I think most sports games will, ultimately, move to the live service model. And honestly, for some of the more niche sports, I think it will be a good thing.
There's no reason to put your dev team on an annual treadmill to push out a boxed product. All it does is limit the amount of time they have to make significant changes or innovations with a limited budget.
Launch each generation with an NHL game, establish it as the platform moving forward, then release the roster updates, gameplay tweaks, engine enhancements over time as updates.
As far as NHL games in particular, I'm not sure the stagnation is due to a lack of competition so much as a lack of direction. There's a pretty big divide between the reality of hockey (what you see on TV) and the fantasy of hockey (which is what the games are built around). 'Hero moments' in hockey are relatively rare. There's a lot of randomness in the game. Making a game that looked and played exactly like the sport on tv would involve a lot more fumbled passes, turnovers, neutral zone play, missed shots, blocked shots, deflected shots, etc. The whole 'I'm the hero screaming down the wing and snapshotting a gorgeous goal isn't really as common in hockey as we know it as it is in videogames... but that power fantasy is what players want from videogames. They want the ability to impact the outcome with their own skill, and to limit random chance. But that's not hockey.
The NHL series right now is basically what Need for Speed was for racing games. Trying to straddle the line between real world and fantasy. Ultimately people tired of need for speed, because if they wanted the fantasy they could go for Burnout and if they wanted simulation they could go for Gran Turismo or Forza.