Table 2.
Back-testing results (99% c.l.) of the conventional delta-normal models
| Total period | 20-day CDNVaR | 125-day CDNVaR | 250-day CDNVaR | 500-day CDNVaR |
|---|---|---|---|---|
| (a) Descriptive statistics of the CDNVaR versions | ||||
| Average VaR | − 2.43% | − 2.57% | − 2.62% | − 2.65% |
| SD | 1.65% | 1.38% | 1.21% | 1.01% |
| Min | − 14.23% | − 8.52% | − 6.69% | − 5.20% |
| Max | − 0.48% | − 0.93% | − 0.97% | − 1.35% |
| Violations | 134 | 120 | 129 | 130 |
| Average value of deviations when an overshooting is reported | − 0.69% | − 0.97% | − 0.99% | − 1.16% |
| Year | 20-day CDNVaR | 125-day CDNVaR | 250-day CDNVaR | 500-day CDNVaR |
|---|---|---|---|---|
| (b) Backtesting results per trading year | ||||
| 2000 | 4 | 6 | 5 | 4 |
| 2001 | 3 | 2 | 3 | 3 |
| 2002 | 4 | 5 | 5 | 5 |
| 2003 | 2 | 0 | 0 | 1 |
| 2004 | 6 | 0 | 0 | 0 |
| 2005 | 4 | 2 | 1 | 0 |
| 2006 | 4 | 5 | 4 | 4 |
| 2007 | 10 | 12 | 15 | 15 |
| 2008 | 8 | 12 | 21 | 28 |
| 2009 | 6 | 0 | 0 | 2 |
| 2010 | 8 | 7 | 6 | 0 |
| 2011 | 7 | 10 | 10 | 9 |
| 2012 | 5 | 3 | 1 | 0 |
| 2013 | 9 | 3 | 2 | 0 |
| 2014 | 11 | 10 | 10 | 7 |
| 2015 | 6 | 7 | 7 | 9 |
| 2016 | 2 | 2 | 4 | 5 |
| 2017 | 5 | 4 | 3 | 0 |
| 2018 | 10 | 16 | 15 | 20 |
| 2019 | 8 | 3 | 4 | 5 |
| 2020 | 12 | 11 | 13 | 13 |
| Sum of overshootings | 123 | 107 | 116 | 118 |
| More than 4 overshootings per year | 14 | 11 | 10 | 10 |
Part (b) reports the backtesting results. Italics indicate the years in which the “4 overshootings rule” is violated