In the the third period or late in the game the team protecting a lead has already scored their goals and often sits back a bit playing a defensive game and generates few shots while the trailing team is desperate and tries to play offense and throw pucks at the net.
But by limiting games to only those that are within one goal in the first or second period you're already limiting an already small (>82 game) sample size.
Sample size is fine.
They also break it out to close/tied/leading/trailing/all so you can come up with your own conclusions.
It seems strange to me. I can only assume team's win games by having the puck more.
It seems strange to me. I can only assume team's win games by having the puck more.