I mean is it? His war profile seemingly shows that he is mostly just miscast as the shutdown guy 5v5. Give him softer minutes with a 3 C that can take some of the Defensive load and he could easily rocket up the war chart.
Also that his line mates (pearson) aren't suited to being on his wings if thats the roe a team wants horvat to do.
This post perfectly makes my point.
What you're saying is that what 'WAR' is actually measuring is effectiveness in a player's role, and is useless without context. And I can show you posts I made 15 years ago explaining that that's literally exactly what +/- does. And, again, all most of these advanced stats are doing is taking +/- of events with bigger sample sizes. And if you have no context ... it's junk, just like +/-.
But further to that, these 'WAR' stats and the people who use them aren't remotely trying to present them as 'effectiveness in role, requires context'. They're presenting them as 'how good this player is'. The 'R' in WAR is referring to replacement-level players, and if you're saying Bo Horvat is a negative WAR player, what you're actually doing is presenting that Bo Horvat is a sub-replacement level player who is a fringy AHL type. Which is obviously nonsense.
Like, no shit a #4 defender playing #1 minutes on a terrible team is going to have bad results. And of course that's going to get captured. But it doesn't mean the player is sub-replacement level or a garbage player, and making that assumption based on a graph without context is just ridiculous.
This isn't baseball and the attempts of people without the required math skills and logical understanding to try turning hockey stats into baseball stats is beyond tiresome.