Project 1

Title: Testing various methods of nonlinear principal component analysis (PCA) on financial time series           

Business: AVIVA       

Brief description: Apply traditional and nonlinear PCA on the overnight index swap (OIS) rates provided by the Bank of England. Investigate the forecasting performance of different time series models and select ones that are more robust to different time periods and at the same time maintain a low forecast error overall. 

Project 2

Title: Large loss prediction       

Business: QBE         

Brief description: Predict rare events (large losses, both in terms of frequency and severity), define the relationship between small and large losses, and find ways to rescale model forecasts generated from imbalanced data sets.

Project 1

Title: Bouquet of interest rate models

Business: Hymans Robertson

Brief description: Compare and contrast various stochastic interest rate models given their advantages/limitations, implementation steps etc. 

Project 2

Title: Calibration risk for interest rates

Business: Hymans Robertson

Brief description: Quantify the calibration risk of interest rate models and demonstrate their impact on company’s projections. 

Title: Correlation models for time series

Business: AVIVA       

Brief description: Identify the optimal length and the optimal frequency for the time series to calculate stable correlations. Establish a method to annualise the correlations calculated with daily or monthly observations. Develop a robust model able to forecast reliable future correlations. 

Undergraduate:

Business: Standard Chartered Bank        

Research work on China’s ESG development history and current status. How Chinese government has been setting this development as a long-term sustainable target (especially Green GBA and Green Shenzhen)? As investor sentiment on ESG is evolving rapidly, how may Standard Chartered better identify how capital is allocated and the appetite for SDGs across our markets’ presence? Conduct a main competitor analysis on market promotion and positioning (especially from competitors’ product offerings – product components, asset scale, social and business impact…etc).   Investigate how Standard Chartered can better identify the latest regulatory priorities and enable factors in supporting sustainable finance growth, especially in GBA and in Shenzhen.  

MSc:

Business: Capgemini, UK 

The Climate Change Act 2008 declares the UK’s target of net-zero carbon emissions by the year 2050, requiring significant transformation across industries to meet this objective. Focusing on motor vehicle trends in the UK, the rapid uptake of Electric Vehicle (EV) usage presents several challenges in sustaining such a growth rate, from vehicle fueling infrastructure to smoothing demand on the national grid. By contrast, while van traffic has grown by 71% over the last 20 years, uptake of alternatively fueled LCVs and HGVs is low, and reducing carbon emissions in this domain currently requires different strategies – for example, Artificial Intelligence and Analytics can be leveraged for optimising routing and distribution centre locations. Identify drivers and risk factors of the increasing market share of passenger and commercial EVs in the UK and produce a model forecasting the carbon emissions of the evolving profile of vehicles on the UK road network (considering the constraint of the Climate Change Act and other government targets). Produce three scenarios which would expedite the de-carbonisation of the UK road network and assess their respective feasibility. Supported by technical analysis, discuss how Artificial Intelligence methods could be leveraged to reduce the carbon emissions of the UK’s current fleet of LCVs and HGVs. 

Undergraduate:

Business: AVIVA

Aviva’s one of largest general and life insurance companies in the UK. One of the key risk factors driving Aviva performance stems from the stock markets. It is vital from a risk management perspective that Aviva develops reliable tools to assess the impact of the equity risk on its Total Balance sheet. At present, the most popular techniques to forecast the returns of the equity indices are based on the GARCH model framework. New models based on Artificial Neural Networks, more specifically, Recurrent Neural Networks (RNN) have emerged. The aim of this challenge is to compare the predictive power of both approaches on real data sets.    

MSc:

Business: IBM/Grillo

Almost one-third of the world’s population live in seismically active regions. For those in the strike zone of an impending earthquake, every second matters. An earthquake early warning (EEW) system limits the impact of future earthquakes on communities living nearby. The system detects an earthquake at its very beginning and rapidly issues alerts to users in its path. The alert outpaces strong earthquake shaking and may provide critical time to take basic protective actions such as seeking cover or exiting a building. The alerts can also trigger automatic systems that stop elevators, prepare backup power, and turn off gas pipes. This project gives students the opportunity to design and train a neural network for earthquake wave detection at Grillo seismic sensors, to contribute to the live development and testing of a revolutionary early earthquake warning system, and to help unlock seismic monitoring for developing nations.

Undergraduate:

Cyber risk and Hazard risk.

Challenge – Addressing a partially insurance or currently uninsurable risk.

Context: Every two years, Aon releases the Risk Management survey whereby Aon asked over 2,300 risk managers and c-suite professionals from 60 countries/territories and 16 industries about their key risks and how they manage and mitigate them. A detailed report is produced and the top 60 emerging risks are announced – some of which are either only partially insurable or uninsurable. (One of key the reasons being it is challenging to price the risk).

View the 2021 GRMS survey & results: Executive Summary - 2021 Global Risk Management Survey (aon.com)

Opportunity: Select one of the partially insurable or uninsurable risks listed and build a model utilizing open source data (or combination of Aon data) to better understand or ‘price’ the risk so it can be developed in the future into an insurable product.

MSc:

Parkison’s disease is a movement disorder manifesting itself as hypokinesia, slowness of movements. Parkinson’s patients are commonly assessed by MDS-UPDRS rating scale that consists of several exercises. This assessment can be captured by video and analysed using computer vision techniques. Deep learning methods allow the detection of keypoints such as fingertips, which in turn can be converted to a clinically relevant signal. One example of such a series is a Finger tapping signal, where the patient repeatedly taps the index finger and the thumb generating a periodic signal. Frequency, amplitude, or abnormalities in such movements are indicative of the disease. 

For this project we propose analysing these series for the detection of clinically relevant information:

1) Develop robust and automated peak detection algorithm for the periodic signals.

2) Develop a hesitation detection algorithm (anomaly when the patient slows or interrupts the movement for a short period).

As an extension of the project we have two follow-up tasks if time permits:

3) Clustering patients into clinically relevant subtypes using the clinical UPDRS scores. This analysis could be repeated for computer vision models developed by us for UPDRS to see if similar subgroups arise.

4) Supervised classification of the series according to the UPDRS score.

Undergraduate:

This year’s annual undergraduate Statistics Practitioners’ Challenge was provided by IDB Invest, the private sector arm of the Inter-American Development Bank Group.  IDB Invest is the private sector arm of the Inter-American Development Bank Group. It finances projects to advance clean energy, modernize agriculture, strengthen transportation systems and expand access to financing. It is committed to promoting economic growth and social inclusion in Latin America and the Caribbean. 

Challenge: To produce a model methodology and model results of credit default correlation factors for the Latin American and Caribbean region. The correlation factors should reflect the strength of the relationship between the credit quality of any two different companies in the Latin American and Caribbean region. These correlations will be used for Credit Economic Capital Calculations.  

MSc:

This year the challenge is provided by Allianz. Allianz is one of the UK's largest personal lines general insurers, with more than 7 million customers and 4,000 employees. They offer services direct to consumers, as well as through brokers, and through strategic partnerships with a wide range of organisations. They also own a number of other financial brands including Britannia Rescue, Highway Insurance, and Frizzell. 

Challenge: Poisson regression on highly zero-inflated claims counts. Claim frequency modelling is normally done by means of a Poisson regression to predict the number of claims a particular policy will have per year for a peril. More information on insurance risk modelling here. These have traditionally been already very skewed towards the zero counts which only get more pronounced when we begin modelling rarer perils for instance in home insurance space. The purpose of this project would be to investigate potential methods of dealing with this highly skewed Poisson regression problem.