Credit Risk Projects Lo¨ıc BRIN and Fran¸cois CRENIN
´ Ecole Nationale des Ponts et Chauss´ ees D´ epartement Ing´ enieurie Math´ ematique et Informatique (IMI) – Master II
Lo¨ıc BRIN Credit Risk - Projects
1/5
Project 1
Project 2
Project 3
Project 4
Project 1 Using boosting algorithms to estimate the Probability of Default
Project 1: Using boosting algorithms to estimate the Probability of Default Description of the project: This project aims at comparing several tree-based boosting algorithms when trying to predict the probability of default. Target: Clear explanations of three boosting algorithms, their implementations and a structured comparison of their results after training them on the Credit database. More common algorithms such as RF or SVM could be used as benchmark. Database: Retail Banking Credit Data
Lo¨ıc BRIN Credit Risk - Projects
2/5
Project 1
Project 2
Project 3
Project 4
Project 2 Feature engineering on market data to estimate the Probability of Default
Project 2: Feature engineering on market data to estimate the Probability of Default Description of the project: This project consists in transforming raw market data into features used as inputs of Machine learning algorithms to predict a probability of default of a set of companies. Target: Use what you have learnt from the lectures and experiments on the database to built features. These features will only be used as inputs in the algorithms seen in the lectures (logistic regression, basic trees, bagging, RF and SVM). Database: Market Data
Lo¨ıc BRIN Credit Risk - Projects
3/5
Project 1
Project 2
Project 3
Project 4
Project 3 Using different copulas to price Collateralized Synthetic Obligations
Project 3: Using different copulas to price Collateralized Synthetic Obligations Description of the project: The purpose of this project to test different extensions of Vasicek models using different copulas to link defaults, and see how the pricing of the tranches’ spreads evolves and compare to the spreads quoted for the CSO based on the NA IG CDX 5Y S29 index on past data. Target: Pricers with different underlying copulas to be used. References: [Burtschell et al., 2009] Database: NA IG CDX 5Y S29 CSO tranches spreads in octobre 2017.
Lo¨ıc BRIN Credit Risk - Projects
4/5
Project 1
Project 2
Project 3
Project 4
Project 4 CVA Wrong Way Risk Multiplier Decomposition and Efficient CVA Curve
Project 4: CVA Wrong Way Risk Multiplier Decomposition and Efficient CVA Curve Description of the project: The purpose of this project is to price a 3-year Italian CDS that you want to buy to Intesa Sanpaolo taking into account Counterparty Value Adjustement (CVA) and Wrong Way Risk (WWR). Target: A pricer. References: [Pang et al., 2015] Database: CDS data from Intesa Sanpaolo, Italy and 3M EURIBOR rates, daily, from 2009 to october 2017.
Lo¨ıc BRIN Credit Risk - Projects
5/5
Project 1
Project 2
Project 3
Project 4
Burtschell, X., Gregory, J., and Laurent, J. (2009). A comparative analysis of CDO pricing models. Link. Pang, T., Chen, W., and Li, L. (2015). CVA Wrong Way Risk Multiplier Decomposition and Efficient CVA Curve. Link.
For a given threshold l0 < LGD, compute [(L â l0)+]. ... consists in the securitization of a portfolio of mortgage loans which have the same ... Loan to value: 75 %;.
This exercise is based on the following S&P transition matrix: .... order to make the results reproducible, we set the seed of the random number generator before.
Swap: the reference is a reference pool (usually between 5 and 10 reference entities) ..... The CRM is a risk metrics that, as the IRC, captures the risks due to.
the common threshold and Ri a normal variable equal to: Ri = ÏF + 1 â Ïεi, where F and ... consists in the securitization of a portfolio of mortgage loans which have the same ... depend on the initial amount provided (1 - Loan To Value), the rat
Write the value of the debt of the firm for the debt-holders, of the shares of the firm for the shareholders, as an option on the value of the firm and with maturity the ...
counterparty risk management: for internal purpose and for regulatory capital .... An airline usually protects itself against a rise in fuel prices by entering into long.
1 Credit risk models to fulfill regulatory requirements and prevent the bank from failure .... IRB Advanced: modeling of PD, LGD and EAD. .... Asset Management.
Lecture 2 â Statistical tools for scoring and default modeling. François CRENIN .... The logistic regression model can be defined the following ways: ... therefore lead to different estimations depending on the algorithm/software chosen. .... Ther
With a barrier option approach and stochastic interest rate [Brys et al., 1997];. â· Taking into account ..... Journal Of Finance. Link. Modigliani and Miller (1958).
that PD = Q(Ri < s) = Φ. ︸︷︷︸. Normal cdf ... A copula C, is a function that is used to model dependencies: â(x1, ..., xd ) .... Gumbel compulas, Student copulas, grouped t-copulas, individual t-copulas, etc.;. â· the so-called ..... Delt
Class structure and assignments Credit risk and economics Credit risk outcomes Credit risk: The basics ... Credit Risk. Lecture 1 â Introduction, reduced-form models and CDS .... Source: Aspects of Global Asset Allocation, IMF. and personal cross-c
Let us consider the loan portfolio of a bank made of 200 000 loans for an average amount of 100 000 EUR. ... default occurs when the value of equity is smaller that a threshold. .... Exercise 3: The subprime mortgage crisis: a model risk crisis?
year conditional on no earlier default is 3 %. 1. Estimate the ... that buys protection on A during the period starting in 3 years and ending in 7 years. We assume ...
2 Banks are financial intermediaries which fill the gap between supply and demand on money market and the risk ..... Banking services. Proprietary Trading.
Risk-Neutral Valuation: Pricing and Hedging of Finance Derivatives. N.H, Bingham and R. ... to credit risk, our goal was to introduce the basic concepts and related no- tation, rather than to ...... 3.5.3 Incomplete Accounting Data. 113 ..... high le
For details of our global editorial offices, for customer services and for ..... spelling mistakes and bad grammar to provide some invaluable suggestions. .... A lack of proper assessment of credit exposure and default probability was a key ...... In
Credit derivatives. OTC derivatives. Corporate. Retail. Basket default swaps. Contingent ... one can easily deal with a large number of names,. â« Tackle with ...
announcements, should lead to a drop in the company's bond prices. .... associated with the names in the credit portfolio, that can be either historical or market.
Jul 1, 2003 - This correlation parameter is not estimated in principle by the banks but computed ..... 16 It is easy to prove additivity in the single factor setting. ..... 20 As the for the quantile based risk measure, the expected shortfall based r