Peng Zhou

• Team leader of model analysis group. The main responsibility includes ongoing performance analysis (OPA), BAU backtesting and model revalidation for counterparty risk models (e.g. simulation model, pricing model, margin and aggregation model). The model scope includes FX, Rates, Credit, Equity and Commodity.
• Develop and enhance multi-curve interest rate simulation model (MCIR) for interest rates simulation. This model is used in conjunction with pricing model to determine exposure profiles for the internal risk evaluation and regulatory capital calculation.
• Rates and equity front office pricing model integration. Conduct comprehensive analysis on the front office pricing models (e.g., backtesting, sensitivity and stress test, convergence test) to incorporate the rates and equity front office pricing models in counterparty risk framework. Gain knowledge in the front office pricing models such as rates curve construction, vanilla interest rates swap, cap, floor, swaption, cross currency swap, european/american option, equity swap, etc.

Near surface oil seepage detection using Microbial Markers.

• Build a binary classification model using light gradient boosting machine to predict near surface oil seepage. Only 252 data samples are available for analysis and the number of features (DNA sequences) are more than 5000.
• Develop a method to solve the covariate shift issue where there is a distribution mismatch between training and testing data.
• Develop a method for feature selection to have the feature dimension reduced from 5000 to 16.
• Prediction AUC sore is improved from 0.66 to 0.85