wilmott.com

Has anyone tried to calibrate a short rate model to historical rates? I am interested to know what the success is. I tried to do it for HW using standard econometrics - represented HW as an AR(1) process, and transforming the rates such that the mean = 0, looked for a sigma and alpha. However I failed to find historical data series that would consistently provide a good fit to the AR(1) model. I was wondering if anyone encountered similar problems for HW, or if another model was more successful (CIR, Ho-Lee etc).

HW model will by construction fit the historical data since it has a parameter that is time dependent, and you can adjust that parameter to make it fit.For fitting Vasicek, CIR type of model, you can go check out the papar by longstaff: http://www.anderson.ucla.edu/documents/ ... ison.pdfit has excellent discussion on how to fit the models, and how well they fit the market data.