New to the site. I am currently working on a project that involves analyzing two pricing sources (IDC and Markit) on fixed income assets (Corp, High Yield, Muni, and Structured). I am trying to develop a method that is effective in identifying outliers by price in comparing these sources. IE, Markit prices bond A at 105, IDC prices bond A at 95.
My current method on corp and HY I use Z-Score, IQR, and Median Avg Deviation. I group these based on Liquidity score provided by Markit (their confidence score), and then further segment by maturity/term. Any asset that meets 2 of these 3 measures as on outlier is moved on to phase 2. Phase 2 is a materiality test, where I examine the total position held across all of our pets. The cut off is 1mm. The final phase is a test of % variance of position (diff/100)*(pos).
I then look at the previous 3-4 months of data to see if any of the bonds are consistently out of tolerance. My question to you guys, is there any way to make this more robust? Any other factors I could throw in?