Hostname: page-component-7b7567c8c-2n9c9 Total loading time: 0 Render date: 2026-11-18T09:33:31.966Z Has data issue: false hasContentIssue false

Quantitative approaches to model uncertainty: update on recent research ‐ Abstract of the London Discussion

Published online by Cambridge University Press:  03 July 2015

Rights & Permissions [Opens in a new window]

Abstract

Information

Type
Sessional meetings: papers and abstracts of discussions
Copyright
© Institute and Faculty of Actuaries 2015 
Figure 0

Figure 1Equity return distributions (FTSE 100). Annual fit at two different rates

Figure 1

Figure 2Statistical definition of bias=A-T

Figure 2

Figure 3Method of moments risk calibration

Figure 3

Figure 4The Monte Carlo backtest (MCBT)

Figure 4

Figure 5Normal distribution without bias correction

Figure 5

Figure 6Adjusting input confidence levels to achieve desired exception rate

Figure 6

Figure 7Distribution fitting process

Figure 7

Figure 8From parameter to model uncertainty. Frequency of exceptions, sample size=30. MLE, maximum likelihood estimation; KS, Kolmogorov–Smirnov

Figure 8

Figure 9Example calculations of unbiased shortfalls. Figures based on Monte Carlo backtest, assuming a normal random walk

Figure 9

Figure 10Robust example (Nicholson & Smith, 2013)Source: Figure 10 is a reproduction of figure 3, from the Actuary magazine http://www.theactuary.com/features/2013/10/gi-prepare-for-the-worst/

Figure 10

Figure 11Expect rejection when testing big data

Figure 11

Figure 12What makes a good capital calculation