An analysis of different rating systems

Let me state at the outset that I'm not as of yet jumping on the sportradar hate bandwagon. However, I agree that the ratings seem very random. So what I did was take the example of a particular team and compare player ratings across sportradar, sofascore (rating system last year) and whoscored (ratings system two years back). I chose Watford for my comparison (ratings from Watford v Liverpool matchday 1). I ranked players based on their ratings in decreasing order (highest rated player comes first)

Sportradar

Britos
Okaka
Pereyra
Gray
Chalobah
Doucoure
Holebas
Janmaat
Richarlison
Cleverly
Amrabat
Gomes
Kaboul

Sofascore

Doucoure
Okaka
Holebas
Britos
Kaboul
Pereyra
Janmaat
Chalobah
Richarlison
Gray
Amrabat
Cleverly
Gomes

Whoscored

Doucoure
Holebas
Okaka
Britos
Kaboul
Amrabat
Janmaat
Pereyra
Chalobah
Gray
Richarlison
Cleverly
Gomes

Observations -
1) Some players are consistent amongst all three apps. Gomes and Cleverly for example. Both near the bottom. Okaka is always near the top. Janmaat is consistently middle of the pack. So that's a good sign.
2) There are a few glaring discrepancies. Look at kaboul. Lowest rating in sportradar. Top 5 in both sofa and whoscored. Amrabat is low in sport but high in whoscored. Doucoure is highest in sofa and whoscored but average in sport. This is strange.

How is it possible that a player who got the highest rating in one app gets a bad rating in another? Are the criteria for scoring so different? In fact, what is sportradar's criteria? With sofa and whoscored, it was transparent. We knew exactly why they arrived at the ratings they did. Here it is shrouded in secrecy. It is understandable if there are minor differences - Even sofa and whoscored were not the same. Sofa was a softer grader system while whoscored was a high risk high reward scoring system. And yet, if you compare those two, the ratings are fairly consistent. They have the same top 5, the same bottom 5. Sportradar on the other hand...

Requests to the comunio admin - kindly work on the transparency aspect ASAP. We need to know how these ratings are being decided. Because clearly we have a problem. Yes, it is important for us to have a website to check these ratings but it is absolutely crucial that we know on what basis the ratings are given. People who play 10 minutes are being given higher ratings than ones who have played the whole game. Clearly that's a normalisation issue. And like so many other have said, people are going to lose interest quickly with a bad rating system. So it is in your best interests to clarify the situation immediately.

Thanks

TL;DR - we need more transparency and an explanation of the new system. It is extremely counter intuitive and has no resemblance with previous rating systems. You are going to lose your fan base real fast if there is no clarity.

RoadLFC wrote:

Let me state at the outset that I'm not as of yet jumping on the sportradar hate bandwagon. However, I agree that the ratings seem very random. So what I did was take the example of a particular team and compare player ratings across sportradar, sofascore (rating system last year) and whoscored (ratings system two years back). I chose Watford for my comparison (ratings from Watford v Liverpool matchday 1). I ranked players based on their ratings in decreasing order (highest rated player comes first) Sportradar Britos Okaka Pereyra Gray Chalobah Doucoure Holebas Janmaat Richarlison Cleverly Amrabat Gomes Kaboul Sofascore Doucoure Okaka Holebas Britos Kaboul Pereyra Janmaat Chalobah Richarlison Gray Amrabat Cleverly Gomes Whoscored Doucoure Holebas Okaka Britos Kaboul Amrabat Janmaat Pereyra Chalobah Gray Richarlison Cleverly Gomes Observations - 1) Some players are consistent amongst all three apps. Gomes and Cleverly for example. Both near the bottom. Okaka is always near the top. Janmaat is consistently middle of the pack. So that's a good sign. 2) There are a few glaring discrepancies. Look at kaboul. Lowest rating in sportradar. Top 5 in both sofa and whoscored. Amrabat is low in sport but high in whoscored. Doucoure is highest in sofa and whoscored but average in sport. This is strange. How is it possible that a player who got the highest rating in one app gets a bad rating in another? Are the criteria for scoring so different? In fact, what is sportradar's criteria? With sofa and whoscored, it was transparent. We knew exactly why they arrived at the ratings they did. Here it is shrouded in secrecy. It is understandable if there are minor differences - Even sofa and whoscored were not the same. Sofa was a softer grader system while whoscored was a high risk high reward scoring system. And yet, if you compare those two, the ratings are fairly consistent. They have the same top 5, the same bottom 5. Sportradar on the other hand... Requests to the comunio admin - kindly work on the transparency aspect ASAP. We need to know how these ratings are being decided. Because clearly we have a problem. Yes, it is important for us to have a website to check these ratings but it is absolutely crucial that we know on what basis the ratings are given. People who play 10 minutes are being given higher ratings than ones who have played the whole game. Clearly that's a normalisation issue. And like so many other have said, people are going to lose interest quickly with a bad rating system. So it is in your best interests to clarify the situation immediately. Thanks TL;DR - we need more transparency and an explanation of the new system. It is extremely counter intuitive and has no resemblance with previous rating systems. You are going to lose your fan base real fast if there is no clarity.

Similar is case with J. King's rating in the match day 3, he got 7.9 and 7.5 in whoscored and sofa respectively and was by far the best player on pitch for Bournemouth but got a lowly 4 in sportradar.

There needs to be some basis on which the ratings are given, usually its based on the player's performance and his influence in the game. But it looks like sportradar just uses a random number generator to give ratings to losing team