Pricing & portfolio management

Sophisticated statistical modelling and machine learning techniques balanced with the commercial imperative.

We build technical models for attritional cost, customer demand and competitor deconstructions, which underpin the pricing for many insurers and financiers in Australia and New Zealand.

In addition to off-the-shelf techniques and bespoke machine learning algorithms designed for core components of pricing modelling resulting in enhanced predictive power, and ultimately better pricing outcomes.

Our extensive decision support tools enable clients to make better informed pricing decisions.

An effective, transparent framework which harnesses the value of your full arsenal of predictive models is key to a value maximising process.

We have experience with a wide range of commercially available optimisation algorithms and can also offer a unique approach to price optimisation successfully deployed by a number of insurers, which consistently delivers improved business performance across a varied range of pricing objectives. 

We have a vision – the portfolio manager turns the computer on in the morning and the monitoring system gives immediate feedback about changes in the portfolio, or deviation from budget projections at a granular, segment level. With this ultimate goal in mind, our machine learning algorithms used for portfolio monitoring go a long way to making this a reality.

We hear about great BI systems that can be sliced and diced by any combination of factors, which lay unused by time-poor portfolio managers. Identifying changes in trend or deviation from projections by key segments saves a lot of time in identifying the “story behind the numbers”.

Strong predictive models are at the heart of good pricing decisions - but the predictive power of even the best model can deteriorate over time as market conditions evolve. Knowing when change happens and how to quickly adapt models is essential.

We have developed a model monitoring framework for this purpose. This includes metrics developed specifically for model monitoring as standard model fit metrics are not necessarily commercially appropriate.

We also use machine learning techniques to identify poorly fitting segments, enabling efficient model augmentation. 

We have access to unique industry data, bespoke templates and can help you answer strategic queries, such as; Should I expand my product range over the next 5-10 years? How? 

We take the time to learn your business and to understand your objectives, helping you consider important scenarios such as trends in competition, price sensitivity, market demand, regulation, vehicle safety, self driving cars, urbanisation and other social trends.

The Commercial Lines classes present challenges, including poor, scarce or incomplete data, the impact of large claims and the greater variety in the types of risk being insured.

We recognise the importance of underwriter judgement, whether at the individual risk level or in setting book premium rates, and typically work closely with the underwriter in the course of our work in Commercial Lines.