We address the needs of investment banks to respond to current and future regulations, including a demonstrated ability to survive a two-week downturn of specific given severity. The Basel III accords, not yet finalized, are projected to substitute expected shortfall for current Value at Risk.
Under current Basel II and projected Basel III requirements, the daily requirement is determined on a rolling basis, with two-week survival probabilities calculated according to a proscriptive formula.
We improve on the expected shortfall analysis relative to current standards of risk management. As examples, we estimate the risk with the same analysis modules as those used in optimization to mitigate or hedge the risk. Moreover, we can estimate risks of greater severity that Basil III will require.
Beyond the severity for an expected shortfall risk at 1%, at traded frequency from 30 minutes to 2 weeks, we estimate the expected shortfall, as well as VaR, at all quantile levels from 0.1% to 99.9%. This improvement in accuracy increases the severity window from worst case scenario to worst case scenario.
These risk levels are designed to avoid excessive losses, to ensure capital reserves against margin call requirements, and to allow aggressive investment when the risk is justified. Our multi time horizon risk assessment goes beyond that offered by other risk models. We offer extreme accuracy, at levels of 1% tail risk expected shortfall and even 0.1% tail risk. The path dependent risk models of maximum drawdown and margin call risk which we offer are not common for most risk model packages.
A unique capability of the GlimmAnalytics modules is the interoperability of its risk and optimization modules, conducted within arbitrage pricing theory. These modules allow a common software base for risk assessment, risk budgeting and optimization.
Our performance evaluation module assesses trader performance. This assessment is used to allocate capital, to determine compensation, termination or promotion, and to prove success to outsiders for the purpose of attracting assets under management.
Each day, or investment period, performance is measured relative to the GlimmAnalytics assessment of returns as a probability distribution function. These returns are accurate at all quantiles (Kolmogorov Smirnoff stat). Statistical returns above or below expectation can be integrated over time to give a cumulative, accurate, picture of performance.
Many of the issues which are specific to commodity returns are highly technical and commodity specific, such as weather patterns (crops), geopolitics (petroleum), regulations (petroleum), evolving demand patterns (lithium, nickel), and long term inflation fears (gold). We do not attempt to enter into such details, but take advantage of the fact that they are largely reflected in prices. On this basis, we model commodities from a strictly price history point of view. Development of this plan and proof of concept is under advanced development, based on the GlimmAnalytics scenario engine.
We address duration risk through examination of federal bonds of various duration. Through bootstrapping, equivalent couponless bond histories are constructed. These are analyzed by the GlimmAnalytics scenario engine, which features heavy tails and other typical features of financial returns. Initial results are promising. When they are fully documented, we will offer the package as a GlimmAnalytics product. Risk assessment of defaultable loans is also under advanced development.