|
Hello!
Here are stats on win rates for each matchup, broken out by league. It's based on about 100,000 1v1 Ladder matches played since the 1.5.3 Balance Update patch. There are also stats available for another ~100,000 1v1 Ladder matches played under the 1.4.3 Balance Update patch.
http://ggtracker.com/winrate_by_matchup
The most imba matchup currently is Bronze Protoss vs Terran, in which Protoss wins 53.6% of the games. Of course we wish it were exactly 50%, but 53.6 isn't that IMBA in my book (especially since I play Protoss).
So here's to hopefully putting an end to "race X is OP" forever!  jk
|
Yes, if we apply Blizzard logic to this: win rate is ~50% on ladder therefore the only logical explanation is that game is balanced.
+ Show Spoiler +Not quite perfect but it does have a hint of sanity in my book.
|
|
|
Interesting how consistent the percentage above or below is between leagues. Over that many games, I guess we CAN say Z < P < T at the low level. Not in an imba way or anything to be concerned about though.
|
On January 15 2013 13:34 Solarsail wrote:
Interesting how consistent the percentage above or below is between leagues. Over that many games, I guess we CAN say Z < P < T at the low level. Not in an imba way or anything to be concerned about though.
You've got both of the signs backwards. At the low level, Z > P > T according to these statistics.
|
the systems tries to keep you to have ~50% winrate
i dont see how these stats are helping by any means
|
On January 15 2013 13:42 ChriseC wrote:
the systems tries to keep you to have ~50% winrate
i dont see how these stats are helping by any means
That's an interesting point.
Let's say, for sake of argument, that my race Protoss really is totally EZEZEZ. Since the system tries to keep me at a 50% winrate, but my race is totally OP, so then Protoss like me would be promoted into higher leagues. We'd see more Protoss in the higher leagues, and the other races more in the lower leagues.
Just eyeballing the stats, there seems to be somewhat fewer Terran cross-race matches played in Masters/Diamond than the other races. Does that mean Terran is the hardest race and there are fewer of them in the top leagues?
|
It means nothing, the system is designed to keep everything at 50% independently of game balance.
A more meaningful measure would be the variation of population profile through the league. And even that has flaws.
|
On January 15 2013 13:51 freakhill wrote:
It means nothing, the system is designed to keep everything at 50% independently of game balance.
A more meaningful measure would be the variation of population profile through the league. And even that has flaws.
Well, one thing it means is that when a Gold-league Protoss meets a Gold-league Zerg, you should expect the Zerg to win ~50.9% of the time (+/- 1.2%). These stats don't tell you if these two players are equally "skilled" in the cosmic sense of the term. But it does tell you that they are evenly matched, given their races and leagues.
|
On January 15 2013 13:50 dsjoerg wrote:Show nested quote +On January 15 2013 13:42 ChriseC wrote:
the systems tries to keep you to have ~50% winrate
i dont see how these stats are helping by any means That's an interesting point. Let's say, for sake of argument, that my race Protoss really is totally EZEZEZ. Since the system tries to keep me at a 50% winrate, but my race is totally OP, so then Protoss like me would be promoted into higher leagues. We'd see more Protoss in the higher leagues, and the other races more in the lower leagues. Just eyeballing the stats, there seems to be somewhat fewer Terran cross-race matches played in Masters/Diamond than the other races. Does that mean Terran is the hardest race and there are fewer of them in the top leagues?
This assume that we start with the same number of players of each race and equal skill. Really though, we know that terran numbers, at least in lower leagues if not overall, are likely inflated by the single player.
On January 15 2013 13:56 dsjoerg wrote:Show nested quote +On January 15 2013 13:51 freakhill wrote:
It means nothing, the system is designed to keep everything at 50% independently of game balance.
A more meaningful measure would be the variation of population profile through the league. And even that has flaws. Well, one thing it means is that when a Gold-league Protoss meets a Gold-league Zerg, you should expect the Zerg to win ~50.9% of the time (+/- 1.2%). These stats don't tell you if these two players are equally "skilled" in the cosmic sense of the term. But it does tell you that they are evenly matched, given their races and leagues.
This is also not true because league and mmr aren't the same. Perhaps most gold league zergs find themselves often matched against higher league opponents, or at least more often than their protoss peers. If this is the case, than we would expect them to peat gold protosses more often. Of course, this is just an example.
|
Sadly, this is the same way DB views balance. Of course it's 50-50, that's what the system enforces lol.
|
This doesn't actually mean anything due to how MMR works, but it's nice to have for the people that think that their race is underpowered in gold league, who ignore the aforementioned fact.
|
On January 15 2013 14:04 Jarree wrote:
Sadly, this is the same way DB views balance. Of course it's 50-50, that's what the system enforces lol.
Blizzard's matchmaking developers have metrics specifically designed to measure the extent to which game balance affects matchmaking. No, they haven't described exactly how these metrics work, but they probably do a decent job.
|
So terran suck on the ladder, how surprising.
On January 15 2013 14:32 Lysenko wrote:Show nested quote +On January 15 2013 14:04 Jarree wrote:
Sadly, this is the same way DB views balance. Of course it's 50-50, that's what the system enforces lol. Blizzard's matchmaking developers have metrics specifically designed to measure the extent to which game balance affects matchmaking. No, they haven't described exactly how these metrics work, but they probably do a decent job.
And that explains why the only race foreigners are able to play at the highest lvl is Z?
|
I would think that this would not mean too much, as you are supposed to have a 50% win/loss ratio on ladder.
If you are going to do ladder analysis you should start at the very top of the ladder. But the only "practical" analysis of statistics you're going to get are from tourney results, anyhow.
Pretty cool to see 100k games worth of data, though.
|
Toss imba. Evrysingle tournament theres at least one toss in top 3
|
So even with Blizzards imba matchmaking the system still is unfavorable towards terran?
|
These are some nice statistics regarding lower leagues, but I love how everyone is like "ehmahgerd system tries to keep a 50% so X race is still imba", when if the system was like 80% for Zerg everyone would be like "LOL blizzard is so right i love you guys"
|
Yeah, balance is with us. Just Terrans are fucked. Nothing new.
|
On January 15 2013 14:45 sAsImre wrote:So terran suck on the ladder, how surprising. Show nested quote +On January 15 2013 14:32 Lysenko wrote:On January 15 2013 14:04 Jarree wrote:
Sadly, this is the same way DB views balance. Of course it's 50-50, that's what the system enforces lol. Blizzard's matchmaking developers have metrics specifically designed to measure the extent to which game balance affects matchmaking. No, they haven't described exactly how these metrics work, but they probably do a decent job. And that explains why the only race foreigners are able to play at the highest lvl is Z?
Balance across ladder =/= balance at pro levels. You'd be surprised to know they have many, many more casual players in lower leagues whom they're also beholden to, and whom are probably interested in knowing where their race sits at x league.
|
On January 15 2013 13:50 dsjoerg wrote:Show nested quote +On January 15 2013 13:42 ChriseC wrote:
the systems tries to keep you to have ~50% winrate
i dont see how these stats are helping by any means That's an interesting point. Let's say, for sake of argument, that my race Protoss really is totally EZEZEZ. Since the system tries to keep me at a 50% winrate, but my race is totally OP, so then Protoss like me would be promoted into higher leagues. We'd see more Protoss in the higher leagues, and the other races more in the lower leagues. Just eyeballing the stats, there seems to be somewhat fewer Terran cross-race matches played in Masters/Diamond than the other races. Does that mean Terran is the hardest race and there are fewer of them in the top leagues?
Yeah, I think the most interesting thing about this is the discrepancy in sample size for each match up by league. Damn few Terrans higher up.
tvx winrates are almost entirely within a good balance ratio, but there are nearly 4x as many zvp's as there are tvp's recorded in gm (by this program, so I guess it could be coincidence). Whatever it means, if it even means anything, that's pretty sad 
|
On January 15 2013 14:55 TimKim0713 wrote:
Toss imba. Evrysingle tournament theres at least one toss in top 3
Yes, if only.
|
|
|
tvx winrates are almost entirely within a good balance ratio, but there are nearly 4x as many zvp's as there are tvp's recorded in gm (by this program, so I guess it could be coincidence). Whatever it means, if it even means anything, that's pretty sad
My stats on GM are not very complete, I need to do some more work to be able to comprehensively retrieve GM matches.
Someone who reviewed these stats suggested focusing on the stats of Random players. What do y'all think? Would those be more indicative of balance?
|
On January 16 2013 01:11 dsjoerg wrote:Show nested quote +tvx winrates are almost entirely within a good balance ratio, but there are nearly 4x as many zvp's as there are tvp's recorded in gm (by this program, so I guess it could be coincidence). Whatever it means, if it even means anything, that's pretty sad My stats on GM are not very complete, I need to do some more work to be able to comprehensively retrieve GM matches. Someone who reviewed these stats suggested focusing on the stats of Random players. What do y'all think? Would those be more indicative of balance?
Im curious who suggested it, lol. If you mean the balance between the three races, it wouldn't indicate anything. It'd be interesting to see, though.
|
On January 16 2013 01:15 rd wrote:Show nested quote +On January 16 2013 01:11 dsjoerg wrote:tvx winrates are almost entirely within a good balance ratio, but there are nearly 4x as many zvp's as there are tvp's recorded in gm (by this program, so I guess it could be coincidence). Whatever it means, if it even means anything, that's pretty sad My stats on GM are not very complete, I need to do some more work to be able to comprehensively retrieve GM matches. Someone who reviewed these stats suggested focusing on the stats of Random players. What do y'all think? Would those be more indicative of balance? Im curious who suggested it, lol. If you mean the balance between the three races, it wouldn't indicate anything. It'd be interesting to see, though.
Focusing on Randoms stats was suggested here and here.
I agree that Random stats are hard to interpret because the player is at a "training handicap" to their non-Random opponent. But the comparison of RTvP to RPvT seems pretty clean. Unless you believe that Random players are more practiced in one race than another?
|
lol, and so across the board except in masters, terran loses to the other races more often than they win. good to know =P
|
On January 15 2013 14:50 Qwyn wrote:
I would think that this would not mean too much, as you are supposed to have a 50% win/loss ratio on ladder
If TvP is flawed at your level, you could get let's say a 60% win ratio as Terran against players "your level" on that matchup. Since you have an overall winrate at 50%, your winrates will probably be something like 55/50/45 with 45% on your TvZ even though matchup is correct.
What is truly difficult is that a flawed ZvT with 60% winrate Z will end up with exactly the same result overall.
If T is favored agains both Z and P, you end up with a 50/50/50, with a population shift in MMR that could potentially be noticed (see http://www.teamliquid.net/forum/viewmessage.php?topic_id=351786).
|
On January 16 2013 01:33 dsjoerg wrote:Show nested quote +On January 16 2013 01:15 rd wrote:On January 16 2013 01:11 dsjoerg wrote:tvx winrates are almost entirely within a good balance ratio, but there are nearly 4x as many zvp's as there are tvp's recorded in gm (by this program, so I guess it could be coincidence). Whatever it means, if it even means anything, that's pretty sad My stats on GM are not very complete, I need to do some more work to be able to comprehensively retrieve GM matches. Someone who reviewed these stats suggested focusing on the stats of Random players. What do y'all think? Would those be more indicative of balance? Im curious who suggested it, lol. If you mean the balance between the three races, it wouldn't indicate anything. It'd be interesting to see, though. Focusing on Randoms stats was suggested here and here. I agree that Random stats are hard to interpret because the player is at a "training handicap" to their non-Random opponent. But the comparison of RTvP to RPvT seems pretty clean. Unless you believe that Random players are more practiced in one race than another?
There is an inherent disconnect between a random player's subjective skill and his rating when he can potentially be thrown against statistically superior opponents riding on the advantage of playing random in lower leagues. It would be clean as RTvP and RPvT but it wouldn't really be indicative of TvP/PvT etc.
edit: It's just as possible for a random player to be better in a match-up than a normal Terran be better than a normal Protoss in TvP/PvT -- but that wasn't the point.
|
|
|
|
|
|
EPT
