MHR is a godsend for managers trying to figure out what tourneys to enter or what independent games to schedule. That was the initial purpose and it is very good at it. Like any algorithm, it needs data to become more accurate so it is much better in the mid to late season compared to the start. It is also the best tool we have for determining the best teams in the country for things like Nationals.
I have some knowledge of 18U hockey right now, can't speak much for the younger ages, and I've watched games of many top 40 18U AAA teams this year. The ranking of the top 18U teams by MHR is way more accurate than the Krach rankings of some of these teams in THF. Very easy to game the Krach system by playing an easier schedule and racking up wins. People say you can game the MHR system by playing a tougher schedule, but I haven't found that is the case. MHR does run into issues when there are lots of games where teams are beating each other by more than the 7-goal maximum. The whole point of MHR in the first place was to try to provide data to prevent those match-ups from happening. If tourney and league organizers spent more time focused on that goal, there would be less issues all around.
Not always possible but here's what every coach/manager should be striving for: try to build a schedule where you are predominantly playing teams rated within 3 goals of you, and then you just need to worry about winning the game (not covering the spread). I'm talking about higher level, older-age youth hockey here, where coaching to win should be occurring.