Bank of America 2015 Annual Report Download - page 95

Bank of America 2015 93
The graph below presents the daily total market-based trading portfolio VaR for 2015, corresponding to the data in Table 56.
Additional VaR statistics produced within the Corporation’s
single VaR model are provided in Table 57 at the same level of
detail as in Table 56. Evaluating VaR with additional statistics
allows for an increased understanding of the risks in the portfolio
as the historical market data used in the VaR calculation does not
necessarily follow a predefined statistical distribution. Table 57
presents average trading VaR statistics for 99 percent and 95
percent confidence levels for 2015 and 2014.
Table 57 Average Market Risk VaR for Trading Activities – 99 percent and 95 percent VaR Statistics
2015 2014
(Dollars in millions) 99 percent 95 percent 99 percent 95 percent
Foreign exchange $ 10 $ 6 $ 16 $ 9
Interest rate 25 15 34 21
Credit 35 20 52 26
Equity 16 9 17 9
Commodity 5 3 8 4
Portfolio diversification (46) (31) (78) (43)
Total covered positions trading portfolio 45 22 49 26
Impact from less liquid exposures 8 3 7 3
Total market-based trading portfolio 53 25 56 29
Fair value option loans 26 15 31 15
Fair value option hedges 14 9 14 9
Fair value option portfolio diversification (26) (16) (24) (14)
Total fair value option portfolio 14 8 21 10
Portfolio diversification (6) (5)(12) (8)
Total market-based portfolio $ 61 $ 28 $ 65 $ 31
Backtesting
The accuracy of the VaR methodology is evaluated by backtesting,
which compares the daily VaR results, utilizing a one-day holding
period, against a comparable subset of trading revenue. A
backtesting excess occurs when a trading loss exceeds the VaR
for the corresponding day. These excesses are evaluated to
understand the positions and market moves that produced the
trading loss and to ensure that the VaR methodology accurately
represents those losses. As our primary VaR statistic used for
backtesting is based on a 99 percent confidence level and a one-
day holding period, we expect one trading loss in excess of VaR
every 100 days, or between two to three trading losses in excess
of VaR over the course of a year. The number of backtesting
excesses observed can differ from the statistically expected
number of excesses if the current level of market volatility is
materially different than the level of market volatility that existed
during the three years of historical data used in the VaR calculation.
We conduct daily backtesting on our portfolios, ranging from
the total market-based portfolio to individual trading areas.
Additionally, we conduct daily backtesting on the VaR results used
for regulatory capital calculations as well as the VaR results for
key legal entities, regions and risk factors. These results are
reported to senior market risk management. Senior management
regularly reviews and evaluates the results of these tests.
The trading revenue used for backtesting is defined by
regulatory agencies in order to most closely align with the VaR
component of the regulatory capital calculation. This revenue
differs from total trading-related revenue in that it excludes revenue
from trading activities that either do not generate market risk or
the market risk cannot be included in VaR. Some examples of the
0
25
50
75
100
125
150
12/31/2014 3/31/2015 6/30/2015 9/30/2015 12/31/2015
Dollars in millions
Daily Total Market-based Trading Portfolio VaR History