ahead Bias
22.2.1. Reference Errors
22.2.2. Price Stealing
22.2.3. Look-Ahead Bias Caused by Price Adjustment
22.2.4. PIT (Point-in-Time) Data
22.4. Trading Rules
22.4.1. T+1 Trading
22.4.2. Price Limits (Up/Down Limits)
22.5. Overfitting
22.6. Backtest Duration
22.7. Differences Between Backtesting and Live Trading
22.7.1. Signal Flickering
22.7.2. Impact Costs
22.7.3. Impossible Execution Prices
22.7.4. Matching Issues
22.8. Monopoly Backtesting Framework
22.8.1. Backtest Function Overview
22.8.1.1. Architecture and Style
22.8.2. Strategy Framework
22.8.2.1. Data and Data Formats
22.8.2.2. Multi-Timeframe Data
22.8.2.3. Drive Mode and Performance
22.8.2.4. Backtest Reports
22.8.3. Complete Strategy Example
22.8.4. Parameter Optimization
22.9. References
VI. Integrating with Live Trading
All preparations are ultimately for integrating with live trading. The last two lessons will introduce various integration solutions.