I have cajoled my single-game predictive model Magnus into a playoff prediction model. Curious readers will find lots of detail in that link, but the short version is that I consider shots generated and allowed, including their locations, shooting and goaltending results, special teams, penalties drawn and taken, rest, and home ice advantage. Injuries, suspensions, and likely roster compositions are taken into account.
For each matchup, I simulated each game ten thousand times, given injuries and coaches' past allotments of icetime, and computed a probability for the home-ice team to win. Different teams benefit in different amounts from home-ice advantage because of their different strengths and weaknesses.
This bracket shows the probabilities for each team to win each round. Numbers smaller than 5% are not shown.
Furthermore, I have prepared estimates of the quality of the likely roster for each team, as of the beginning of the playoffs.