Thanks for the excellent response! I'm still not clear where you're getting your data, but it looks like it may be Natural Stat Trick. Is that correct?
The data I'm looking at, which is not the Bruins internal numbers (although they are also more kind to DF than NST), show Forbort with a > 50% XPG%, and in the top quartile in expected goals against per 20 minutes (out of 211 defensemen with 500 or more 5 on 5 minutes). Having said that, DF was slightly below league average in expected goals for, and last among the 7 Bruins defensemen with over 500 5 on 5 minutes. He was also next to last on the Bruins expected goals against (Orlov was worse). He had the lowest XPG% among the regulars, with the 2nd lowest OZ start percentage (Carlo was lower). Relative to team he was nothing special 5 on 5.
If I'm being honest, I was not a big fan of Forbort's signing. He's not a puck mover (tons of dump outs of his own end), has weak pass completion percentages on outlets (which he uses a ton), and has very low puck possession. He's strong on contested loose pucks, and was very good on the PK with the 8th lowest expected goals against per 20 minutes among the 123 NHL defensemen with over 100 minutes in shorthanded situations (4th on the Bruins, however).
But I guess what I find interesting here is how much the publicly available systems can differ from the more advanced. I'm old and losing interest, but still think it would be cool to do a comparison of the different hockey metric systems, looking at how similar their results are, and how their expected goal data compares to actual on ice goals over a longer (3 year?) period. Also, Zboril, per the data I'm looking at, was a significant drop down from Forbort, with a 45% XPG% and expected goals against 0.20/20 minutes higher than Forbort's.