Me and my team has recently bought access to market data for sovereign bonds from Latin America countries. We are interested in finding a suitable outlier detection method to detect eventual errors in our data.

We have been googling multivariate outlier detection and found Robust Mahalanobis Distances as a possible tool. Has anyone in here had experience with bond data outlier detection using RMD?

Please discuss/make suggestions how we could/should tackle this problem

  • $\begingroup$ are you doing historical data cleanup or online outlier rejection? The former is supposed to be easier since causality doesn't matter. I bet RMD wouldn't be to too different from many other methods. The online algo would be harder and I'm not sure RMD would give you any edge either. $\endgroup$ – Will Gu Apr 19 '18 at 19:21

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.