Tweet
How Power Ratings Are Calculated
Hoping to head off questions about this later on.
1. Six different stats categories make up the "performance" component of the power rating. They are:
2. Also included in the power rating is a roster rating. It is based on player ratings, positional importance (i.e. QB & WR are wayyyyyyy more important than the FB), and position-based depth. (i.e. RB2, WR3, and CB3 are the most important relative to the starters at those positions.)
3. Preseason, the power rating is 100% roster rating. As the season goes on, it becomes more and more performance-based, but always with at least *some* roster component. I don't remember the exact numbers off the top of my head, but it's something like...
Preseason: 100% roster, 0% performance
After week 1: 85% roster, 15% performance
after week 2: 75%/25%
after week 3: 60%/40%
after week 4: 45%/55%
after week 5: 30%/70%
after week 6: 15%/85%
after week 7: 10%/90%
rest of season: 10% roster, 90% performance
As I said, I don't recall the exact numbers, but I think 10% roster/90% performance is the ending point (it's either that or 15/85) and it's somewhere around week 7 or 8 that we reach that point.
4. In the first 2-3 weeks of the season, expect to see heavy movement at times, and every now and then some crazy point spreads. I suspect that the reason that Jim doesn't adjust his power ratings until week 2 is over is to avoid this. The specific situation where point spreads can go crazy is if a team has a monster performance in week 1, especially a game where they allow 3 points or less. Given that the ppg average is usually around 21ish, the code sees this as a scoring defense that is seven times better than the league average. This works itself out just fine over the course of the season. (Once a few weeks have past, it's not terribly common for a team to have any number that is even 1.5 times the average. A big-time high-scoring team might put up 32 or 33 ppg in a league with an average of 21.)
So, bottom line, if in week 2 a team that won 35-3 in week 1 plays a team that lost 41-7, you might see a 20-25 point spread. I've found that this isn't a common enough occurrence to make it worth the considerable effort to work around.
Hoping to head off questions about this later on.
1. Six different stats categories make up the "performance" component of the power rating. They are:
- total offense
- total defense
- scoring offense
- scoring defense
- winning percentage
- opponents' winning percentage
2. Also included in the power rating is a roster rating. It is based on player ratings, positional importance (i.e. QB & WR are wayyyyyyy more important than the FB), and position-based depth. (i.e. RB2, WR3, and CB3 are the most important relative to the starters at those positions.)
3. Preseason, the power rating is 100% roster rating. As the season goes on, it becomes more and more performance-based, but always with at least *some* roster component. I don't remember the exact numbers off the top of my head, but it's something like...
Preseason: 100% roster, 0% performance
After week 1: 85% roster, 15% performance
after week 2: 75%/25%
after week 3: 60%/40%
after week 4: 45%/55%
after week 5: 30%/70%
after week 6: 15%/85%
after week 7: 10%/90%
rest of season: 10% roster, 90% performance
As I said, I don't recall the exact numbers, but I think 10% roster/90% performance is the ending point (it's either that or 15/85) and it's somewhere around week 7 or 8 that we reach that point.
4. In the first 2-3 weeks of the season, expect to see heavy movement at times, and every now and then some crazy point spreads. I suspect that the reason that Jim doesn't adjust his power ratings until week 2 is over is to avoid this.
The specific situation where point spreads can go crazy is if a team has a monster performance in week 1, especially a game where they allow 3 points or less. Given that the ppg average is usually around 21ish, the code sees this as a scoring defense that is seven times better than the league average. This works itself out just fine over the course of the season. (Once a few weeks have past, it's not terribly common for a team to have any number that is even 1.5 times the average. A big-time high-scoring team might put up 32 or 33 ppg in a league with an average of 21.)So, bottom line, if in week 2 a team that won 35-3 in week 1 plays a team that lost 41-7, you might see a 20-25 point spread. I've found that this isn't a common enough occurrence to make it worth the considerable effort to work around.
Comment