Market risk assessment constitutes a significant part of the estimation of financial
risk. It is calculated using risk measures, expressed in risk metrics, and has traditionally
been estimated using historical data. This has the disadvantage that it provides a retro-
spective indication of risk that may not be a proper indication of current and future risk
under unstable market conditions. Currently, the popular measures used for estimating
market risk are Value at Risk (VaR) and Expected Shortfall (ES) metrics. It is vital for
companies to know about risks at the moment that decisions are made, and VaR or ES
enable this by incorporating future prospects into their risk calculation by using prob-
ability distributions. However, classical VaR calculation assumes that only the risk of
single assets and their correlation (or dependence) matters. It does not take sudden
market changes into account. As a consequence, this makes VaR inflexible and unre-
sponsive with regard to abnormal market conditions, such as with the instability caused
by high-impact news events or an economic crisis.
17.3 Refining VaR and ES calculation using semantic news analysis
Abnormal market conditions exert a much higher risk than normal market conditions
and it is therefore vital to include them in risk management strategies. The inflexibility of
market risk measures with regard to such abnormal conditions can be countered by
developing a system that takes financial news messages into account. Financial news
reports all events that are relevant for the value of an equity. Such events, or chains of
such events, might cause unstable market conditions. Detection of these events can be
used to estimate the probability of emerging abnormal market conditions. By incorpor-
ating news into the risk calculation, sudden impactful events can help to determine
the kind of distribution that should be attributed to individual parameters of the
calculation. In order to do so, news messages need to be given an impact value.
17.4 The implementation of semantic news analysis
In order to use news as a source for certain market risk evaluation, it needs to be
determined what the impact of a news item is on the equities of portfolios in that specific
market. Recent technological developments, like SemLab’s ViewerPro semantic analysis
platform, have enabled the creation of data-mining tools that can interpret live news-
feeds. Combining such technologies with risk metrics, such as VaR, could lead to
quicker, more flexible, and more accurate risk assessment calculation.
Historical data can be analysed to determine the magnitude of specific events or event
combinations. This would yield an estimate of the probability that abnormal market
conditions occur. It would also result in an estimate of the effect of unstable market
behaviour on a certain equity. As such, news event analysis could be used as input for
risk measures to calculate risk metrics during abnormal market conditions and answer
two important questions: ‘‘Will there be abnormal market conditions?’’ and ‘‘What will
be the effect on a certain portfolio?’’
The efficient market hypothesis states that prices of traded assets (stocks, bonds, etc.)
reflect all known information. Typically, risk measure calculations only take into
320 Industry insights, technology, products, and service providers