Course Agenda

Agenda

Agenda: Fundamentals of Machine Learning - London

Day one: Wednesday, June 17, 2020

08:3009:00

Registration and refreshments

08:30 - 09:00

09:0010:30

Introducing machine learning (ML)

09:00 - 10:30

  • Differences between modern statistics, ML and AI

  • Core components of the ML process

  • How does ML apply to financial organisations?

  • Current applications of ML

  • How ML is changing financial markets

Dr Richard Saldanha

Founder & Co-Head

Oxquant

Dr Richard Saldanha is Founder and Co-Head of Oxquant, a consultancy firm that provides expertise and advice on risk management, investments and the impact of artificial intelligence. He is also an Independent Adviser to Oxford Portfolio Advisers.

Richard has over 20 years’ experience in asset management and investment banking in the areas of risk, trading and investments. In particular, he was Global Head of Risk at Investec Asset Management (2010–15), headed a quantitative global macro initiative for the same firm (2006–09) and founded and ran his own hedge fund Oxquant Capital Partners (2004–06).

In addition to his consulting and advisory activities, Richard lectures on statistical machine learning and its applications in finance at Queen Mary University of London. He attended Oriel College, University of Oxford, and holds a doctorate (DPhil) in statistics.

Drago Indjic

Data Science Practitioner

Oxquant

Drago’s hedge funds career continued in fintech as a data science practitioner at Oxquant (Oxford; FCA #232160), co-founder of Soft-Finance (Geneva), Technological Partnership (Belgrade) and Richfox Capital (Amsterdam). ETFmatic.com “robo” app runs his implementation of a fully digital investment process, following quant/PM roles in various hedge funds, a sovereign wealth fund and a multi-family office, Drago has co-edited the Legaltech book for Wiley (2020) and lectures part-time at Queen Mary and Regents. Member of the IEEE and IET, PhD (Imperial College), Dipl Ing (Belgrade), fintech and AI funding reviewer for the EU and @dindjic on Twitter.

10:3010:45

Morning break

10:30 - 11:00

10:4512:00

Understanding data

10:45 - 12:00

  • The importance of identifying appropriate data sources and types

    • Alternative data: smart city, spatial and flow data, mobile sensor data (gestures, movement)

    • Semantic data: text, documents 

  • Data quality, quantity and representativeness 

  • The effect of data quality on ML output

  • Data engineering and data quality management

    • The ETL process: Extracting - Transforming - Loading data

    • Data models

  • Analysing data

    • Exploratory data analysis

    • Visualisation 

  • Choosing the best data set 

    • Available data or trial study design

    • Active experimentation 

Dr Richard Saldanha

Founder & Co-Head

Oxquant

Dr Richard Saldanha is Founder and Co-Head of Oxquant, a consultancy firm that provides expertise and advice on risk management, investments and the impact of artificial intelligence. He is also an Independent Adviser to Oxford Portfolio Advisers.

Richard has over 20 years’ experience in asset management and investment banking in the areas of risk, trading and investments. In particular, he was Global Head of Risk at Investec Asset Management (2010–15), headed a quantitative global macro initiative for the same firm (2006–09) and founded and ran his own hedge fund Oxquant Capital Partners (2004–06).

In addition to his consulting and advisory activities, Richard lectures on statistical machine learning and its applications in finance at Queen Mary University of London. He attended Oriel College, University of Oxford, and holds a doctorate (DPhil) in statistics.

Drago Indjic

Data Science Practitioner

Oxquant

Drago’s hedge funds career continued in fintech as a data science practitioner at Oxquant (Oxford; FCA #232160), co-founder of Soft-Finance (Geneva), Technological Partnership (Belgrade) and Richfox Capital (Amsterdam). ETFmatic.com “robo” app runs his implementation of a fully digital investment process, following quant/PM roles in various hedge funds, a sovereign wealth fund and a multi-family office, Drago has co-edited the Legaltech book for Wiley (2020) and lectures part-time at Queen Mary and Regents. Member of the IEEE and IET, PhD (Imperial College), Dipl Ing (Belgrade), fintech and AI funding reviewer for the EU and @dindjic on Twitter.

12:0013:00

Lunch

12:30 - 13:30

13:0014:30

Objective function

13:00 - 14:30

  • What is an objective function?

  • Representing the business problem in mathematical terms

  • Measuring the sensitivity of a solution 

  • Types of objective function

    • Regression and classification

    • Advanced: text, fidelity, trajectory mismatch

  • Specifying constraints: custom business vs compliance/regulatory 

  • Specifying the objective function based on outcomes 

    • Risk and loss functions – investment, credit

    • Ratings – client analytics (churn, fraud, AdTech), ESG 

  • Model selection

    • Principles of model and variable selection

    • Confidence estimation

    • Determining the complexity of a model

    • Training speed and scalability

Dr Richard Saldanha

Founder & Co-Head

Oxquant

Dr Richard Saldanha is Founder and Co-Head of Oxquant, a consultancy firm that provides expertise and advice on risk management, investments and the impact of artificial intelligence. He is also an Independent Adviser to Oxford Portfolio Advisers.

Richard has over 20 years’ experience in asset management and investment banking in the areas of risk, trading and investments. In particular, he was Global Head of Risk at Investec Asset Management (2010–15), headed a quantitative global macro initiative for the same firm (2006–09) and founded and ran his own hedge fund Oxquant Capital Partners (2004–06).

In addition to his consulting and advisory activities, Richard lectures on statistical machine learning and its applications in finance at Queen Mary University of London. He attended Oriel College, University of Oxford, and holds a doctorate (DPhil) in statistics.

Drago Indjic

Data Science Practitioner

Oxquant

Drago’s hedge funds career continued in fintech as a data science practitioner at Oxquant (Oxford; FCA #232160), co-founder of Soft-Finance (Geneva), Technological Partnership (Belgrade) and Richfox Capital (Amsterdam). ETFmatic.com “robo” app runs his implementation of a fully digital investment process, following quant/PM roles in various hedge funds, a sovereign wealth fund and a multi-family office, Drago has co-edited the Legaltech book for Wiley (2020) and lectures part-time at Queen Mary and Regents. Member of the IEEE and IET, PhD (Imperial College), Dipl Ing (Belgrade), fintech and AI funding reviewer for the EU and @dindjic on Twitter.

14:3014:45

Afternoon break

15:00 - 15:30

14:4516:15

From modern statistics to machine learning models

14:45 - 16:15

  • From linear regression to nearest neighbours and decision trees

  • Ensembles of models

  • Supervised and unsupervised learning 

  • Reinforcement learning 

  • Deep learning

  • Tackling the overfitting problem and tuning hyper parameters

Dr Richard Saldanha

Founder & Co-Head

Oxquant

Dr Richard Saldanha is Founder and Co-Head of Oxquant, a consultancy firm that provides expertise and advice on risk management, investments and the impact of artificial intelligence. He is also an Independent Adviser to Oxford Portfolio Advisers.

Richard has over 20 years’ experience in asset management and investment banking in the areas of risk, trading and investments. In particular, he was Global Head of Risk at Investec Asset Management (2010–15), headed a quantitative global macro initiative for the same firm (2006–09) and founded and ran his own hedge fund Oxquant Capital Partners (2004–06).

In addition to his consulting and advisory activities, Richard lectures on statistical machine learning and its applications in finance at Queen Mary University of London. He attended Oriel College, University of Oxford, and holds a doctorate (DPhil) in statistics.

Drago Indjic

Data Science Practitioner

Oxquant

Drago’s hedge funds career continued in fintech as a data science practitioner at Oxquant (Oxford; FCA #232160), co-founder of Soft-Finance (Geneva), Technological Partnership (Belgrade) and Richfox Capital (Amsterdam). ETFmatic.com “robo” app runs his implementation of a fully digital investment process, following quant/PM roles in various hedge funds, a sovereign wealth fund and a multi-family office, Drago has co-edited the Legaltech book for Wiley (2020) and lectures part-time at Queen Mary and Regents. Member of the IEEE and IET, PhD (Imperial College), Dipl Ing (Belgrade), fintech and AI funding reviewer for the EU and @dindjic on Twitter.

16:1516:15

End of day one

17:00 - 17:30

Day two: Thursday, June 18, 2020

08:3009:00

Morning Refreshments

08:30 - 09:00

09:0010:30

Interpreting and validating a model

09:00 - 10:00

  • Automated model building and benchmarking: AutoML, EvalAI suites and tournaments

  • Transparency and explanation: Understanding why a model provides a given or fair prediction

  • Deep learning: challenges in leveraging state-of-the-art approaches 

  • The requirement of explainable models from a regulatory perspective 

  • Approaches to satisfy explainability 

10:3010:45

Morning break

10:00 - 10:30

10:4512:00

Machine learning in investment management and portfolio optimisation

10:30 - 12:00

  • Data science & machine learning in quantitative investment management

  • The impact ML has on portfolio optimisation 

  • Multi period portfolio optimisation and asset allocation frameworks

  • New tools and techniques in large-scale machine learning and analytics

  • ML based portfolio hedging strategies

12:0013:00

Lunch

12:00 - 13:00

13:0014:30

ML for risk practitioners

13:00 - 14:30

  • Identifying and monitoring risk

  • Applications of ML in different risk areas

  • Reducing credit risk

  • Enforcing regulatory compliance: risk profiling and fairness

  • Mitigating MLs own risk to the business model

14:3014:45

Afternoon break

14:30 - 15:00

14:4516:05

Machine learning in finance: putting it into practice

15:00 - 16:30

  • ML for quantitative investment: challenges and opportunities 

  • Where does ML have more chance of winning? 

  • Start with a baseline model 

  • Two examples: 

    • Forecasting dividend ends 

    • Building a stock selection model 

  • Future applications of machine learning in finance

16:1516:15

End of course

16:30 - 17:00