- Actuarial and risk
- 6 skills
- Data and analytics
- 8 skills
- Quantitative finance
- 6 skills
Actuarial and risk
A BSc in Actuarial Science from the University of the Free State, with ASSA exemptions A111, A112 and A113, and an Honours year in Risk Analysis.
Pricing with GLMs
A negative binomial GLM for claim frequency and a Gamma GLM for claim cost, combined into a risk premium with a 95% interval, validated on a holdout portfolio and by a coverage study over 200 simulated portfolios.
Profit testing and assumption checks
A funeral-insurance rider priced across 7,280 rating combinations, with a 61-year monthly profit test and sensitivity and scenario analysis.
Value at risk and expected shortfall
Four VaR models backtested with the Kupiec, Christoffersen and Basel traffic light tests, and expected shortfall with the Acerbi-Szekely test.
Stress testing
Historical crises replayed on current holdings, plus hypothetical shocks to correlation, volatility and the rand.
Tail dependence and economic capital
Copulas joining mortality and property-catastrophe losses, and what each dependence assumption does to diversification and capital.
Healthcare claims analytics
Risk signals and peer comparison that point investigators to medical scheme claims worth a closer look. In development.
Data and analytics
Most projects start with messy public data and end with a dashboard or app that other people can use.
Python and R
pandas, NumPy, SciPy and scikit-learn in Python; rugarch, gstat and Shiny in R. Where a model is ported from R to Python, tests check the two agree.
SQL
Queries on Google BigQuery to count GDELT news coverage of each bank by day.
Excel and VBA
Learned at university: a VBA module with its Excel work, and Excel models for CM2.
Data cleaning and validation
Removing bad price prints, rebuilding flight legs from gaps in receiver coverage, and data-quality checks on incoming claims.
Dashboards and visualisation
R Shiny apps, React dashboards and a D3 and MapLibre map with linked brushing.
Forecasting and time series
Arrival-time forecasts with empirical prediction intervals, and intervention analysis of tourism income.
Spatial statistics
Global and local Moran's I with permutation tests, variogram modelling and kriging.
Automated reporting and cloud
A monthly job that refreshes prices, reruns the analysis and redeploys the dashboard. The apps run on Azure App Service, AWS Lambda and CloudFront, with data queried on BigQuery.
Quantitative finance
Models of JSE markets, each tested on data it did not see, with costs included and pass marks written down in advance.
Volatility modelling
ARMA-GARCH, GJR-GARCH and EGARCH models with normal or Student t errors, chosen by BIC on training data only.
Copula dependence
Student t copulas and conditional probabilities that measure how far one share has moved relative to another.
Portfolio construction
Minimum variance, equal risk contribution and maximum Sharpe with Ledoit-Wolf covariance shrinkage, rebalanced monthly.
Walk-forward backtesting
Refitting on data up to each trading window, charging costs and transfer tax, and testing Sharpe ratio differences rather than reading them off a table.
Discrete choice modelling
Conditional logit with ridge regularisation, recovering investor risk preferences from choices between JSE sectors.
Options data
An explorer for contracts across strikes, expiries and calls or puts. MSc Financial Engineering coursework is under way at WorldQuant University.