You're quoting stats, but the conclusions you're drawing from them are based on misconceptions.
Ok, I'll explain what me and other users have been trying to explain with data.
View attachment 459635
A team averages 2.32 goals per game 5-on-5.
2.32gpg/60(mins in a game) = 3.75% (chance a team has of scoring in any given minute 5-on-5)
0.0375(scoring per minute prob.) x 0.9625(chance of opposing team not scoring) x 100 =
3.6% (chance of scoring in any given minute 5-on-5 when your opponent doesn't score)
i.e chance of tying the game in a minute without pulling the goaltender.
View attachment 459636
The most recent data shows an extra attacks results in a goal resulting in sending the game to OT, or winning
14.5% of the time.
Whether the other team scorers a goal on your empty net is irrelevant with this data, because with this data if they score on your EN you can't tie and that results in a failure of the objective, the same way neither team scoring would, or (100-14.5) 85.5% of the time. A failure in the alternative of not pulling the goalie gives you a 3.6% chance to score 5-on-5 or a 96.4% chance of failure.
Either way you calculate it:
96.4% - 85.5 = 10.9%
or
14.5 - 3.6 = 10.9%
Therefore, pulling the goalie gives a
10.9% higher success rate of achieving your objective. Any numerical observation of goals for vs. goals against while 6-on-5 with the goaltender pulled are completely irrelevant in the analyzation of whether using an extra attacker has a positive opportunity cost.