Researches/Analyses โ Risk in the Banking Sector
I have conducted or collaborated on the following research and analyses:
Study of Overlapping Effects/Probabilities in Parameter Aggregation for the ECL Concept
Topics covered: Lifetime recovery probabilities, probabilities of selling real estate, sample representativeness in the context of model application methods, unnecessary model complexity.
Analysis of Possible Approaches to Measuring Uncertainty in Loss Models (Lifetime PD): Estimation Errors, Confidence Intervals.
Topics covered: Population vs. sample โ a highly non-obvious distinction in business reality, CTG (Central Limit Theorem), estimation errors for regression parameters, estimation errors for probability distributions, non-obvious interpretations of errors and their outcomes.
Analysis of the Unexpected Value Model Mechanics for Credit Losses
Topics covered: Business visualization of unexpected scenario situations, simplification strategies, model-based descriptions of unexpected situations, benchmarking against the IRB approach.
Approach to Describing the Probability Distribution of Extreme Values in the Context of Operational Risk
Topics covered: EVT (Extreme Value Theory) โ types I and II, global distribution vs. tail distribution, problematic definition of the tail threshold, conditional distribution and limit distribution, Mean Excess Function, stability issues in distribution estimation, GPD, application of truncated distributions, problematic applications of lognormal distribution, VaR, VaR scaling.
Analysis of Possible Approaches to Building Regression Models for Describing the Dependence of the Number or Value of Operational Losses on Independent Variable
Topics covered: Data normalization, selection of variable characteristics, selection of business variables, multicollinearity of variables, appropriate time definition for model observations, pitfalls related to model quality quantifiers (p-value, Rยฒ), misleading graphical presentations.
Comparative Model of Changes in Credit Risk (PD) Over Time in the Context of Lifetime Path Issues and Probability Conditioning (SICR Issue)
Topics covered: Conditional probability, different approaches to Weibull probability distribution, immeasurable probabilistic effects resulting from excessive complexity, probability trajectory plotting (Weibull) over time
Inadequacy of Credit Risk (PD) Assessment Due to Lack of Volume Class Homogeneity, Along with an Analysis of Risk Differentiation Levels Based on Exposure Volume
Topics covered: Portfolio risk approach, symmetry of volume distribution and moderate dispersion, dependence: exposure volume โ risk, choice of homogeneity type: business similarity of exposure vs. volume similarity, small-number problem and low-probability issues.
Approach to Analyzing Operational Loss Dependencies Considering the Nonlinear Nature of Correlation (Increased Dependencies in Distribution Tails
Topics covered: Linear vs. nonlinear correlation, rank correlation, Copula functions (Gaussian, t-Student), concept of full dependence/full independence/correlation dependence, correlation measurement methods in distribution segments (e.g., in tails), cross-correlations, appropriate time definition for observations
Loss Mechanics (PD) in a Lifetime Perspective and Its Description Using Statistical Methods
Topics covered: Survival analysis, Weibull distribution, correction of probability spaces (0-1), real saturation problem, business interpretation of parameters (ฮบ, lambda), estimation methods, optimization algorithms and their challenges, extrapolation of information to unknown spaces.
Development of Approaches for Backtesting Lifetime PD
Topics covered: Survival analysis, Weibull distribution, various time approaches.
Developed Methodologies โ Risk in the Banking Sector
I have developed or collaborated in the development of the following methodologies:
Methodology: Holistic Assessment of the AMA Operational Risk Model
More than 50 original analytical studies, highly rated by the national regulator.
Methodology: Holistic Assessment of Lifetime PD Models (IFRS9)
Original approach to backtesting, comprehensive approach to assessing business logic.
Methodology: Holistic Assessment of the ECL Model
Comprehensive approach to evaluating the ECL model.
Methodology: Holistic Assessment of Internal Capital Models
Methodology for economic capital models.
IT Projects โ Risk in the Banking Sector
RiskOPERON System: Development of the AMA Methodology and Module in an IT System
Comprehensive AMA methodology and computational functions.