GSoW Squad Ranking - Discussion Topic

All Squad Wars related topics. Questions, comments, issues, shout outs, call outs.
DarkRebel
Grand Admiral
Posts: 1282

Re: GSoW Squad Ranking - Discussion Topic

Post#11 » Wed Apr 17, 2019 6:57 pm

Well, even the prediction logic has no trust in the ranking, and that’s laughable. LOL.

Can anyone in the Galaxy explain why the underdog is predicted to win, base on the same ranking system that it used ? It’s like hypocrisy is built right into the system.

There is just no logic in it, whatsoever.

Yes, I did check out the page which discussed Trueskill system with Mu and Sigma. It is way too complicated to fully comprehend.

The way I understand it, Mu is the true ranking, and sigma is just the degree of variation (standard deviation) from true ranking. High sigma doesn’t mean that the squad is bad, just that the ranking is more unstable.

The prediction logic is right in using Mu only.
The ranking uses something like Mu - 3 x sigma, so that’s where the flaw is, I think. Standard deviation means +/- from Mu, but ranking uses a pessimistic approach of only subtracting sigma from Mu.

Ranking should just the same logic as the match prediction and only uses Mu.

A squad with high sigma just means that its ranking is unstable and can change a lot, and doesn’t mean that it is not as good as another squad with lower sigma value.


User avatar
Spock
Grand Admiral
Posts: 1028

Re: GSoW Squad Ranking - Discussion Topic

Post#12 » Wed Apr 17, 2019 7:08 pm

DarkRebel wrote:Well, even the prediction logic has no trust in the ranking, and that’s laughable. LOL.

Can anyone in the Galaxy explain why the underdog is predicted to win, base on the same ranking system that it used ? It’s like hypocrisy is built right into the system.

There is just no logic in it, whatsoever.

Yes, I did check out the page which discussed Trueskill system with Mu and Sigma. It is way too complicated to fully comprehend.

The way I understand it, Mu is the true ranking, and sigma is just the degree of variation (standard deviation) from true ranking. High sigma doesn’t mean that the squad is bad, just that the ranking is more unstable.

The prediction logic is right in using Mu only.
The ranking uses something like Mu - 3 x sigma, so that’s where the flaw is, I think. Standard deviation means +/- from Mu, but ranking uses a pessimistic approach of only subtracting sigma from Mu.

Ranking should just the same logic as the match prediction and only uses Mu.

A squad with high sigma just means that its ranking is unstable and can change a lot, and doesn’t mean that it is not as good as another squad with lower sigma value.


I don’t disagree but there is only so much we can do and we will continue to do tweaks and tests to try and improve the site within the boundaries that are actually within our control. For a while now we have been discussing and testing a potential change. If it’s made the main thread will be updated.
Image

Leader of the Imperial_Enclave.

Check out our YouTube channel with OVER 1 THOUSAND no outpost replays being shared. Quite possibly the largest collection of commander replays out there.

https://m.youtube.com/channel/UCW4iVRbrSDJHTBZobNNy7LA


DarkRebel
Grand Admiral
Posts: 1282

Re: GSoW Squad Ranking - Discussion Topic

Post#13 » Wed Apr 17, 2019 8:09 pm

Spock wrote:I don’t disagree but there is only so much we can do and we will continue to do tweaks and tests to try and improve the site within the boundaries that are actually within our control. For a while now we have been discussing and testing a potential change. If it’s made the main thread will be updated.

Appreciate the response Spock.


Return to “Squad Wars Discussions”

Who is online

Users browsing this forum: No registered users and 6 guests