agenda

Friday, August 31
7:00 a.m. – 8:00 a.m.

Session Coordinator(s)

Facilitator(s)

8:00 a.m. – 8:15 a.m.

Presenter(s): Kitty Ching, FSA, Vice President, The Actuarial Institute of Chinese Taipei

Session Coordinator(s)

Facilitator(s)

8:15 a.m. – 9:05 a.m.

Presentation(s): View Presentation

Presenter(s): Marco Hoogendijk, Managing Director Asia, Ortec Finance Asia Ltd

Economies and financial markets are well known for their continuous changing nature leading to significant up and down movements. A better understanding of the drivers behind these fluctuations contribute to better investment decisions. In one world view, these fluctuations are driven by cyclicality. This session will investigate how predictive analytics can lead to better estimating two types of cycles: the well-known business cycle and the less well-known, but important financial cycle -the latter building on the seminal ideas of Hyman Minsky. This session will explore cycle estimates, test their sensitivity and ask how these cycles matter for the risk and return on asset classes within a prudent investment framework for life insurance companies.  

 

Session Coordinator(s)

Facilitator(s)

9:05 a.m. – 9:55 a.m.

Presentation(s): View Presentation

Presenter(s): James Lin, FIAA, Senior Manager, Ernst & Young

Session Coordinator(s)

Facilitator(s)

10:15 a.m. – 11:05 a.m.

Presentation(s): View Presentation

Presenter(s): Yi-Ling Lin, FSA, MAAA, FCA, Principal, The Terry Group

A world is abuzz with hot topics of predictive analytics, machine learning and big data. This session will demystify some of it through case studies in health care. Traditional data analysis, focusing on what happened and why, is expanding to include predictive and decision-empowering data analytics to help solve business problems related to future outcomes. Emerging new data sources, sophisticated analytical tools, powerful computers, as well as traditional statistical approaches and data visualization lend a powerful hand in innovative solutions to complex business problems. In this session, case studies will be used to illustrate solving business problems in the health care arena by progressing through the data analytics spectrum, from hindsight to insight to foresight. Drawing from an extensive data analytics toolbox, applications of traditional and emerging methodologies, such as linear and logit regression, data visualization, and random forest model will be shown.

After the session, attendees will be able to describe the steps in the data analytics spectrum, develop a plan and apply data analytics to business problems, and interpret and evaluate results from different types of predictive models.

Session Coordinator(s)

Facilitator(s)

11:05 a.m. – 12:40 p.m.

Session Coordinator(s)

Facilitator(s)

12:40 p.m. – 1:30 p.m.

Presentation(s): View Presentation

Presenter(s): Richard Liao, ASA, Associate Actuary, Milliman; Stanley Hsieh, Actuarial Analyst, Milliman

 Life insurance companies rely on experience studies for predicting lapse rates of insurance policies. Traditional actuarial lapse studies require subjective judgement and consider only limited number of factors such as policy durations and product grouping. For life insurance companies to improve the prediction of lapse behavior, better methods and more factors, such as more internal policy related data or external economic data, should be considered. With the improvement in technology and advent of big data, machine learning techniques can be applied for the prediction of lapse rate. In this session, how machine learning technology is being applied to predicting lapse rate by going through a real-life case study will be discussed. We will introduce machine learning methods which are being applied on the case study, including Generalized Linear Model, Decision Trees, Random Forest and Gradient Boosting Machine. How the machine learning models generate a more accurate result than traditional experience study will be demonstrated. The session will conclude by discussing what life insurance companies must do to start taking advantage of data analytics.

Session Coordinator(s)

Facilitator(s)

1:30 p.m. – 2:20 p.m.

Presentation(s): View Presentation

Presenter(s): Yun-Tien Lee, FSA, Consulting Actuary, Milliman; Janice Wang, ASA, CERA, Actuarial Associate, Milliman

How predictive analytics can benefit enterprise risk management (ERM), with a specific application to policyholder behavior risk will be discussed. This session will focus on identifying risks related to policyholder behavior and setting up procedures for active monitoring and management of those risks. The case study considered will relate to risks among variable annuities, but the topics addressed will relate to all insurance products. After the session, attendees will know ways in which predictive analytics can benefit ERM as it relates to actuarial assumptions and associated risks.

Session Coordinator(s)

Facilitator(s)

2:20 p.m. – 2:45 p.m.

Session Coordinator(s)

Facilitator(s)

2:45 p.m. – 3:35 p.m.

Presentation(s): View Presentation

Presenter(s): Alfred Ma, PhD, CFA, ASA, Executive Director & Chief Investment Officer, CASH Algo Finance Group

The session is mainly divided into two parts. The first part of the session will include some predictive analytics for risk management in financial trading especially in algorithmic trading setting will be introduced. It involves cutting edge techniques in machine learning including support vector machines and deep learning models like Long Short-Term Memory model. In the second part of the session, some ex post statistical analysis tools will also be introduced for risk management under a post-trade session. A few standard statistical tests such as Welch’s t test, Mann-Whitney U test, Welch’s T test on ranked data and Kolmogorov–Smirnov test will be introduced and discussed regarding the advantages and disadvantages.  At the end of the session, audiences would learn how to apply machine learning models in risk management setting Design risk management framework for other setting based on machine learning models Apply statistical tests on ex post performance measures and evaluate the risk involved Explain the weaknesses of each statistical tests on various application and select the best choice for a situation.

Session Coordinator(s)

Facilitator(s)

3:35 p.m. – 3:45 p.m.