- Joined
- Jan 8, 2016
- Messages
- 1,863
- Likes
- 4,497
TL;DR The 247 composite is just an average of all the other services ranking but has lots of issues.
Ive seen a lot of posts discussing the 247 composite and how they calculate their rankings. So I set out to try and figure out their methodology. I started by looking at their own explanation of how they do it. Heres the link: http://247sports.com/Article/247Rating-Explanation-81574. There are two numbers at play here that get used somewhat interchangeably so I wanted to clarify. Ranking is the relative position of a recruit compared to other recruits, i.e. #23 in the country. Rating is either a numerical score, i.e. 96 or 0.9456, or number of stars. The 247 composite strictly uses national rankings to calculate their score and doesn't care about the ratings.
The basic process is to take the average of all the sites rankings. Ill use Cade Mays as an example. This average ranking (19.75) is used to calculate the 247 composite rating (0.9925). Then the recruits are ordered highest to lowest and stars are assigned. I pulled everyones rankings from each site this week and built my own spreadsheet to try this out and was able to replicate (mostly) their rankings. The basic process seems reasonable but there are a lot of issues with 247s execution of this.
First, there are kids who are in the list multiple times (Jackson Carman at #12 and Jack Carman at #19 same school). Second, its clear that 247 doesnt have or isnt using the latest rankings from each of the sites. Third, some of the sites only rank a certain number of players. Rivals only ranks the top 250 and Scout only ranks the top 300 (although they only list 250). This leads to instances were a player doesnt have a ranking on one or more of the sites, especially those on the edge of those numbers.
This leads to number four, the composite ranking of kids who dont have a ranking on one of the sites are skewed. Its clear that something isnt right with their formula for these kids. Al Blades should be ranked around #37 based on his average (no ranking on Scout) but instead shows up at #79. There is a similar trend for others. Whats interesting is that even without a ranking on Scout, 247 indicates that they are using 4 rankings. I calculated what the missing ranking would be in order to get the composite score listed for him and found the missing ranking was the same for most of the kids who dont have a ranking on one site. 247 is using a ranking of 252 in lieu of the missing ranking rather than calculating the average of only the sites with a ranking. This is skewing a lot of kids down from where they should be.
What I havent figured out is how they determine number of stars. They say The 247Sports Composite Rating assigns stars based on an approximate average distribution of stars from the industry. However the composite has more 5 star players listed (26) than any individual service (Rivals 23, ESPN 12, 247 9, Scout 17). Maybe its based on rating or % of 5 star players versus the overall number of rankings?
My impressions are there are a lot of issues with 247s implementation. It does seem to give a reasonable answer for most kids in the top 200 or so players (with exceptions for kids like Al Blades who are missing a ranking). Beyond that, there are so many kids not ranked by one of the services that the composite rankings seem pretty random. It doesnt seem like the composite is updated as the services update their individual rankings but rather its updated periodically so the composite may be out of sync with the other services rankings.
I hope this helps explain the composite rankings and how theyre calculated.
Ive seen a lot of posts discussing the 247 composite and how they calculate their rankings. So I set out to try and figure out their methodology. I started by looking at their own explanation of how they do it. Heres the link: http://247sports.com/Article/247Rating-Explanation-81574. There are two numbers at play here that get used somewhat interchangeably so I wanted to clarify. Ranking is the relative position of a recruit compared to other recruits, i.e. #23 in the country. Rating is either a numerical score, i.e. 96 or 0.9456, or number of stars. The 247 composite strictly uses national rankings to calculate their score and doesn't care about the ratings.
The basic process is to take the average of all the sites rankings. Ill use Cade Mays as an example. This average ranking (19.75) is used to calculate the 247 composite rating (0.9925). Then the recruits are ordered highest to lowest and stars are assigned. I pulled everyones rankings from each site this week and built my own spreadsheet to try this out and was able to replicate (mostly) their rankings. The basic process seems reasonable but there are a lot of issues with 247s execution of this.
First, there are kids who are in the list multiple times (Jackson Carman at #12 and Jack Carman at #19 same school). Second, its clear that 247 doesnt have or isnt using the latest rankings from each of the sites. Third, some of the sites only rank a certain number of players. Rivals only ranks the top 250 and Scout only ranks the top 300 (although they only list 250). This leads to instances were a player doesnt have a ranking on one or more of the sites, especially those on the edge of those numbers.
This leads to number four, the composite ranking of kids who dont have a ranking on one of the sites are skewed. Its clear that something isnt right with their formula for these kids. Al Blades should be ranked around #37 based on his average (no ranking on Scout) but instead shows up at #79. There is a similar trend for others. Whats interesting is that even without a ranking on Scout, 247 indicates that they are using 4 rankings. I calculated what the missing ranking would be in order to get the composite score listed for him and found the missing ranking was the same for most of the kids who dont have a ranking on one site. 247 is using a ranking of 252 in lieu of the missing ranking rather than calculating the average of only the sites with a ranking. This is skewing a lot of kids down from where they should be.
What I havent figured out is how they determine number of stars. They say The 247Sports Composite Rating assigns stars based on an approximate average distribution of stars from the industry. However the composite has more 5 star players listed (26) than any individual service (Rivals 23, ESPN 12, 247 9, Scout 17). Maybe its based on rating or % of 5 star players versus the overall number of rankings?
My impressions are there are a lot of issues with 247s implementation. It does seem to give a reasonable answer for most kids in the top 200 or so players (with exceptions for kids like Al Blades who are missing a ranking). Beyond that, there are so many kids not ranked by one of the services that the composite rankings seem pretty random. It doesnt seem like the composite is updated as the services update their individual rankings but rather its updated periodically so the composite may be out of sync with the other services rankings.
I hope this helps explain the composite rankings and how theyre calculated.
