EPTBlizzard releases latest Win Percentages - Page 12
| Forum Index > SC2 General |
|
andrewlt
United States7702 Posts
| ||
|
DarkPlasmaBall
United States45191 Posts
"Well, that obviously proves that X is OP!" or "Well, that obviously proves that X is UP!" The numbers don't indicate balance issues -.-' You need to expect variation (not perfect 50/50). Stop QQing. Quite frankly, I'm pleased with how close the percentages are. It shows that Blizzard is doing a pretty damn good job so early on in the game's life. If a race was 30/70 all across the board in a match-up, then I'd be concerned. I think people who have trouble with playing against certain races or builds should focus on strategies and asking for help, rather than insisting on buffs and nerfs. If there's a problem, then the problem is YOU. | ||
|
ReketSomething
United States6012 Posts
| ||
|
Scruff
Singapore509 Posts
On November 07 2010 17:35 StarcraftMan wrote: The stats show that Zerg is OP in 3 out of 4 regions. Not surprising - this is something the community suspected aleady. And this post just shows how dumb and ignorant you are | ||
|
FunnelC4kes
Ireland462 Posts
| ||
|
Sprouter
United States1724 Posts
| ||
|
clickrush
Switzerland3257 Posts
On November 08 2010 13:48 andrewlt wrote: So it looks like Protoss is slightly underpowered but TvZ is around even (50.5%). That's not that bad considering T>Z (52-53%), Z>P (54%) and PvT (51%) roughly historically in professional BW. yeah but bw is so different. the only matchup that sometimes looks similar is TvZ imo. PvZ only in a few cases. sangho made it look pretty similar. PvT is nothing like bw. anyway, blizz does a great job with their balancing I really trust in those guys to do a good job also in the future. | ||
|
darmousseh
United States3437 Posts
What's an adjusted win percentage? Well, while the math behind calculating an adjusted win percentage is extremely complex, an adjusted win percentage can be summed up as the 'true' win percentage of a given race, produced by removing the skewing effects of the matchmaker and factoring in player skill. I find this to be a hilarious statement. A rating is a measure of your performance, skill is something completely difference. A rating depends on multiple factors including Skill, matchup, conditions(maps), and other factors(maybe a player is really good at laddering or at 4 gating, but bad in lategame) I've tried to think of any formula that is possible to use in this situation, but here is the problem. They are using a bayesian algorithm. Bayes predicts winning odds against another player. So when the system believes that you should win and you lose, then either 1. You aren't as good as the system thinks 2. Your opponent is better than the system thinks 3. There is recent change which has caused the system to change. There is no other information that can be gathered. When blizzard releases their magic formula please let me know. | ||
|
DaemonX
545 Posts
| ||
|
Naoko
Sweden53 Posts
None of these numbers mean anything else. | ||
|
MrCon
France29748 Posts
On November 08 2010 16:26 darmousseh wrote: they released the formula @blizzcon, havn't they ?I find this to be a hilarious statement. A rating is a measure of your performance, skill is something completely difference. A rating depends on multiple factors including Skill, matchup, conditions(maps), and other factors(maybe a player is really good at laddering or at 4 gating, but bad in lategame) I've tried to think of any formula that is possible to use in this situation, but here is the problem. They are using a bayesian algorithm. Bayes predicts winning odds against another player. So when the system believes that you should win and you lose, then either 1. You aren't as good as the system thinks 2. Your opponent is better than the system thinks 3. There is recent change which has caused the system to change. There is no other information that can be gathered. When blizzard releases their magic formula please let me know. | ||
|
EmilA
Denmark4618 Posts
And I am also suspicious about the release of these numbers. Last time a patch hit shortly afterwards - maybe this is the case once more? | ||
|
paralleluniverse
4065 Posts
On November 08 2010 16:26 darmousseh wrote: I find this to be a hilarious statement. A rating is a measure of your performance, skill is something completely difference. A rating depends on multiple factors including Skill, matchup, conditions(maps), and other factors(maybe a player is really good at laddering or at 4 gating, but bad in lategame) I've tried to think of any formula that is possible to use in this situation, but here is the problem. They are using a bayesian algorithm. Bayes predicts winning odds against another player. So when the system believes that you should win and you lose, then either 1. You aren't as good as the system thinks 2. Your opponent is better than the system thinks 3. There is recent change which has caused the system to change. There is no other information that can be gathered. When blizzard releases their magic formula please let me know. They released the formula at Blizzcon: http://www.teamliquid.net/forum/viewmessage.php?topic_id=163309#5 It's a formula of a Bayesian estimator. And Bayesian inference is the correct way to go here. Also, rating is the system's estimator of player skill. Thus the two terms are use synonymously. | ||
|
TERRANLOL
United States626 Posts
On November 08 2010 12:09 HughJorgen wrote: If a player's skill rating and league are determined by their wins and losses, then stats on win/loss percentages at a particular skill rating mean nothing. Think of it this way, if they nerfed Protoss into the ground by halving the shields and life points of every unit, then for a while Protoss would lose every match. Then, higher skilled Protoss players would start beating lower skilled Zerg and Terran players despite the massive handicap. Then Battle.Net looks at Huk and assign him a skill rating of :mid gold level:. Then Blizzard run the stats and find that Huk is a mid gold player winning 50% of his matches. Balance achieved! It's not just a matchmaking problem, it's a circular argument. Protoss win 50% of their matches at this skill level = A player's skill level is determined by where they start winning 50% of their matches. There are a few things that slightly improve the situation: 1) random players since they can maintain their 'skill level' by winning more with one or two of the races. 2) The very, very top since you can maintain a win% higher than 50% (there might be more of a particular race here) 3) The very, very bottom where you can maintain a win% below 50% (there might be more of a particular race here) 4) some other measure of player skill that doesn't entirely depend on wins vs losses (apm? resources? game score?) Just to reiterate: saying that they've removed the matchmaking bias is all well and good but unless Blizzard have a method of determining player skill that doesn't depend on wins/losses (e.g. apm, game score, whatever) you'll still get statistics that are heavily biased in favour of balance. We can use average points to make sure that what you're proposing isn't true. | ||
|
Tonyoh
France218 Posts
2 Players with equal skills, and one no matching the other. But yeah... it's totally balanced ? Nowadays I play random and I enjoy zerg in team for their speed and techswitching. This race is way too powerfull. They should nerf the speed bonus given to speedling. Look two basics units : 4 speedgling : outrun reaper, cost almost the same. But you need factory to research the speed upgrade for reapers... No wonder why TLO switch to take zerg, he is a smart and creative players, and see that now zerg : 1) can fast expand without any troubles, 2) have the fastest units, the best mapcontrol, 3) fungalgrowth and badling own m&m ball, as well as speedling/roach, 4) better eco... (can make like 12 drones at the same time) Face it, zerg is currently the best race, and it's more apparent in team game ( I almost never loose as zerg, went 9-1 in 2vs2 with a friend in gold league.) If no furthers changes are made, I will play with the imbalance in my side. =) Nowadays, since reaper speed nerf, and roach up, fast expand is way too easy, and zerg have some really cost effective units (zerglings, roach, badling etc) Queen do very fine against fast banshee or voidray. | ||
|
madnessman
United States1581 Posts
| ||
|
HuHEN
United Kingdom514 Posts
| ||
|
Patriot.dlk
Sweden5462 Posts
| ||
|
Pewt
Canada201 Posts
Diamond, a league which ranges from 6poolers to 4gaters to decent players to pros, is not really representative of anything. | ||
|
dekuschrub
United States2069 Posts
that matchup is so bogus! T v Z is fine! | ||
| ||
