Pretty sure he's had the most profitable betting model of any NHL model publicly published. Its not perfect, but I think he deserves credit for that. He's missed on the Blues' more than most teams, though.
There's a sports statistician (can't remember his name off the top of my head) whose model soundly beat all of the advanced stat hockey nerds a few years back based on the impact of officiating. He left some standard stuff out in favor of the predictability of penalties and referee biases - basically, reffing the score and home arena advantages. I don't think he publishes a yearly prediction, and it's possible some others have started to incorporate such things into their models, but we have good recent evidence that the advanced stat guys still have a lot to figure out.
Just because it's the most profitable, that doesn't mean it's accurate enough for any given fan's tastes when speaking about the game, especially when speaking about finer grain aspects. And I'm starting to see a little too much hockey discourse being squashed because of religious model adherence. So, even though I think that stuff is interesting, how fans choose to use it is often not great. Sometimes the evidence is overwhelming or at least solid, but it's okay to admit "the best we have" can still be not good enough to base conclusions on.