Pietro Rossi had a problem. An insurance company needed a model that could price bonds based on the likelihood of changes in credit ratings. The standard, off-the-shelf models are based on probability of default – not, specifically, credit rating moves.
So, he – together with a team from Bologna-based consultancy Prometeia – built his own.
The resulting model, which is detailed in a January paperis in production at the client’s business, says Rossi, adjunct professor of computational finance at the University of Bologna.
“We realised that it was possible to build a framework where you create a stochastic scenario for transition matrices and … reproduce prices of observable bonds according to their actual ratings,” he says.
Credit transition matrices indicate the probability of a rating moving up, moving down or staying the same. Credit ratings are plotted along the horizontal and vertical axes. The largest probabilities are normally positioned along the diagonal of the matrix, suggesting that the most likely scenario is that in the next period a bond will maintain its current rating. The elements outside the diagonal indicate the probabilities of a change in rating.
In the approach Rossi and co-authors developed, the pricing is based on scenarios simulated using stochastic multi-period credit rating transition matrices. That allows bond prices to be estimated not only at maturity but also at intermediate monthly steps throughout the life of the portfolio.
The main applications the team envisages are computing the distribution of the present value of a credit portfolio’s P&L and simulating the rating composition of a credit portfolio.
But credit risk is not all that Rossi has turned his talents to recently. He also published a paper earlier this month on volatility models and their calibration to S&P and Vix options.
The work deploys a faster calibration technique than previous models. “It’s a technical contribution on how to calibrate the path-dependent volatility model proposed by Julien Guyon and Jordan Lekeufack,” explains Rossi.
Rossi’s solution – developed with co-authors Fabio Baschetti of the University of Verona and Giacomo Bormetti of the University of Pavia – proposes to bypass the computational bottleneck from Monte Carlo simulations. “We can teach a network to learn both [SPX volatility and Vix volatility]so the calibration procedure is lightning fast,” Rossi says.
The calibration of a volatility model so that it fits both the S&P volatility surface and the volatility curve implied by the Vix index has been attracting the attention of quants for about a decade, including Julien Guyon, Mathieu Rosenbaum and Jim Gatheral. Rossi admits that the intellectual challenge is a major draw for the quant community – possibly more so than the practical aspects of the problem.
Looking ahead, Rossi says he is pursuing two main research directions. One is a closer examination of widely used interest rate models, testing the correctness of the SABR model introduced by Hagan and others.
The other area of research is connected to the work on credit rating transition. Rossi and academic colleagues are investigating whether the stochastic transition framework can be used to price Bermudan or American options on defaultable bonds using least‑squares Monte Carlo. Either could become a Risk paper soon!
To listen to this podcast…
Index:
00:00 Introduction
01:41 Background on credit and rating transition
04:00 Uses of the stochastic approach
07:45 Applications and calibration
15:10 Joint calibration of S&P and Vix options
17:50 Speed-up from previous methods
23:10 Why quants want to solve the joint calibration problem
29:35 Future projects
To hear the full interview, listen in the player above, or download. Future podcasts in our Quantcast series will be uploaded to Risk.net. You can also visit the main page here to access all tracks, or go to Spotify, Amazon Music or the iTunes store to listen and subscribe.