Hockey Outsider
Registered User
- Jan 16, 2005
- 9,394
- 15,463
In another recent thread, I made an offhand comment about "three stars" selections from playoff games. Someone (wisely) asked @pnep if he had any data about this and, not surprisingly, he did. I've taken a quick look at the file he posted. I figured this was interesting enough to warrant a separate thread. For now, I'm going to focus on the playoffs (though I might look at the regular season data later on).
A quick note on data quality - pnep's file has results from 1992 to 2022. There have been 2,635 playoff games during that period. We'd therefore expect 2,635 * 3 = 7,905 "three stars" to have been selected. The file actually has 7,640 entries. Therefore the data is approximately 97% complete. I haven't done a deeper dive to determine if the omissions are random, or if certain years are missing (which could disadvantage certain players).
I've come up with a basic weighting system. Being named first star of the game is worth three points, second star is worth two points, and third star is worth one point. There are a number of different weights that can be used, but this is to get the conversation going.
A quick note on data quality - pnep's file has results from 1992 to 2022. There have been 2,635 playoff games during that period. We'd therefore expect 2,635 * 3 = 7,905 "three stars" to have been selected. The file actually has 7,640 entries. Therefore the data is approximately 97% complete. I haven't done a deeper dive to determine if the omissions are random, or if certain years are missing (which could disadvantage certain players).
I've come up with a basic weighting system. Being named first star of the game is worth three points, second star is worth two points, and third star is worth one point. There are a number of different weights that can be used, but this is to get the conversation going.