DIR Return Create A Forum - Home
---------------------------------------------------------
Zerg Hex Forum
HTML https://zerghex.createaforum.com
---------------------------------------------------------
*****************************************************
DIR Return to: Balance Discussion
*****************************************************
#Post#: 1261--------------------------------------------------
Skill Bracket Data from NA, Previous Patch
By: deorai037 Date: December 11, 2020, 3:31 am
---------------------------------------------------------
I don't know how other people define skill levels, but in my
spreadsheet full of player winrate data (375), I arbitrarily
divided people up into a number of skill brackets based on
winrate (obviously this is not a perfect measure of player
skill, but I digress)
-------- Info about Method/Data set--------------
Over the course of about 4 months, I casually collected player
winrate data from most of the games I played on the NA server. I
recorded the following data: SC2 Handle, Terran wins, Terran
Losses, Zerg Wins, Zerg Losses, date of data collection,
possible side notes (such as "bank reset"). Because the data was
pulled exclusively from the NA server, noticeably, many players
based largely in EU are going to be missing from this data set.
Terran W/L rate and Zerg W/L rate were calculated separately;
there are many players with significant skill disparities
between the two factions. For obvious reasons, W/L rate was
defined as (wins as faction) / (wins + losses as faction). This
is an extremely crude measure of skill that makes broad
assumptions, but it is a useful measure nonetheless. The
boundaries between categories are somewhat arbitrary, as I quite
simply divided the player base into 7 equal-winrate chunks. So,
the categories look like this:
New: 0~14% winrate
Very low skill: 14~28% winrate
Below Average/Low skill: 28~42% winrate
Average skill: 42~57% winrate
Above average skill: 57~71% winrate
High skill: 72~86% winrate
Very High skill: 86~100% winrate
I found this to be an excellent way of dividing up the
playerbase. A player with a 45% winrate tends to share a lot in
common with players with 55% winrate, so I found that the
"average" skill category covered things more nicely when it
included a broad center (as opposed to having, say, 35~50% be
considered below average and 50~65% be considered above
average). The brackets are just about the right size to make
useful observations about player trends.
There are many potential flaws with this crude data set (these
were gathered casually and for personal use). First, players
with low numbers of games often become outliers, comprising a
decent chunk of the "New", "High Skill," and "Above Average"
categories. Second, there are quite possibly a number of
potential smurfs hiding in the woodworks of the data. Third,
winrate data can be manipulated in both directions; a player who
always plays with a guaranteed teammate or clanmate will have an
inflated winrate compared to a pub player who needs
significantly higher skill in terms of being able to communicate
with and coordinate a group of strangers in order to win. On the
other hand, a player who arbitrarily quits a lot of games before
victory can intentionally tank their winrate for reasons
unknown.
----------------Analysis-------------------
There is only one player in my recorded data who ever reached a
winrate high enough to be counted as "Very High skill:" Najdorf,
obviously. This is for Terran play; there were no Zerg players
of "Very High Skill" at the time recorded.
Of the 375 players in the data set, only 17 players are ranked
as "High Skill" for Terran. However, some of these players have
small numbers of games (whether due to smurfing, bank resets, or
being originally from EU), so if we exclude players with less
than 30 total games from this category, we end up with 12
players. Of these 12, one is a known smurf of another (lol, it's
my alt. I accidentally included my main and my alt, both of
which have at minimum over a 100 games apiece, in the data set
separately). So of the 375 players, we have only 11 who are
"High Skill."
Brief side note about my alt, I play on it when I'm tired so the
win rate on RedNova should be lower than my main.
"High Skill" Terran players, then, represent roughly 3% of the
NA population. What about "High Skill" Zerg players? Similarly
to the above, there are about 17 players who are ranked as "High
Skill" for Zerg by the same crude win rate metric. Of them,
three have fewer than 30 combined games on Zerg, so excluding
them from the metric we end up with 14 "High Skill" Zerg
players. We also have to exclude me, *again*, so that leaves a
total of 13 "High Skill" Zerg players representing 3.5% of
players on NA.
Brief personal note about my Zerg skill level, my skill level is
entirely relative to pubs, where most Terrans are not that good,
so as an above-average Zerg player I can very easily crush them
while making a bunch of really crucial mistakes (unless there
are multiple above average Terran players and/or I am having to
teach my partner how to play Zerg, which makes up a significant
chunk of losses). I am certain I would fare substantially worse
as Zerg in an all-around high skill game.
-----------------Terran Skill Bracket
Trends--------------------------
Despite how rough of a metric this is, I've found that there are
certain patterns, playstyles, and in-game decisions that tend to
be common throughout a skill bracket. The following description
is for Terran players.
Most players in the "High Skill" bracket are capable of playing
very aggressively with early marine pushes, and can easily
destroy a noob Zerg's base by the 21~22 minute mark while
playing solo (if not even sooner than that) if they so desire.
They tend to have fairly strong base awareness and once they
have established their base's walls and gotten early eco up, are
nearly impossible to push in. Unless you're able to
significantly pressure them very early on and force them to stay
off of 3 gas, they will probably kill you as Terran. To have a
high T and Z win rate, their micro skills must also be fairly
well developed enough to handle the early game. They are also
often much more strategically flexible, able to adjust to the
situation at hand on the fly. If they are left alone, they are
able to maximize their greed to emerge into a powerful midgame;
if they are under pressure, they know how to build just enough
forces to have some money left over to build a decent economy.
Even under heavy pressure, high skill players tend to be able to
hold at least the first outside gas and a fruit farm for the
first 5 minutes. They are also able to recognize when a Zerg has
overspent themselves, and swap builds accordingly (i.e., an
overspent zerg means that t2 marines will be in high enough
numbers to rip through them mid/late game, so they might not
rush to try and wipe said zerg with t1s and go for a stronger
later push).
People in the "Above Average" skill bracket tend to play
similarly to "High" skill players, at least, after the 10 minute
mark. Usually, there tends to be some fundamental skill area in
which they are weaker -- usually in terms of aggression or
timing. Still, when push comes to shove, most players in the
"Above Average" skill bracket tend to know what they're doing
even if they're not perfect at it. They're usually able to make
a partial recovery if they're initially focused, and if the game
drags on late without their positions collapsing they tend to be
very strong in the late game. By no means are "Above Average"
players bad, and they make for quite reliable partners most of
the time, especially if there's a "High" skill player available
to take the early heat for them.
People in the "Average" skill bracket tend to have multiple
fundamental flaws in their playstyle and a poor understanding of
the flow of the game. Where "High" skill players have a very
strong, intuitive understanding of how much the other team is
spending on upgrades (aka, how much their eco is slowed) and so
when the power spikes of Terran and Zerg are, "Average" players
tend to be completely clueless with regards to game flow, and so
they tend to lack a strong plan (ask them, "when do you plan on
pushing?" and their response is essentially shockedpikachu.gif).
If the game drags on ultra late, then "Average" players can
sometimes shine since they'll still be able to field a
formidable force by late game, but prior to the 30 minute mark,
it's hard to expect Average players to push beyond their first
hatchery before then since they don't know how to push a big
spine without vets or elites. Some Average players also have
fairly bad micro skills and low awareness, and so it's easy to
snipe an unwatched side-wall of theirs. When placed under early
pressure, Average players tend to buckle fairly quickly, and
have little to no recovery. When given free eco, Average players
lack the same understanding of greedy terran eco that Above
Average and High skill players have, so they will sometimes have
ecos the same size as a high skill player who's been forced onto
high ground.
"Below Average" players tend to have severe systemic issues in
their understanding of the game; their micro skills are often
remarkably poor, and they build very, very little eco throughout
the game. They have very little understanding of the flow of the
game, and more often than not you'll see them attempt to push
the first hatchery.... at 22 minutes..... with mass reapers....
after +2 movespeed +4 attack fully upgraded tar roaches are out.
Actually, "Below Average" players will almost never actually
attack past the first hatchery, if at all. Unlike Average
players, who tend to at least have a vague awareness of the fact
that they'll need to attack the Zerg at some point, "Below
Average" players seem to not be aware that Zerg Hex is, in fact,
not a tower defense game, and have no idea what the goal of the
game is.
"Very Low" skill players are essentially players who have no
clue how RTS mechanics work in the first place, or just haven't
figured things out at all. Sometimes, they will build large
walls in imitation of better players, but behind those walls is
just a bunch of empty space, unoccupied by and farms or
buildings. With some "very low" skill players, there is a risk
of accidentally killing them off too early as Zerg, because some
are bad enough at the game that they can't handle split zergling
rallies; their micro skills are too poor to figure out "hold
position marines, patrol the battery," thereby requiring active
effort from the Zerg to keep them alive. You're lucky if a very
low skill player actually figures out how to construct a one-gas
turtle wall (a typical hallmark of below average and low skill
players) that is actually covered, in its entirety, by a
construction yard's aura. Occasionally, you will find con yards
built in the hidden back area of the Terran base. Even if they
do manage to turtle, it is highly unlikely that a very low skill
player will ever actually ion; their very presence in the game
is a major boon to the zerg team.
Also, many "Low" and "Very Low" skill players do not know what
MIA calls are.
One of the most fascinating examples of a low skill player is
one of the Kachinsky-profile barcodes on NA, who has a Terran
W/L rate of 336/601 (Low Skill) and a Zerg W/L rate of 6/21
(Very Low Skill). This is a player with nearly a thousand games
and yet remains absolute garbage at the game. Having played
against them a few times, I've realized that they can construct
an impressive looking base that is made entirely out of paper;
as Zerg, I could safely ignore them up through roaches, after
which they would inevitably collapse.
-------------------------Big Number? I
focus------------------------------
Many players do not seem to realize that the stat screen doesn't
actually tell you the level of player skill (it shows you the
number of wins + (1/4?) the number of losses), and instead they
see "big number; I focus that guy then."
Here I will propose multiple interesting effects that arise from
this misconception based on the stat menu.
First, high skill players will see their winrates slightly
decrease as they accumulate more games in their banks. I'm sure
many other high skill players have already felt this; you get
focused, it's harder to win under extra pressure, you don't get
to shine and instead play "Zergling Hell Simulator" for 20
minutes. This much is obvious.
Second, and more hilariously, low skill players will see their
winrates slightly increase as they accumulate more games in
their banks. This is, however, not really due to an increase in
player skill (player winrates often tend to stabilize after the
first 150 or so games). Instead, because a low skill player with
a big number gets focused more often, they end up taking
pressure in place of better players on their team (this is, in
fact, why I started keeping track of winrates in the first
place). So, a low skill player actually becomes useful because
they have 7 spawners built on them (because "omg 340 'wins'?")
while their above average skill neighbors get to eco and carry
their sorry asses.
One example of a player in this category is a guy named Jack,
who I've mentioned elsewhere in these forums. Although this
player, with 1311 wins and 926 losses, is technically considered
"Above Average," he is, at best, an average/mediocre skill
player. He automatically defaults to turtling and has incredibly
poor recovery from early pushes. I very, very vividly remember a
game where I aggressively drew off a Zerg player's attention
thinking it would give him space to re-take his other two gases.
Nothing. That guy literally got zero units on him for about 10
minutes straight, and he didn't build a single extra gas. What a
terrible player, lol. If I hadn't intentionally drawn off focus
(with a really aggressive early push), I probably could've won
that on my own.
Third, high and above average skill players who reset their
banks to have lower game numbers see their winrates improve.
This much is obvious; you get focused less by people who don't
recognize your name, so you can eco harder and are more likely
to win.
Fourth, low skill players who reset their banks (under the
delusion that this will increase their win rate) will actually
see their win rates TANK. It's actually worse for them that
they're not being focused, because their team contribution went
from "resident butt monkey who takes hits for others" to
"useless load that can't eco even with no pressure."
----------------------------------Coordinated
strategy-----------------------------------------------
This data set focused entirely on pub matches, most of which
involved largely random players with relatively small amounts of
coordination. The observations made about skill and player
behavior trends are not as relevant when it comes to coordinated
teams and strategies. For example, it doesn't take much
mechanical skill to pull off clan Italis' Italis cheese strat
(which I love dearly because it's so fucking hilarious to watch
happen), but it does take a lot of coordination and agreement
beforehand to pull it off. Likewise, mass ion rush is an
extremely effective coordinated strategy that requires very
little mechanical skill on part of the players involved (it's
literally a test of how quickly you can eco up on a small base).
---------------------------------Conclusion---------------------
-----------------
This concludes a brief analysis of some of the winrate data I've
gathered and some of the Terran player trends that go along with
it. I hope this provides some useful perspective on what I mean
when I talk about "high" skill, "above average," "average"
players, and I hope it also fosters discussion about what a
useful frame of reference would be in terms of discussing player
skill.
#Post#: 1262--------------------------------------------------
Re: Skill Bracket Data from NA, Previous Patch
By: Adam Date: December 11, 2020, 3:48 am
---------------------------------------------------------
all in all great analysis, only thing that I can think of that
is little wrong when compared to what I see in games is how
above averge put people with 57% in the same category as 70%
I dont know if its only my impression but I feel that terrans
with 66% wr push big spines WAY more often (and successful) than
those with 56%, it seems like only 10% difference but for some
reason it rly shows in practice
btw out of curiosity do u have me in your spreadsheet or I
managed to escape u with my bank res :P
#Post#: 1264--------------------------------------------------
Re: Skill Bracket Data from NA, Previous Patch
By: RickSanchez Date: December 11, 2020, 4:54 am
---------------------------------------------------------
in which category am i? ???
#Post#: 1265--------------------------------------------------
Re: Skill Bracket Data from NA, Previous Patch
By: Najdorf Date: December 11, 2020, 6:03 am
---------------------------------------------------------
Najdorf checking in here. One of the more interesting things
that i have felt regarding win rate is that i actually win at a
slightly higher rate on my main account than on my Alternates.
I find that there are only a few players who can hold focus long
enough for a player to have their greed kick in without getting
wiped out by an f2. But there are many players who can build an
army and do damage if left alone. So when on my main account i
just hold focus and even my negative win rate allies play
average or above average by virtue of not having to deal with
any problems before they have shatter. However when i play on
my alternate other players will get focused and never leave
their base again, or perhaps rage quit making it much more
difficult to carry games.
#Post#: 1268--------------------------------------------------
Re: Skill Bracket Data from NA, Previous Patch
By: deorai037 Date: December 11, 2020, 7:19 am
---------------------------------------------------------
You were listed in my data set as "possible EU or smurf," Adam.
If your SC2 handle on NA was "Rickdeeznuts" I also had you
listed as a "possible smurf"
Again, it's an extremely crude measure of gameplay skill, but
cutting the brackets thinner in the middle would mean that the
brackets would cease to be useful at the top!
---------------------------------------------------------------
I think that's more specific to you, Naj, you have a very
powerful grasp on little details like stub walls and early micro
(they add up!) that lets you hold 2/3 gas where even most other
high skill players would fall back to 1/2 gas. So I think in
most cases, even if your allies are subpar, you yourself often
remain a potent force and potential threat throughout the game,
which wouldn't be the case for most players under heavier
pressure (who would be able to buy their allies maybe 5~10
minutes, but would ultimately fall a lot further behind than you
would). Essentially, the positive effect of drawing aggro
remains but the negative effect of being excessively heavily
focused is mitigated by raw player skill.
I mean.... I only got decent at this game from watching a few
replays of games I had with you lol my winrate shot up from
being like 60% to like 80% within like 15 games. Your t1 game
opened my eyes to the possibilities, there is no spoon
#Post#: 1271--------------------------------------------------
Re: Skill Bracket Data from NA, Previous Patch
By: elon.schools@yahoo.com Date: December 11, 2020, 9:02 am
---------------------------------------------------------
adam is hard smurf, he can easily destroy a zerg single handed
if the zerg does not have skill to match him if. rick is not as
good but defiantly not someone to leave alone.
#Post#: 1272--------------------------------------------------
Re: Skill Bracket Data from NA, Previous Patch
By: Speed Date: December 11, 2020, 9:27 am
---------------------------------------------------------
[quote author=That Strait Smith link=topic=134.msg1271#msg1271
date=1607698953]
adam is hard smurf, he can easily destroy a zerg single handed
if the zerg does not have skill to match him if. rick is not as
good but defiantly not someone to leave alone.
[/quote]
Sorry, but who cannot destroy a zerg when the zerg is way worse
than you? xd
#Post#: 1283--------------------------------------------------
Re: Skill Bracket Data from NA, Previous Patch
By: ZergTriumph Date: December 11, 2020, 4:32 pm
---------------------------------------------------------
I find it best to quickly check match history of players before
a game begins, focusing on Structures Razed and Units Killed.
If a terran has low Structures Razed and Units Killed, you can
go zerg and safely ignore them.
*****************************************************