Well, the site is just like every other site that tries to measure skill. It has its ups and downs, and it's good to know which are they.
Particularly, lolskill focuses on a popular approach (KDA, gold per game, CS) which ends up being a ****py measurement tool for actual skill, and even when focusing on that, it does so in a really mediocre way.
Now about their algorythm (and why it's mediocre), they calculate your "lolskill" based on 3 "skills". "Experience", "General skill" and "Performance".
Experience is just a number that escalates with the approximate MMR of the league/tier/LP you were in the last time you played with that champion. Playing a game in platinum 5 gives you a base of around 2900 in Experience, whereas playing in a lower league gives less and less. This makes it really easy for master/challenger and high diamond players to get to the top 1000, since they already start with around 4000 points in anything they play in ranked.
General skill and Performance aren't as clear to me.
AFAIK, general skill measures how many games have you played with that champion, and may be affected by win rate and general KDA or some other variable, I only know that it goes up with games played with a champion.
Performance goes up with the percentage measurements over (or below) the average for that champion. It doesn't measure how long a game was, or what role you played, so supports and jungle champions will always have wonky stats with lots of red percentages in CS and kills, even when performing great, and early game enemy surrenders might actually decrease your champion's average performance. This is what makes their calculation really mediocre, and the reason why I stopped plainly trusting the number. It's still nice to see it go over 100% when you get really fed and carry a game though :P
- Not only gold is meaningless without the game lenght factor. KDA and CS also vary from shorter to longer games, and this is something really easy to take in account, which lolskill plainly ignores for some reason.
- Diamond/Master/Challenger players can have negative performance levels. There is no reason to think otherwise. Sure, if you're bronze and playing against a diamond player it might seem impossible for him/her to feed, but up there in their elo they're just one more player, and they face people like them that sometimes just stomp them. Sure, it's less probable that an awful feed happens, but having a negative score is not that uncommon (an early surrender when you're 1/3/0 will probably net a really awful negative performance, mostly due to how ****py the algorythm is).
Particularly, lolskill focuses on a popular approach (KDA, gold per game, CS) which ends up being a ****py measurement tool for actual skill, and even when focusing on that, it does so in a really mediocre way.
Now about their algorythm (and why it's mediocre), they calculate your "lolskill" based on 3 "skills". "Experience", "General skill" and "Performance".
Experience is just a number that escalates with the approximate MMR of the league/tier/LP you were in the last time you played with that champion. Playing a game in platinum 5 gives you a base of around 2900 in Experience, whereas playing in a lower league gives less and less. This makes it really easy for master/challenger and high diamond players to get to the top 1000, since they already start with around 4000 points in anything they play in ranked.
General skill and Performance aren't as clear to me.
AFAIK, general skill measures how many games have you played with that champion, and may be affected by win rate and general KDA or some other variable, I only know that it goes up with games played with a champion.
Performance goes up with the percentage measurements over (or below) the average for that champion. It doesn't measure how long a game was, or what role you played, so supports and jungle champions will always have wonky stats with lots of red percentages in CS and kills, even when performing great, and early game enemy surrenders might actually decrease your champion's average performance. This is what makes their calculation really mediocre, and the reason why I stopped plainly trusting the number. It's still nice to see it go over 100% when you get really fed and carry a game though :P
- Not only gold is meaningless without the game lenght factor. KDA and CS also vary from shorter to longer games, and this is something really easy to take in account, which lolskill plainly ignores for some reason.
- Diamond/Master/Challenger players can have negative performance levels. There is no reason to think otherwise. Sure, if you're bronze and playing against a diamond player it might seem impossible for him/her to feed, but up there in their elo they're just one more player, and they face people like them that sometimes just stomp them. Sure, it's less probable that an awful feed happens, but having a negative score is not that uncommon (an early surrender when you're 1/3/0 will probably net a really awful negative performance, mostly due to how ****py the algorythm is).
You need to log in before commenting.
LSS --- what level is gold, silver, etc on their scale?
Perf --- is this anything to be proud of? I did notice that the champs I consider myself good with I had a higher Perf, and champs I have played I don't consider myself good with I had a bad Perf with.
Also ---- do plat or diamond players have any -negative or 'red' perf levels?
Gold --- Is this stat meaningless since game length can greatly alter it and high level players tend to have shorter games?
Thanks in advance.