Twelve Capital uses machine learning to enhance understanding of hurricane risk

Investment in technology is key for insurance-linked securities (ILS) managers looking to deliver high-performing portfolios of insurance and reinsurance linked investments, according to Twelve Capital.

The Swiss headquartered manager of catastrophe bond, collateralized reinsurance, private insurance debt and equity investment funds, believes that its own investments made in modelling elemental risks will help it to deliver enhanced returns going forwards.

Twelve Capital has been working alongside technology provider Reask, a specialist in the fields of catastrophe risk management, modelling and forecasting.

As part of the work to enhance Twelve Capital’s assessment of hurricane risk, the pair are testing out how advanced technologies such as machine-learning and automated pattern recognition methods can assist with the forecasting and analysis of hurricane systems.

Twelve Capital explained in a paper that Reask, “Applies state-of-the-art methods derived from artificial intelligence (AI) as pattern recognition systems without the need for explicit human input, as well as image processing, in order to forecast hurricane risk.”

The ILS fund manager further explained that using machine learning in this way means that models can recognise meteorological and climate-related patterns that could be relevant to formation and persistence of Atlantic hurricanes.

The work Reask is doing aims not just to predict the number of hurricanes that will form in a single year, but also to predict landfall potential and which geographic areas should be considered most at risk.

In addition, the work focuses on assessing the risk of clustering of hurricane systems within a specific hurricane season.

In working alongside Reask on this research into how machine learning can enhance our understanding of hurricane risk Twelve Capital is seeking to apply these methods and any learnings to the hurricane-exposed risk selection process for its ILS and catastrophe bond investment portfolios.

Twelve Capital is working to bring together human expertise and experience with computational power to provide it an edge when it comes to delivering ILS portfolios that perform for its investor clients.

“Clients who invest in ILS with Twelve Capital therefore gain an improved perspective of the riskiness of their assets,” the manager explained.

But Twelve Capital does not believe models are the answer alone, saying that no matter how sophisticated a risk model is, it will always contain assumptions and simplifications compared to real-world experience.

Because of this, “It is important to understand the strengths as well as the weaknesses of a model in order to derive conclusions with the appropriate level of confidence, while remaining conscious of its inherent level of uncertainty which ultimately drives the price to be asked of the protection buyer,” Twelve Capital believes.

The ILS fund manager said that the initial results of using machine learning to better its understanding of hurricane risk in the Atlantic “look promising.”

Twelve Capital is now looking into how this new data source provided by advanced technology can be put to work in helping it to better manage and optimise its portfolios of insurance and reinsurance related assets.

Putting the latest technology to work in understanding risk is vital for ILS funds, as it can provide them with an edge when it comes to risk selection, portfolio optimisation and also help them to reduce volatility across portfolios.

———–

Register today to attend our next ILS conference in Singapore, ILS Asia 2019.

We’re returning to Singapore for our fourth annual ILS market conference for the Asia region. Please register today to secure the best prices. The last few Early Bird tickets are still available.