|
Introduction
This is something I hear a lot of fucking arguments about, especially from people absolutely convinced that Protoss is either stronger or weaker than the other races. I saw a thread recently about race balance at D level in ICCUP, and while interesting, it doesn't really have much impact on the higher levels of the game, which are a lot more interesting to me than my own D-level (on a good day) play. So, I grabbed some information off of TLPD and analyzed that shit.
Note: I'm a philosophy student. Math and statistics are not my strong point. There are probably at least a few minor errors in here that I didn't catch, even though I made my computer do most of the figgerin' for me.
Details
Now, on to the good stuff. Basically, the charts below represent the average ELO of each player, for all races (including 'all') against all races (including 'all'). The ELO is calculated separately for each opposing race (vA, vT, vZ, vP) but together for all races (Protoss are measured against all races in their vT, but vT is not measured against vP). What this means is that if Terrans have a higher ELO than everyone else at vZ, that's significant, but if Terrans have a higher ELO vZ than vP, that isn't.
All players:
Oh yeah, protoss is imba, all right. Except it isn't. Terrans are far and away outperforming everybody else, largely due to dominance in TvZ. Zergs come in second due to strength ZvP, and Protoss comes in last because they're barely better vT (their "strong" matchup) than Terrans are. And note that Terrans are better against Protosses than other Protosses are. This means that assuming completely 'average' gamers, your best chance of beating a Protoss is with a Zerg, your best chance of beating a Zerg is with a Terran, and there's no significant difference between countering Terran with Protoss or another Terran.
Top 50 players:
Wait, what? Now Terran is significantly better vT than Protoss, and Zerg is catching up? Admittedly, Protoss are doing better vZ now, but they're still dead last there, as well as in the overall standings. What's going on?
Oh, wait, but look! Now Protoss are doing slightly better mirror than Terrans are against them. Basically, what this means is that a lot of top Terran players are relatively strong at TvP and TvT (but weak at TvZ), and the top Protoss players are relatively strong at PvZ.
Conclusions
Wow, that was actually reasonably enlightening. Basically, in progaming, T>Z, Z>P, P=T (look how TvP is just barely stronger than PvP, and PvT is just barely stronger than TvT). This isn't to say that Terran is inherently the strongest race and Protoss is inherently the weakest - I'd have to actually do work to account for all the potential factors that influence why this turned out the way it did - but Terran is currently performing the best overall, and Protoss is currently performing the worst overall.
Now what's really interesting to me is how things change when you look at the top 50 instead. TvP loses ground to PvP, and PvT loses ground to TvT. TvZ looks way less disgustingly good and PvZ and ZvT look like they actually stands a snowball's chance in hell. What this means is that, on average, the difference between the best of starcraft players and the rest of starcraft players is that the Protosses are way better against Zergs, and the Zergs are way better against Terrans, but the Terrans improve relatively equally across the board. I'm not good enough to speculate as to why that is, but I'm sure some of the rest of you have theories.
|
Valhalla18444 Posts
This doesn't take into account things like maps and player skill. You can't determine game balance by looking at statistics so simply.
|
Most of this stuff is known, but thanks for the input you put in  . (If your another protoss whos angry that people say its easier, then  , but otherwise good work.)
EDIT: Also what fakesteve said about maps.
|
Terrans improve across the board because they are mechanic driven race. Protoss is more reactive and since scouting in ZvP is so crucail it makes it so that when you miss something as toss you have a high chance of losing, increasing the loss rate for them. Zerg I think are about balancing larvea with drones and units and the balancing differs from MU to MU and from situation to situation. Terran on the other hand just have to have good macro and micro and be aware of what is going on. Their role does not shift dramaticly from MU to MU unlike toss or zerg.
|
Protoss is the hardest race to play (and perform well with) at the top level, this has been well known for a long time. I attribute a lot of this to PvP has the lowest peak ELO of any matchup, and the lowest elo average as well (it's the most luck based. One unit, DTs.).
|
On July 30 2009 15:54 FakeSteve[TPR] wrote:
This doesn't take into account things like maps and player skill. You can't determine game balance by looking at statistics so simply.
Listen to fakie.
Still, thanks for the hard work.
|
On July 30 2009 15:54 FakeSteve[TPR] wrote:
This doesn't take into account things like maps and player skill. You can't determine game balance by looking at statistics so simply. Well...even if you take statistics from the entirety of progaming history, or just generally accepted truths about starcraft, they both agree with his above statistics. T>z>p=t
thats always been the formula. Tell me a period in starcraft history when it was otherwise outside of maybe 1999?
|
On July 30 2009 16:07 Dazed_Spy wrote:Show nested quote +On July 30 2009 15:54 FakeSteve[TPR] wrote:
This doesn't take into account things like maps and player skill. You can't determine game balance by looking at statistics so simply. Well...even if you take statistics from the entirety of progaming history, or just generally accepted truths about starcraft, they both agree with his above statistics. T>z>p=t thats always been the formula. Tell me a period in starcraft history when it was otherwise outside of maybe 1999?
I think another part of this is that protoss is the race who has always been the most dependent upon maps for balance, especially for pvz (FE wasn't easily viable on many maps until recently, etc. Neo Medusa is a great example of the way a map that was once balanced can drastically swing the other way after Zergs figure out zvp on it).
|
Look at maps , players skill these are with pro gamers. terran always dominated in the pro scene but how about us in iccup everyone complain because of how much cheese they can do and get away with it terran really does require more apm to pull off
|
Nice job, it seems in general right now a lot of terrans are struggling with TvZ, though not nearly to the degree that protoss players have been struggling with PvZ. I guess there's just more mid-level terrans than mid-level zergs right now.
|
The argument isn't that protoss is stronger, but that it is easier to be good with them.
i.e. the learning curve is much easier for protoss than the other races, even if it doesn't go up further
Not saying that I necessarily have this viewpoint...
|
your balance conclusions are exactly right and the people saying otherwise have already drank the koolaid. fuck the haters. the cavailer dismissal of your statistics is pretty disgusting when the results seem to reflect the opinions of the professional league. what are the chances that it is a mere coincidence?
this of course confirms my suspicion that protoss players are, as a group, the most talented, hardest working and best looking.
as for your top 50 results the explanation is that you can't get into the top 50 if you can only win 2 of 3 matchups consistently. protoss's weak matchup is pvz, and without being able to win pvz, protosses can't get into the top 50. zerg's weak matchup is zvt, and without being able to win zvt zergs can't get into the top 50. meanwhile, since there is no weak terran matchup (or tvp very marginally so) terran players don't have a similar deficit and so you see the terran in the top 50 improve relatively homogenously in each matchup
|
Valhalla18444 Posts
uh for the record all i said is that there needs to be more thought put into the matter, accounting for maps, shifts in the metagame and how long it took the playerbase at that time to adapt, etc. i didn't say the conclusion was incorrect 
There has also been a significant lack of talent in the Protoss player pool compared to the Zerg and Terran player pool at certain points in progaming. Protoss players emulating Stork's cheese builds (who only he was consistent in winning with) and getting rolled for a period of like 8 months messes the statistics, Protoss players being unable to deal with Savior's skullcrushery for so long until Bisu provided an answer, stuff like that.
|
I agree terran is the strongest race but it also requires the most skill to play.
|
Baa?21244 Posts
All this says is to ban Terran =D
"I agree terran is the strongest race but it also requires the most skill to play."
Zerg...D:
|
Very interesting stats. Thanks for the hard work.
Of course, as everyone has already said, it's nothing to draw concrete conclusions from.
|
Uhm.... Terran vs Terran ELO over all active progamers is a completely meaningless figure. In Terran vs Terran, one Terran is always going to win and one Terran is always going to lose. In other words, the Terran vs Terran ELO should in theory be at 2000. The fact that it is higher than that only proves that there has been ELO inflation.
For Top 50 gamers, these numbers like "Terran vs Terran ELO" are relevant, and what they show is the following: in Terran vs Terran, there is less of a luck component than in Zerg vs Zerg and Protoss vs Protoss. We can deduce this by seeing that the top 50 Terran players are more consistent against their own race than the top 50 Zerg and Protoss players. However, again we have to be concerned against the possibility of ELO inflation and how it affects each race separately. This is not an easy task to do.
What is ELO inflation and why does this come about? Unfortunately, this particular topic is liable to open up a veritable shitstorm on the forums, particulraly regarding the "why" as in theory ELO inflation should not happen.
However, observe the following: over time, player ELO ratings have risen across the board in Starcraft. How do we know this? Consider Boxer, for instance. His peak ELO was 2224. If you look at players right now, there are no less than 5 gamers with an ELO higher than Boxer's peak. At the time Boxer attained that ELO value, he was the undisputed bonjwa, with a whopping 87% wins in KeSPA sanctioned matches. A good observer might point out that back then there were simply fewer games being played, so it's harder to attain a high ELO rating (and this may in fact be the cause of inflation for players at the high end -- players today have more opportunities to demonstrate their skill while at peak level; certainly Boxer and NaDa and Oov never had schedules where they were playing in competition on an almost daily basis), but if we look at his first 98 games played, his record was still over 68% against the most difficult players of his day. Furthermore, because there wasn't a proleague back then, it should in theory be easier for a high ranked player to maintain a high ELO rating, since proleague is a venue in which a high ranked player is statistically more likely to end up competing against a player with a low ELO value and have a chance to lose!
Basically, the bottomline is that your points, while supported by fancy looking graphs, are not as well grounded as you think they are.
|
Oh yeah, protoss is imba, all right.
This isn't the argument at all. The argument is that protoss is EASIER to play, not the better race. If you could play the game in slow motion, at 5% speed and do everything perfectly, protoss would probably be the weakest race.
|
In the history of competitive starcraft, every race has manifested through prodigies and 'bonjwas' who have dominated other players at some point or another.
Ultimately the "imba race" argument is irrelevant because the holes in a player's game are so rarely from game imbalance, but rather their own mechanics and skills.
Of course I have my own opinions about imbalances like anyone, but what makes aspects of the game difficult at an amateaur level is not what makes it difficult at the pro level
|
On July 30 2009 15:54 FakeSteve[TPR] wrote:
This doesn't take into account things like maps and player skill. You can't determine game balance by looking at statistics so simply.
Agreed, and I thought I made that clear. Note that I don't even look at win rates. All this is talking about is what races are best suited to defeating the other rates, given this set of players and the set of maps that they're playing on (and all the other factors that happen to exist)
|
|
|
|
|
|
EPT![[image loading]](http://i27.tinypic.com/2wmhnd5.jpg)
![[image loading]](http://i30.tinypic.com/dnzewn.jpg)
