BEGIN:VCALENDAR
VERSION:2.0
PRODID:https://murmitoyen.com/events/vanille/udem/
X-WR-TIMEZONE:America/Montreal
BEGIN:VEVENT
UID:69db72df86c61
DTSTAMP:20260412T062431
DTSTART:20170929T143000
SEQUENCE:0
TRANSP:OPAQUE
DTEND:20170929T153000
URL:https://murmitoyen.com/events/vanille/udem/detail/784836-spectral-backt
 ests-of-forecast-distributions-with-application-to-risk-management
LOCATION:Université McGill – Burnside Hall\, 805\, rue Sherbrooke Ouest\
 , Montréal\, QC\, Canada\, H3A 0B9
SUMMARY:Spectral Backtests of Forecast Distributions with Application to Ri
 sk Management
DESCRIPTION:Conférence d'Alexander J. McNeil\, professeur d'actuariat à l
 'Université de York depuis septembre 2016. Formé à Imperial College Lon
 don et à l'Université de Cambridge\, il a été professeur adjoint au D
 épartement de mathématiques de ETH Zürich avant d'occuper la chaire Max
 well de mathématiques au Département d'actuariat et de statistique de l'
 Université Heriot-Watt\, à Édimbourg\, où il a fondé et dirigé la Sc
 ottish Financial Risk Academy de 2010 à 2016. 
\n Chaire Aisenstadt
\n 

 \nRésuméIn this talk we study a class of backtests for forecast distribu
 tions in which the test statistic is a spectral transformation that weight
 s exceedance events by a function of the modelled probability level. The c
 hoice of the kernel function makes explicit the user’s priorities for mo
 del performance. The class of spectral backtests includes tests of uncondi
 tional coverage and tests of conditional coverage. We show how the class e
 mbeds a wide variety of backtests in the existing literature\, and propose
  novel variants as well. We assess the size and power of the backtests in 
 realistic sample sizes\, and in particular demonstrate the tradeoff betwee
 n power and specificity in validating quantile forecasts.
END:VEVENT
BEGIN:VTIMEZONE
TZID:America/Montreal
X-LIC-LOCATION:America/Montreal
END:VTIMEZONE
END:VCALENDAR