Download presentation
Presentation is loading. Please wait.
Published byDrusilla Greer Modified over 9 years ago
1
Dependency modelling Data Challenge 12 July 2007
2
Welcome! Short introduction to Quantitative Finance Network Lighthill Risk Network Launch of the Data Challenge Next Steps Pub? Counts as 1.5 hours CPD! “Harness collective intelligence”
3
Quantitative Finance Network Quant & Mammon Study (1999) Centre for the Study of Financial Innovation (CSFI) Quantitative Finance Initiative (QFI) Engineering & Physical Sciences Research Council (EPSRC) Faculty and Institute of Actuaries Four Rounds of Funding (2001 to 2004) EPSRC’s Sustainability Objective Switch from Directed to Responsive Mode To build a Network “Pick the best tool for the job”
4
Quantitative Finance Network Objectives: To build a sustainable network linking researchers and practitioners to support the UK’s Financial Services sector To identify challenges and opportunities to initiate and support research to present funding opportunities To foster collaboration Branding for Profession’s knowledge transfer activities CPD, research groups and workshops “Search out and use prior art”
5
Quantitative Finance Network “If you find a useful tool…”
6
Lighthill Risk Network “…post it and tell the world” Major initiative with other core members Lloyd’s, Benfield, RMS, Guy Carpenter, Catlin Bridge between academia and industry Offers “expert panels” on various risk areas of interest to insurers: Catastrophe modelling Climate Change http://www.lighthillrisknetwork.org http://www.lighthillrisknetwork.org
7
Lighthill Risk Network Not for profit, set up Jan 2006, launched at Lloyd’s in April 2007 Based at the Lighthill Institute of Mathematical Sciences (LIMS) and administered within UCL Role as a facilitator not a sponsor of research No equivalent organisation exists, nearest are: QFN (Quantitative Finance Network) funded by EPSRC and run by Institute of Actuaries RPI (Risk Prediction Initiative) which sponsors research from Bermuda “We need a new relationship with IT”
8
Lighthill Risk Network “Harness collective intelligence”
9
Dependency modelling Timetable 12 July 2007 : Launch Pseudo Data 28 September 2007 : Solutions submitted to us 30 October : Climb the learning curve 8 November : Results and Keynote Speech “Pick the best tool for the job”
10
The data challenge 10 years aggregate claims experience Psuedo classes of business Property - Direct Property - Treaty Directors & Officers/Professional indemnity Workers Compensation/Employers Liability Aviation Marine Terrorism “Pick the best tool for the job”
11
The data challenge The data will be provided as five tables each a separate realisation each model slightly more complex labelled A, B, C, D and E. Assume aggregate claims’ figures are accurate and final not subject to any estimation error. “Pick the best tool for the job”
12
The data challenge E.g model A “Pick the best tool for the job”
13
The data challenge what is 99.5% Aggregate VaR? (across all lines of business) what is 99.5% Aggregate TVaR (E(X|X>VaR(X)) what is Mean for each line of business individually? what is the Price for each line of business individually? “Pick the best tool for the job”
14
The data challenge Methods Purpose = illustrate different approaches Any techniques may therefore be applied As long as participants are prepared to describe these and present them at the autumn seminars. “Pick the best tool for the job”
15
The data challenge Submissions Answers should be presented using the template to be supplied Accompanied by a written description of the approach submissions used to select who will be invited to present identify and understand broadly the methods used without being overly long or going into great technical depth! 1,000 words (ish) Longer descriptions also welcome as attachments. The completed answer templates and written submissions should be sent to Nicky Wilkinson at nicky.wilkinson@actuaries.org.uknicky.wilkinson@actuaries.org.uk by 28 th September 2007. include full identification and contact details or upload yourself to the wiki and send a link to Nicky… “Pick the best tool for the job”
16
The data challenge Prizes (just for fun…) Answers ranked on absolute difference from the organisers’ calculated estimates (OCE). this will give 15 rank figures for each submission the participant with the lowest sum of these ranks will be judged the winner. Missing answers will be given the lowest rank Equally good answers across participants will be given the same rank. The prize for the best submission will be something. The winner and two runners-up will be announced at the autumn event and their names will be published in The Actuary magazine. “Pick the best tool for the job”
17
Pick the best tool for the job Search out and use Prior Art If you find a useful tool, post it and tell the World Harness collective intelligence We need a new relationship with IT The Toolkit Manifesto
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.