RSSeNEWS to be determined...

The West Midlands local group meeting on 3 February at Warwick University was a joint event with the Young Statisticians Section in which there were two talks.

The first talk titled ‘A Bayesian approach for prediction error in chain-ladder claims reserving’ was given by Ji Yao. Ji gave a brief presentation on the application of both classical and Bayesian statistical tools in actuarial science. Specifically, he explained how to measure the prediction error in chain-ladder claims reserving in the field of general insurance using the mean square error (MSE) for both Bayesian and classical approaches.

The chain ladder method and Mack’s model were presented under the classical approach while under the Bayesian approach different priors (proper and improper) were specified for the parameters of the underlying model. A numerical example based on ten years claims reserve data from general insurance was used to illustrate these methods. Ji finished his talk seeking some suggestions on two things: whether MSE could be used under the Bayesian approach and whether using proper priors for parameters under Bayesian approach was reasonable.

The second talk was given by Giuliana Bordigoni from AHL, part of the MAN plc group. Giuliana gave an overview of the hedge fund industry and then moved on to the applications of adaptive data cleaning in systematic trading in particular. The main purpose is to develop an algorithm to deal with market data, which can be used to produce forecasts for market traders. However, the main challenge here is data cleaning.

Giuliana pointed out a number of types of errors such as spikes, stale price and change of scale. She then stressed that the algorithm must be flexible, fast, and sufficiently robust. She then moved on to a few types of filters and their combination that are currently used in the industry to deal with high frequency data, showing partial results with a couple of examples from Eurostoxx, the FTSE index and natural gas.

Report by Hasinur Rahaman Khan and Duy Pham