18 players per team, 2 teams = 36 players per game.
Average goals per game = 5.18 (this year)
Number of periods in a game = 3 (overtime obviously changes it slightly, but not important for this)
So, picking a random player, to score in a random period in any game where they score is: 5.18/36/3 = 0.0479 = 4.79%
With 15 guesses that would mean you'd have a 52% chance of being right at least once, 91% chance of being right at least once in 50 guesses.
Ah, it's for a specific player? Great,
190 gp, 5 goals, 3 periods a game = (190*30/)5 = 0.877% chance of guessing a goal in a period for Bartkowski.
In 50 guesses, a 35.6% chance to be right at least once for him. In 82 guesses, a 51.4% chance to be correct at least once.
The only reason this doesn't happen more often is people don't post specific guesses for when a player will score often. I am sure I did some math wrong but this is nowhere near winning the lottery odds or even close which would be something like a 0.0001% chance of getting at least 1 in 82 guesses correct.