Matchups and Evolution....Need a Math Guy
Short answer: SoR could gain rating but it would require a severe blowout.
Better answer incoming once I check the numbers.
Piken Square
(edited by Murderous Clown.9723)
You are probably correct except for one important fact…
There is a built-in limiter on how much the RNG can modify a Server’s Rating before determining the new matchup and it is currently impossible for SoR and ET to be matched up.
The amount of RNG is still likely too large but I think the “maximum” comes out around 200 Rating which isn’t close to allowing the matchup you are talking about.
You are probably correct except for one important fact…
There is a built-in limiter on how much the RNG can modify a Server’s Rating before determining the new matchup and it is currently impossible for SoR and ET to be matched up.
The amount of RNG is still likely too large but I think the “maximum” comes out around 200 Rating which isn’t close to allowing the matchup you are talking about.
I understand the matchup could never happen. The point was to test the (capabilities and flaws) of glicko
Not on topic.. just.. I love your signature
My fun laughs at your server pride.
I’m back with the numbers as promised. Exactly how many points each server would be expected to hold depends on the third server. That said, we can look at the SoR vs ET head to head in isolation and calculate an expected ratio of points.
Feeding the server’s ranking, deviation and volatility into glicko 2 gives an expected match score of 0.998703101 (this number is always between 0 and 1) for SoR. Feeding this number into the sine transformation Anet uses and rearranging gives an expected ratio of 42.6.
So for the ratings to stay the same, SoR needs 42.6 points for every 1 ET gets. Depending on which server they are up against that translates to ET holding an average of 8-16 points.
Piken Square
I’m back with the numbers as promised. Exactly how many points each server would be expected to hold depends on the third server. That said, we can look at the SoR vs ET head to head in isolation and calculate an expected ratio of points.
Feeding the server’s ranking, deviation and volatility into glicko 2 gives an expected match score of 0.998703101 (this number is always between 0 and 1) for SoR. Feeding this number into the sine transformation Anet uses and rearranging gives an expected ratio of 42.6.
So for the ratings to stay the same, SoR needs 42.6 points for every 1 ET gets. Depending on which server they are up against that translates to ET holding an average of 8-16 points.
So based on todays score, where each match has a distribution of 480,000 points. That would mean in a head to head matchup, that ET would only need 11,162 points compared to SoRs 468,837 points, and the match would be considered even. Noone gains ranks and no one loses ranks ?
My math (likely flawed) 480,000/43 = 11,162 per point
11,162 × 42 =468,837
I got 40.1 rather than 42.6, based on these predicted ratings:
2 Sanctum of Rall 2189.9803 188.7653 0.7489 -30.4963 -1
24 Eredon Terrace 877.9270 183.1378 0.7553 -4.6137
because SoR is losing some rating this week to BG, next week they will be expected to win against ET by slightly less (assuming such a matchup could happen, which it can’t).
but it’s certainly the right ballpark. your math looks fine.
-ken
I was going to message Snowreap, but I figured he would find his way over on his own eventually :P
Menorah | Charr Cat | Some Cat Thing
Still running my old RRR build because why not
I got 40.1 rather than 42.6, based on these predicted ratings:
2 Sanctum of Rall 2189.9803 188.7653 0.7489 -30.4963 -1
24 Eredon Terrace 877.9270 183.1378 0.7553 -4.6137because SoR is losing some rating this week to BG, next week they will be expected to win against ET by slightly less (assuming such a matchup could happen, which it can’t).
but it’s certainly the right ballpark. your math looks fine.
-ken
I didn’t take this week’s scores into account, so that’s where the difference comes from.
Piken Square