Algorithmic Trading Strategies
Level up your career: Understanding advanced trading strategies, Impact of Machine Learning and methods for research into new alpha sources.
Algorithmic trading, a broad term for trading which uses mathematical models and algorithms, has become widespread and many markets are now dominated by high-frequency market-makers and proprietary trading shops. Some quant hedge funds have always relied on algorithms, while others have employed algorithms as part of their approach. Algorithms can be used for all aspects of the business, from trade decisions to portfolio allocations to placing market and limit orders with the time horizons involved being anywhere from a nanosecond to many days.
The aim is to use a scientific approach to develop efficient, scalable models with consistent and
Goals of Class
- Provide a strong foundation in the tools and techniques used in algorithmic trading.
- Cover everything from basic programming concepts to advanced trading strategies and methods for research into new alpha sources.
- Apply everything in hands-on projects throughout the course.
Algorithmic Trading Strategies
Course Running Time: 12 Hours 45 minutes
Goals: This course is for those who wish to
Professionals – Understand the mechanics of standard implementations of the single asset and portfolio based risk-premia trading strategies. Recognize pros and cons of various approaches to designing strategies and the common pitfalls encountered by algorithmic traders. Be able to devise new and improved algorithmic strategies.
Algorithmic Traders – Recognize the reasons commonly-used strategies work and when they don’t. Understand the statistical properties of strategies and discern the mathematically proven from the empirical. Acquire and improve methods to prevent overfitting.
Academics/students – Gain familiarity with the broad area of algorithmic trading strategies. Master the underlying theory and mechanics behind the most common strategies. Acquire the understanding of principals and context necessary for new academic research into the large number of open questions in the area.
About the Presenter:
Dr. Nick Firoozye is a mathematician & statistician with over 20 years of experience in the finance industry, in both buy and sell-side firms, largely in research. He started his career in Lehman Brothers doing MBS/ABS modeling, heading teams in portfolio strategy and EM quant research, later taking a variety of senior roles at Goldman Sachs, and Deutsche Bank, and at the asset managers, Sanford Bernstein, and Citadel, in areas ranging from quantitative strategy, relative value strategy and trading, to fixed income asset allocation. He is currently Managing Director and Head of Global Derivative Strategy, part of the Quantitative Strategy Group, at Nomura. He is currently an Honorary Senior Lecturer in Computer Science at University College London, focusing on Robust Machine Learning in finance. He recently co-authored a book, entitled Managing Uncertainty, Mitigating Risk, about the role of uncertainty and imprecise probability in finance, in light of the many recent financial crises, and he is writing a book on Algorithmic Trading Strategies based on his recent Ph.D. course on the same topic offered at UCL.
You will be able to receive 42.75 CPD points (12 hours 45 minutes of structured CPD and 30 hours of self-directed CPD) taking this course.
The CPD Certification Service was established in 1996 as the independent CPD accreditation institution operating across industry sectors to complement the CPD policies of professional and academic bodies. The CPD Certification Service provides recognised independent CPD accreditation compatible with global CPD principles.
Students will be expected to have a strong grounding in univariate and multivariate statistics. Time-series statistics (e.g., as taught in signal processing, econometrics) will be very useful but not mandatory. The course will be directed towards those with some finance experience (i.e., those working in finance or actively studying financial markets). Financial markets knowledge of the basics of equities, fixed income, fx and futures, and mean-variance optimisation is assumed, although we will cover some of the background material.
- Bachelors or Masters degree in
- Hard Sciences and Engineering
- Computer Science (with a firm understanding of mathematics)
- Economics or Finance (with a firm knowledge of econometrics)
Lecture 1: Overview, Math Background, Trend Following
(RUNNING TIME: 3 Hours 27 Minutes)
Quant trading as an industry
- Systematic Trading as an Industry:
- Structure of Quantitative/CTA market
- Trends in AUM
- Where to find out more
- Shared infrastructure for algo traders
- Platforms and APIs
- Python libraries
- NOSQL, etc
- Overview of Strategies
- Momentum or Trend Following
- Mean reversion
- Relative Value
- Vol Selling
- Statistical Arbitrage
- Equities Quant
- Overview of pitfalls
The basics: Quick Overview of Background Maths/Stats
- Time-series stats/econometrics (Discrete time)
- Stochastic Differential Equations (Continuous time)
Emphasis on modern fitting techniques, methods of solution, properties of solutions
Fads, Fancies and Trends
- Mathematical/Statistical PropertiesImpact on design – option value vs reactivity, skewness vs Sharpe
- Continuous time characterisations – power options and dependence on long-only Sharpes
- Discrete time – skewness term-structure, autocorrelation and volatility
- Non-linear filter – Impact and benefits
- Returns distribution engineering, limitations and further direction
- Momentum signals in practice.Timeseries vs Cross-sectional Momentum
- Crossing moving averages
- Technical indicators
- Econometric forecasting, ARIMA models
- On the street–CTAs and Quant Trend following vs Quant Equities
Lecture 2: Mean-Reversion and RV Trading
(RUNNING TIME: 3 Hours 18 Minutes)
- Mean Reversion
- Rationales: Liquidity provision or Overreaction
- Stationary vs Non-Stationary processes (traditional timeseries analysis)
- Univariate Tests – ADF, KPSS, Var-Ratio
- Multivariate tests – Johansen, Nyblom
- Cointegration and PCA
- Shortcomings – Time-variation
- Relation to Relative Value (RV) Trading
Catching falling knives
- RV Trading and its flavours
- I(1) vs I(0): RV vs Trend
- RV Trades in Delta-One space (pairs/spread, butterflies, baskets, etc)
- Timing Entry points and mean reversion. Optimising
- Stationarity: Are RV trades stationary? Macroeconomic trends and RV.
When trades go bad…
- Change-point Detection and Regime SwitchesStop Losses and scaling
- Breakpoint tests
- In Practice – OOS vs IS fits
- Stop Losses and scaling
Seeing things more simply
- Robust prediction using mean reversion
- Exponentially weighted moving averages (EWMA), Double exponential
- EWMA as Kalman filters
- Signal vs Noise
Lecture 3: Carry and Value and Portfolio Strategies
(RUNNING TIME: 2 Hours 57 Minutes)
When things stay the same
- Carry and Roll
- P vs Q measures.
- Carry as P measure expectation
- Calculating Carry and Roll
- Instruments: Futures, swaps, bonds, equities, fx, options
- Carry strategies and performance
- P vs Q measures.
- Elements of expected returns, Decompositions, and forecasting.
- How much carry can you expect to take home?
What should it really be worth?
- What is value?
- Value Trading, Value Investing, Valuations vs Pricing
- Measures of Value:
- Equities/Credit value
- Securitized transactions
- Long-term proxies for value outside equities
- Timeseries and Stationarity vs Horizon
- Recap: Timeseries – mean-reverting/trending/mean-reverting
Combining lots of things
- Measures of performance and risk:
- Sharpe, Sortino, Calmar
- Skewness, Kurtosis
- VaR, CVaR
- Downside Measures
- Portfolio Strategies
- MVO review
- Optimal Shape Ratios, Risk Parity and Min Variance
- MVO as regression – tests of optimality
- Bootstrap methods in MVO
- Other Methods
- Portfolio Strategy Design
- Forecasting vs optimal weights-measurement of ‘goodness’
- Objectives without theory – overfitting
- Sharpe Ratios – distributions and significance (t stats and asymptotic normality)
- MVO review
Lecture 4: Overfitting, Data snooping, and Rehash
(RUNNING TIME: 3 Hours)
Data snooping, P-Hacking and Bad Science
- Data snooping
- Type I vs Type II errors
- Snooping outcome – Bad models. Poor OOS performance
- P-hacking and non-reproducible results
- Irreproducibility crisis in science
- Overfitting in finance
- Standard fixes – Train/Test/Holdout
- Holdout overfitting, Kaggle competitions
- Multiple testing methods
- P-value adjustments
- Tests of Data snooping
- Various other studies / tests
- Data snooping and (Robust) Machine Learning
- Preventing data-snooping in practice
- Key take-aways
- Role of Machine Learning / Big Data
- Designing your own strategies
- Doing active research
- Next steps