aligulac.com Go here for interactive graphics, including number of games the stats are based on. Below is from Aligulac.
Balance report
The following chart show the historical month-by-month winrates for each non-mirror matchup in the database.
Note that this yields information about metagame balance near the top of the skill ladder, and is not to be confused with (although likely correlated to) actual game balance throughout the whole player population.
The winrate is shown with respect to the race listed first, so if the Protoss vs. Terran graph shoots up, that means Protoss is doing well and Terran poorly.
The performance difference chart shows the approximate difference between actual performance as evidenced by results and predicted performance by rating. For example, if, averaged over a whole month, all Terrans had mean rating difference -100 (which is to say, their mean rating was equal to 100 less than that of their opponents), but they performed as if they had a mean rating difference of +100 (about 54.5% winrate), then that amounts to a performance difference of +200, which is to be interpreted as that Terran players overall performed 200 points better than expected.
This chart can avoid some of the problems with the above, which can potentially be influenced by incredibly hot streaks from one or two singularly great players. However, as ratings catch up to the performances of the players, this chart will tend toward equilibrium, even if balance never changes.
Some stats Using stattrek.com This is by the OP, not aligulac.
I am summing the stats for November to end of January (so all of LotV, plus last week of HotS) for each matchup, and do some binomial tests to get p-values for balanced matchups.
PvZ: 1712/3928 = 43.6% (singel sided) p-value for 50%: less than 1e-6 (single sided) p-value for 45%: 3%
TvZ: 1836/3665 = 50.1% (singel sided) p-value for 50%: 46%
PvT: 1265/2511 = 50.4% (singel sided) p-value for 50%: 36%
These number show incredibly strong evidence that zergs beat protosses most often than not in this set of games. They also show well balanced winrates close to 50% in both TvP and TvZ.
However, the binomial test assumes that the games are independent, which they are not of course. The same players play several maps each time, and each (well, most) players play many series. This can create false significance in this binomial test, like the Z > P above. We can handwavingly take this into account by cutting down the number of independent observations (degrees of freedom), while maintaining the win ratio. Looking at ZvP (the others already have high p-values), we slash the number of games by 5, turning it into
PvZ (decreased #observations): 342/785= 43.6% (singel sided) p-value for 50%: 1e.7e-4 (single sided) p-value for 47%: 2.9% (single sided) p-value for 45%: 22%
These numbers still show a very strong signal of zerg winning more over protoss, although the evidence for a winrate over 55% isn't as compelling any more.
You can also add some multiple hypothesis testing for the three matchups, and double-sided p-values, which would further weaken the signal a bit, pushing back the point of 5% significance to 48% probably. Nonetheless, all these corrections doesn't change the fact that the measured winrate is 43.6%. It only changes how likely it is to be coincidental from a true winrate of 50%. In conclusion, there is striking evidence that zerg has a larger than 50% winrate in this data set, some evidence that it is larger than 53% and indications that it is above 55%.
I am a zerg player, so it is not in my interest to report a Z > P imbalance.
Although I have been trying to push the importance and significance of the signal we saw last month, so I guess I am biased to reproduce what I have been whining about since last month. Which it has been. but would I have written this OP if the January stats had been 50% for ZvP? Probably not. Which is called publication bias.
edit: oops, screwed up PvZ and ZvP. >_> fixed, sorry. Zerg is winning more over toss.
On February 02 2016 21:23 ZeRoX_TV wrote: Protoss is very UP after this patch and any stats cant deny it. I am thinking of quitting game until this is solved. Unplayable atm.
haha, yes, you are right. I accidentally wrote ZvP where it should have been PvZ. :o) Fixed now.
How much influence does it have, that the foreign scene has almost double the amount of notable Zerg players over Terran and Protoss (who seem rather close in notable player numbers) ?.
When we expect, that skill is equally devided between all 3 races, shouldnt have Zerg with the double amount of "notable players" have also a higher amount of strong players, who then kick not only more weaker Zergs in ZvZs out, but also win more games against other races then their counterparts from the other races?
Sorry, my english isnt perfect, I hope you understand what I mean.
the data from november/december is pretty much irrelevant in pvz now ; everyone was playing stalker disruptor and it didnt work. its like taking march/april 2013 pvt data and saying oh wow worst winrate ever for protoss and yet it evens out afterwards with no balance change as protoss players adapt to the new medivac speed
plus this map pool is incredibly zerg favoured in zvp
On February 02 2016 22:17 Acer.Scarlett` wrote: the data from november/december is pretty much irrelevant in pvz now ; everyone was playing stalker disruptor and it didnt work. its like taking march/april 2013 pvt data and saying oh wow worst winrate ever for protoss and yet it evens out afterwards with no balance change as protoss players adapt to the new medivac speed
plus this map pool is incredibly zerg favoured in zvp
January doesn't really seem to change the trend much though. Still below 45% win rate. But yeah, you never know if the meta will change and turn things around I guess. Do you know of any new OP PvZ builds on the rise? Would you tell if you did?
Should say that the HotS PvT thing only went for a month, about 750 games. We have 3 months and almost 4k games now. How long should we wait for the meta to change?
And yes, whenever you speak about balance, it is always on the current maps. So patching the units or the maps will have similar effect.
How will the upcoming unit/map patch affect ZvP do you think?
On February 02 2016 22:17 Acer.Scarlett` wrote: the data from november/december is pretty much irrelevant in pvz now ; everyone was playing stalker disruptor and it didnt work. its like taking march/april 2013 pvt data and saying oh wow worst winrate ever for protoss and yet it evens out afterwards with no balance change as protoss players adapt to the new medivac speed
plus this map pool is incredibly zerg favoured in zvp
Yes, the difference between stalker/disrupter and mass pheonix into chargelot archon is night and day, performance wise.
LotV is too volatile to be drawing conclusions from results right now.
On February 02 2016 22:17 Acer.Scarlett` wrote: the data from november/december is pretty much irrelevant in pvz now ; everyone was playing stalker disruptor and it didnt work. its like taking march/april 2013 pvt data and saying oh wow worst winrate ever for protoss and yet it evens out afterwards with no balance change as protoss players adapt to the new medivac speed
plus this map pool is incredibly zerg favoured in zvp
I dont think PvZ is a matchup that will stabilize on its own in the near future. The fundamental balance of PvZ was just thrown out the window.
In Wol / Hots, protoss had the strongest ground army in the lategame - this meant even if zerg takes 10 more bases, there was always a chance to come back into the game.
Zerg has the faster economy, easier to hold own expansions, ease to punish enemy expansions and now also the better setup in the lategame with ultralisks / lurkers / broodlords / vipers / cracklings.
Aside for some punishing timings from protoss - theres no straight up macro game to win anymore.
Your problem is assuming you should still be in the game if another player takes 10 bases, and that it was fine when that was the case.
A huge crutch was taken away from Protoss this last patch - the ability to defend basically any early aggression from zerg without units.
It'll take some time for that to shake out.
Also FYI - Zerg econ isn't actually faster in the first 5 minutes of the game or so. Protoss out drones Zerg until later on thanks to Chronoboost and doesn't have to choose between workers or units.
EPT![[image loading]](http://i.imgur.com/gkvImgt.png)
![[image loading]](http://i.imgur.com/VPubjUy.png)
![[image loading]](http://fs5.directupload.net/images/160202/vfgnap53.png)
![[image loading]](http://i.imgur.com/UA2ljlf.png)
![[image loading]](http://i.imgur.com/idXiryz.png)
