What is risk analysis?
Risk analysis consists of the identification of risk, the assessment of its likelihood, and the measurement of its would-be impact. Stress-testing measures an entity’s exposure to risks under different types of scenarios.
Traditional stress-testing
Traditional stress-testing frameworks focus on quantification of exposure to real known and imagined risks. Typically this means some combination of:
- “What if” scenario analysis – exploring the impact of hypothesized shocks or events.
- Monte Carlo simulation – testing a range of values or assumptions for different parameters in a model, to explore impact under different conditions.
- HASH supports this sort of comparative-scenario and Monte Carlo analysis via its experiments wizard, but additionally supports advanced experiments which enable learning more, in less time, with a greater degree of confidence.
Advanced Approaches in HASH
Optimization
Optimization experiments are typically used to identify the least-costly or best-performing way of structuring a system (be it a business, or policy) to satisfy some end-goal, within a set of constraints. These same experiments can be used in an adversarial fashion to detect weaknesses in a system or conditions under which it may fail.
Black Swan Modeling
Both novel risk discovery and general forecasting in HASH utilize simulation-at-scale — the running thousands or more simulations in parallel — in order to discover probabilistically-likely outcomes as well as “black swans”, previously unseen but nevertheless possible occurences. Although these events by definition are not reflected in historical data (and therefore impossible to detect through regression-style analyses alone) they may be reasonably expected to occur, and even be likely when factoring in the existence creative competitors, customers, suppliers, or other market participants with diverging interests. This accurately reflects true dynamics observed within the economy and other real-world complex systems.
HASH can be used to predict the emergence of unexpected phenomena in both predefined scenarios and under conditions of uncertainty because it employs generative agent-based simulation. The resiliency of systems can be measured in a general sense, probabilistically across many runs, and specifically in the face of individual scenarios.
Calibration & Memoization
As events unfold in the real-world, an organization’s simulation playbook (past scenarios modeled) can be compared against. Experiments run are saved automatically in HASH and their outputs made available for future analysis.
Attribution & Sensitivity Analysis
HASH also contains built-in tools for conducting sensitivity analysis experiments which indicate the impact that individual actions, policies or parameter changes may have on the end-state of a simulation, helping measure the impact of individual decisions and risk portfolios.
Real-Time Alerting
HASH’s API allows for instant alerts to be sent in the event of unexpected simulation results. If the probability of a concerning or previously unseen event reaches a prespecified level, emails or webhooks can be triggered, alerting users directly or through third-party incident monitoring tools like PagerDuty.