A look at inconsistency

Which players really are guilty of inconsistency, and which are subject to unfair narratives?

The nature of any sport is that consistency trumps all. A player's ceiling is rendered irrelevant if they cannot reach it when their team needs them best. The greatest players of all time have high ceilings, but it is their floor that truly makes them stand out. Most tier one players have dropped 30 bombs; far fewer have strung together 20+ maps in a row above a 1.00 rating.

And it is the players who have ceilings as high as the game's best but lack that floor that are so frustrating. Our eye tests remember their best games but not their average ones. Rather than being adorned with MVPs, these players are relegated to win conditions, labelled as hot-and-cold, as inconsistent players.

Chief among these players is Kristian "⁠k0nfig⁠" Wienecke. The Astralis entry fragger has dazzled as much as he has disappointed. That he held the CS:GO record for kills in regulation, going 47-21 against Renegades in 2018, shows the ceiling he can reach. In this iteration of Astralis, it is he and Benjamin "⁠blameF⁠" Bremer that are supposed to provide the starpower, yet k0nfig has fallen short of those expectations. His talent has been evident, as it always is, but consistency has eluded him.

So the narrative goes, anyway. But is it a fair one? Can we, fairly and objectively, measure a player's inconsistency? Let's find out, using stats from LAN in 2022.

To kick things off, we'll start with a basic measure of map-to-map consistency: What percentage of maps a player finishes above a 1.00 rating. The average for our sample is just over 55%, so all of these players are above and beyond the pack.

However, this is far from perfect. A 1.01 rating might be a successful match for an average player, but for a star player or primary AWPer that might actually harm their team's chances to win the match.

To see how much a player fluctuates compared to their own average rating, there's a bit of mathematical jargon. We'll be using standard deviation, which is a measure of variance. Basically, the lower the standard deviation, the more a player's maps cluster around their mean rating. A high standard deviation, therefore, should indicate a level of inconsistency.

k0nfig, interestingly, doesn't make the list of the top eight players with the highest standard deviation, placing in equal 17th at 0.327 standard deviation. The players that do make the list, however, are of his ilk. Vladislav "⁠nafany⁠" Gorshkov takes second at 0.377, another player who has come under fire for his inconsistency.

The mean rating on the table is different from the overall rating you'll see on player profiles because it is an average rating per map, rather than per round

His high standard deviation shows why Cloud9 were so deadly at IEM Dallas, when nafany was in red-hot form, averaging a 1.14 rating (17 points higher than his year-long average).

Even within that event though, nafany's highs and lows were staggeringly different. Maps like 2.03 ratings against ENCE in the final and 2.16 in groups against Ninjas in Pyjamas were still coupled with 0.84 on map one against BIG and 0.79 and 0.91 ratings in two of the three maps against FaZe.

The story is the same for many of the players on this list. Fredrik "⁠REZ⁠" Sterner is another player to be labeled as inconsistent, and this metric seems to bear that out. Yuri "⁠yuurih⁠" Santos has endured tough spells in 2022 for their standards, while Andrei "⁠arT⁠" Piovezan is even more aggressive than nafany.

That Nikola "⁠NiKo⁠" Kovač has such a high rating and high standard deviation is also very illustrative of his year; his peaks have been as good as ever, but there have been times when G2 could have used more from their star man.

In NiKo's case, we can see a slight flaw in using standard deviation. If a player's mean map-by-map rating is 1.26, is it really fair to call them inconsistent? Take this scatterplot, which compares a player's mean rating with their standard deviation.

Joining NiKo in the high variance and high rating corner of the map is Dmitry "⁠sh1ro⁠" Sokolov, Mathieu "⁠ZywOo⁠" Herbaut, and Oleksandr "⁠s1mple⁠" Kostyliev. This rings alarm bells if we want to use this metric to establish inconsistency. For these players, they 'deviate' from their mean because they post carry-maps so often. A 2.00 rating is as much as 0.80 off one of these players' mean ratings but is treated the same as a 0.60 rating for 'inconsistency' by standard deviation.

What is more interesting are the other sections. Riflers like Keith "⁠NAF⁠" Markovic, Russel "⁠Twistzz⁠" Van Dulken and Sergey "⁠Ax1Le⁠" Rykhtorov pair a low standard variation with high rating, showing that they are consistently in and around the 1.00-1.40 mark in their maps played. Audric "⁠JACKZ⁠" Jug is the player with the lowest standard deviation at just 0.24 as well as a fairly low mean of 0.96; he was consistently below average statistically this year.

This comparison also works for players with high variance and low ratings, with Alejandro "⁠mopoz⁠" Fernández-Quejo Cano, arT, Dan "⁠apEX⁠" Madesclaire and nafany slotting in near where we expected: Ultra-aggressive, X-factor, rifles. Yet, we will need a different solution for the very best players: The presence of s1mple and ZywOo in the top right proves that standard deviation is not good enough on its own to label a player as inconsistent.

One solution is to rank players by their floor, their bad maps, rather than their good ones. To do this we will be using the first quartile, or 25th percentile, of a player's maps this year on LAN. The 25th percentile is a cousin of the median average except where the median is the middle value of an ordered list, the 25th percentile (often called Q1 in stats) is a quarter of the way through. For ease, when we use the term "floor" from here out we are referring to a player's 25th percentile.

Now, we see those names that standard deviation treated unfairly in a different light. The floor for s1mple is a 1.08 rating, for ZywOo 1.06, and NiKo 1.01. Lurkers Ax1Le, NAF, and blameF are also in the top ten, perhaps in part due to their role allowing them to frag in their team's losses as well as their wins.

This shows us who the most consistent players are, but what about inconsistent? If we just look at the players with the lowest floor, we get the likes of Epitacio "⁠TACO⁠" de Melo (0.64), Richard "⁠shox⁠" Papillon (0.67) and Rasmus "⁠HooXi⁠" Nielsen (0.69). Yet, this does not denote inconsistency in itself, since all these players had pretty bad average ratings.

To find the inconsistent players, we're going to need our high school math textbooks again. By subtracting the floor (25th percentile) from the ceiling (75th percentile: The same as before, going three-quarters the way through of an ordered list) we get something called an inter-quartile range (IQR). This, like standard deviation, is a way to measure variance — think of it as the difference between a player's good and bad maps — and should be more useful for our purposes.

Here's a graphical explanation of the same stat. Each bar is one of k0nfig's maps on LAN in 2022, arranged from lowest to highest. Q1 is one-quarter of the way through, Q2 two quarters, and Q3 three quarters. Then we subtract Q1 from Q3 to give us the IQR.

With that explained, here are the players with the highest IQR:

NiKo is again featured, thanks in part again to his ridiculously high ceiling, something sh1ro also suffers from. Valeriy "⁠b1t⁠" Vakhovskiy and Lotan "⁠Spinx⁠" Giladi had a very good 2022, sharing a median rating of 1.15 but find themselves pretty high on this list. Spinx actually had a pretty good 0.98 rating as his floor; his high IQR is a result of his 1.45 rating ceiling, a huge 0.30 higher than his median rating. This puts him in the same camp as NiKo (1.50 ceiling), a comparison that has been made before and for good reason.

IQR is better than standard deviation, but we are still viewing numbers without their full context. To solve that, here's a scatterplot that visualises a player's floors at the same time as their ceiling. The size of each player's dot corresponds to their inter-quartile range, which should help picture this further. Naturally, there is a lot of correlation between floor and ceiling but there are still points of interest in the players that stand out from the trend line.

The players in the orange and green bubbles have higher ceilings than they do floors, while those in the red and yellow have the vice versa. Now, players like NiKo and sh1ro are rewarded for their high ceilings, being placed in the green bubble of consistently good players. The very top right of the chart also illustrates the differences between Ax1Le and NAF, two players who had very similar standard deviation and IQR, with the Cloud9 man much further up and right than NAF.

Further left, we have the main 'inconsistent' orange bubble — the players who have high ceilings but fairly low floors. The majority of these players are fittingly orange dots, showing that they are aggressive riflers (with more than 20% opening kill attempts on T-side) which makes a lot of sense. These players, on a good day, can farm rating with high-impact entry frags and multi-kills. On their bad days, however, their survival rate tumbles, leaving them in the red.

Boris "⁠magixx⁠" Vorobiev is the biggest outlier here, somewhat surprisingly. On his good days he is rated as highly as b1t and Mareks "⁠YEKINDAR⁠" Gaļinskis but he has a floor lower than Andreas "⁠Xyp9x⁠" Højsleth. Joining him are more of who you would expect: nafany, Hampus "⁠hampus⁠" Poser, Fredrik "⁠roeJ⁠" Jørgensen, and Michael "⁠Grim⁠" Wince are all aggressive X-factors rather than consistent forces. Asger "⁠Farlig⁠" Jensen is the AWPer with the lowest floor in our sample, which fits with the narrative around the Dane.

Here, we have presented three different ways to look at inconsistency: standard deviation, inter-quartile range, and the 'orange zone' in our scatterplot (those players with a low rating in bad maps but a strong one on their good days). All have their flaws when used in isolation, so let's now combine the different methods for an 'inconsistency rating' formula.

To recap, we're taking into account:

— Percentage of maps with 1.00+ rating
— Standard deviation
— Inter-quartile range (Q3-Q1)
— Difference between a player's average and floor (Q2-Q1)
— Difference between a player's average and ceiling (Q3-Q2)

Here's a list of players with the highest inconsistency rating, to give the most 'inconsistent' players. Remember, though, that this is consistency compared to a player's average rating; these players are consistently around their average, not consistently good. Only about 20% of the formula is related to being a 'good' player statistically, in that players are awarded inconsistency rating for having a low percentage of maps above 1.00.

While k0nfig is absent — and b1t has rarely been accused of inconsistency on desks — the list as a whole seems to match with the eye test and community narratives. nafany, mopoz and apEX are all archetypes of the explosive yet inconsistent players that have cropped up throughout this piece, the very same archetype k0nfig is part of.

So, have we solved the great 'inconsistency' question? Sort of — but there are still holes. And, as we said in the introduction, proper consistency will elude 99% of professional players. The narrative around k0nfig and REZ being inconsistent is probably based on the idea that these players should be consistent, given their obvious talent and mechanical skill on the eye test.

When we look at a bigger sample, however, we can see that the vast majority of riflers suffer from the same problem. NiKo had three months at the back end of 2021 where he was as good as anyone in the world — even the AWPers. Now, though, he has fallen back into 'just' being the best rifler in the world. When we flip the list to give us the least inconsistent players, it is a collection of AWPers and supportive, more passive, riflers.

In a game as difficult as CS:GO, off-days and bad patches are inevitable. But, it is clear that some players have better off-days than others. And, as we said in the introduction, this only makes the players who can put up 1.00+ ratings on bad days more valuable, especially those who do take a lot of opening duels like Ax1Le and NiKo.

The problem is that those two players are the only aggressive riflers who have put up a floor higher than 1.00 on LAN so far this year. Just eight others managed that feat — five of them being primary AWPers — which is a tiny fraction of the professional playerbase. True consistency at a high level is every sport's El Dorado, and Counter-Strike is no different.


For similar deep-dive articles, check out the links below:

Read more
Is the modern AWPer really too passive?
Read more
When do Counter-Strike players peak?
Read more
Expert take: Age and motivation in Counter-Strike
Read more
Why are modern IGLs so aggressive?
Read more
Is double AWPing worth it?
Read more
Who are CS:GO's map specialists?
Read more
What are CS:GO's easiest and hardest CT positions?
Read more
What are CS:GO's easiest and hardest T-sided positions?
Read more
We paired current and former academy players with their top tier doppelgängers
Read more
Should rosters be given more time?
Brazil Epitacio 'TACO' de Melo
Epitacio 'TACO' de Melo
Age:
27
Team:
Rating 1.0:
0.92
Maps played:
1427
KPR:
0.61
DPR:
0.64
Canada Keith 'NAF' Markovic
Keith 'NAF' Markovic
Age:
25
Team:
Rating 1.0:
1.09
Maps played:
1594
KPR:
0.73
DPR:
0.63
France Richard 'shox' Papillon
Richard 'shox' Papillon
Age:
30
Team:
Rating 1.0:
1.07
Maps played:
2053
KPR:
0.73
DPR:
0.66
Brazil Yuri 'yuurih' Santos
Yuri 'yuurih' Santos
Age:
22
Team:
Rating 1.0:
1.12
Maps played:
1034
KPR:
0.75
DPR:
0.64
United States Michael 'Grim' Wince
Michael 'Grim' Wince
Age:
22
Team:
Rating 1.0:
1.05
Maps played:
915
KPR:
0.72
DPR:
0.69
Russia Boris 'magixx' Vorobiev
Boris 'magixx' Vorobiev
Age:
19
Team:
Rating 1.0:
0.99
Maps played:
649
KPR:
0.65
DPR:
0.63
Denmark Fredrik 'roeJ' Jørgensen
Fredrik 'roeJ' Jørgensen
Age:
29
Team:
Rating 1.0:
1.06
Maps played:
926
KPR:
0.74
DPR:
0.68
Russia Dmitry 'sh1ro' Sokolov
Dmitry 'sh1ro' Sokolov
Age:
21
Team:
Rating 1.0:
1.23
Maps played:
894
KPR:
0.77
DPR:
0.53
Russia Vladislav 'nafany' Gorshkov
Vladislav 'nafany' Gorshkov
Age:
21
Team:
Rating 1.0:
0.96
Maps played:
895
KPR:
0.66
DPR:
0.69
Denmark Andreas 'Xyp9x' Højsleth
Andreas 'Xyp9x' Højsleth
Age:
27
Team:
Rating 1.0:
1.02
Maps played:
2015
KPR:
0.66
DPR:
0.61
France Dan 'apEX' Madesclaire
Dan 'apEX' Madesclaire
Age:
29
Team:
Rating 1.0:
1.01
Maps played:
2026
KPR:
0.71
DPR:
0.70
Canada Russel 'Twistzz' Van Dulken
Russel 'Twistzz' Van Dulken
Age:
23
Team:
Rating 1.0:
1.11
Maps played:
1415
KPR:
0.73
DPR:
0.62
Latvia Mareks 'YEKINDAR' Gaļinskis
Mareks 'YEKINDAR' Gaļinskis
Age:
23
Team:
Rating 1.0:
1.07
Maps played:
1123
KPR:
0.75
DPR:
0.70
France Audric 'JACKZ' Jug
Audric 'JACKZ' Jug
Age:
30
Team:
Rating 1.0:
0.98
Maps played:
1126
KPR:
0.67
DPR:
0.67
Bosnia and Herzegovina Nikola 'NiKo' Kovač
Nikola 'NiKo' Kovač
Age:
25
Team:
Rating 1.0:
1.16
Maps played:
1657
KPR:
0.80
DPR:
0.67
Sweden Hampus 'hampus' Poser
Hampus 'hampus' Poser
Age:
24
Rating 1.0:
1.01
Maps played:
1389
KPR:
0.70
DPR:
0.70
Spain Alejandro 'mopoz' Fernández-Quejo Cano
Alejandro 'mopoz' Fernández-Quejo Cano
Age:
26
Rating 1.0:
1.00
Maps played:
1174
KPR:
0.68
DPR:
0.68
Brazil Andrei 'arT' Piovezan
Andrei 'arT' Piovezan
Age:
26
Team:
Rating 1.0:
1.00
Maps played:
1020
KPR:
0.71
DPR:
0.71
Ukraine Valeriy 'b1t' Vakhovskiy
Valeriy 'b1t' Vakhovskiy
Age:
19
Rating 1.0:
1.06
Maps played:
478
KPR:
0.70
DPR:
0.63
Russia Sergey 'Ax1Le' Rykhtorov
Sergey 'Ax1Le' Rykhtorov
Age:
20
Team:
Rating 1.0:
1.10
Maps played:
967
KPR:
0.73
DPR:
0.63
Israel Lotan 'Spinx' Giladi
Lotan 'Spinx' Giladi
Age:
22
Team:
Rating 1.0:
1.09
Maps played:
465
KPR:
0.74
DPR:
0.65
Denmark Rasmus 'HooXi' Nielsen
Rasmus 'HooXi' Nielsen
Age:
27
Team:
Rating 1.0:
0.86
Maps played:
818
KPR:
0.58
DPR:
0.69
Denmark Asger 'Farlig' Jensen
Asger 'Farlig' Jensen
Age:
23
Team:
Rating 1.0:
1.08
Maps played:
762
KPR:
0.72
DPR:
0.63
France Mathieu 'ZywOo' Herbaut
Mathieu 'ZywOo' Herbaut
Age:
22
Team:
Rating 1.0:
1.27
Maps played:
1043
KPR:
0.84
DPR:
0.61
Denmark Kristian 'k0nfig' Wienecke
Kristian 'k0nfig' Wienecke
Age:
25
Team:
Rating 1.0:
1.08
Maps played:
1593
KPR:
0.75
DPR:
0.69
Denmark Benjamin 'blameF' Bremer
Benjamin 'blameF' Bremer
Age:
25
Team:
Rating 1.0:
1.15
Maps played:
915
KPR:
0.76
DPR:
0.61
Ukraine Oleksandr 's1mple' Kostyliev
Oleksandr 's1mple' Kostyliev
Age:
25
Rating 1.0:
1.25
Maps played:
1589
KPR:
0.85
DPR:
0.64
Sweden Fredrik 'REZ' Sterner
Fredrik 'REZ' Sterner
Age:
24
Rating 1.0:
1.03
Maps played:
1378
KPR:
0.70
DPR:
0.67
lets be real REZ is king of inconsistency, him and draken
2022-09-26 14:37
22 replies
what about Pefecto?
2022-09-26 14:40
12 replies
#23
 | 
Finland iamjoe
perfecto consistently average
2022-09-26 14:45
#25
ele | 
Finland krep
Seems to be 8th most consistent
2022-09-26 14:46
5 replies
didn't you notice a mistake in his nickname?
2022-09-26 14:49
4 replies
#42
ele | 
Finland krep
I didn't, was it made on purpose?
2022-09-26 14:52
2 replies
look at last picture of least inconsistent, it says pefecto, just a "defecto" meme from twitch came to my mind
2022-09-26 14:53
1 reply
#47
ele | 
Finland krep
Alright
2022-09-26 14:54
#190
 | 
Sweden O7D
his inconsistency is perfecto
2022-09-27 05:32
What you, Sir, have is protagonist syndrome
2022-09-27 11:58
1 reply
sorry about that
2022-09-27 15:38
#212
Xyp9x | 
India NinX
For his role and positioning he is okay.
2022-09-28 05:59
2 replies
I know, there was a mistake in Perfecto nickname that's why I put an attention exactly on "Pefecto"
2022-09-28 15:11
1 reply
#214
Xyp9x | 
India NinX
Ah got it.
2022-09-28 15:20
Now you're just being fake, not real. It clearly shows that magixx is the king of inconsistency. But considering you answered in one minute after the article was released i guess you didn't even bother to look it up.
2022-09-26 14:48
5 replies
clearly you are taking my comment to seriously, brain is too thick to understand.
2022-09-26 14:51
4 replies
#52
 | 
Sweden elyeet
he's sweden, what expect?
2022-09-26 15:00
2 replies
+1
2022-09-26 17:48
Friendly fire!
2022-09-26 19:04
standard cowardice backpedaling xDDD
2022-09-26 19:59
Rez and k0nfig
2022-09-26 15:08
I disagre with electronic, he is a beast playing dk whos the genius that put him as IGL
2022-09-26 18:49
Presented with facts... response: "yeah but let's be real" and describes something else not based
2022-10-04 20:43
nice bro
2022-09-26 14:37
2 replies
syrson not inconsistent????? rly? XD and b1t inconsistent🤣🤣 i think some errors were made in this articles mens)s)s)s
2022-09-26 15:45
1 reply
PS k0nfig very consistent with hand/arm/wrist surgeries... author plz add this xd
2022-09-26 15:49
#3
 | 
United States cr4zy_m3
Nice
2022-09-26 14:37
1 reply
now it's official: xyp is consistently bad
2022-09-26 17:43
s1mple >>>>>> ALL PROS BTW
2022-09-26 14:39
1 reply
#221
 | 
Scotland 4ngu5
actually this shows that "statistically speaking" he's not. we all know he can be a game changer and a match winner with that x factor he brings but, this doesn't actually paint s1mple to be the untouchable god that people sometimes think he is.
2022-10-04 05:49
#5
 | 
Poland DennisGabZ
Nice, Bymas was consitently shit
2022-09-26 14:39
1 reply
#36
 | 
Lithuania CEOofwomen
:(( mens
2022-09-26 14:50
zywoo (group stage) vs zywoo (playoff)
2022-09-26 14:40
3 replies
true, he is great in group stage and a beast in playoff
2022-09-26 14:42
2 replies
+1
2022-09-26 16:57
+1
2022-09-26 17:40
pefecto
2022-09-26 14:40
low variance, poor rating = ALEKSIB IN A NUTSHELL
2022-09-26 14:40
Cool
2022-09-26 14:41
#11
 | 
United States brxcedw
so what you're saying is that degster is the best player? love to see it
2022-09-26 14:41
Interesting
2022-09-26 14:41
#13
 | 
Spain batxe
The last list features also players who are consistently bad, right?
2022-09-26 14:41
1 reply
#204
 | 
Yugoslavia Yungletti
Do you not know how to read?
2022-09-28 00:03
did not expect magixx to top that at all
2022-09-26 14:41
R eZ af
2022-09-26 14:42
Thought REZ would be higher on that list ngl
2022-09-26 14:43
3 replies
wasnt many more places higher he could be tbh
2022-09-26 15:07
+1 It's either 30-5 or 5-30 literally nothing inbetween with Rez
2022-09-26 15:19
1 reply
this
2022-09-26 22:12
#19
 | 
North America ares4458
Any NAFFERS?
2022-09-26 14:44
2 replies
#87
NAF | 
Canada Chordal
SeriousSloth
2022-09-26 15:34
1 reply
#149
 | 
North America ares4458
NA THIS EU THAT IM JUST A NAFFER NAFFING IN CHAT
2022-09-26 18:10
Rez and k0nfig are two riflers with superbly high ceilings and yet always undeliver especially in big stages
2022-09-26 14:44
1 reply
Hmm, makes sense. A nice supplimentary analysis would be comparing their results on group stages versus playoffs. KSCERATO, for example, if he delivered on playoffs he would easily be among the top10 riflers in the world.
2022-09-26 15:22
Sh1ro and ax1le gods but hobbit is washed 😢
2022-09-26 14:44
1 reply
Nah not washed. Lower form than the past but he still has bursts of firepower. Top frags some odd maps as well. Shiro and Axile are supposed to be the star riflers anyway
2022-09-27 08:34
#22
 | 
CIS TheBigSAM
Abdul po defoltu ebet EZ
2022-09-26 14:44
makes me laugh that Syrson is in least inconsistency list even after not having one game of life in this year on Lan kekW...
2022-09-26 14:45
5 replies
He is a fraud
2022-09-26 14:46
2 replies
+1 idk why tabsen and gob b still giving him more chances and keeps rotating other players rather than try to move him on bench and try someone more confident player, i do think he is the reason why most of the rifler's are not getting space or the impact in the team and this BIG is just not seeing it as the way they should now.
2022-09-26 14:49
1 reply
There is no reason for syrson to be trash on lan we have young players like sh1ro ax1le b1t monesy who are gods syrson excuses doesnt exist
2022-09-26 14:51
#153
 | 
Germany xSchons
stats dont lie, he is pretty consistent with a high floor, though he cant reach the ceiling other star awpers have
2022-09-26 18:51
1 reply
#162
 | 
United Kingdom novaseer
+1, this article isn't meant to objectively say that a consistent player is good or an inconsistent player is bad, it's meant to say which players have lots/not a lot of statistic variance from game to game, whatever their floor or ceiling may be
2022-09-26 19:23
#26
ele | 
Finland krep
Very cool stats
2022-09-26 14:46
Great article. 👍
2022-09-26 14:46
#29
hades | 
North America minte
How does this affect Lebron's legacy?
2022-09-26 14:47
1 reply
#165
 | 
Uzbekistan R_Sick
By invalidating it
2022-09-26 19:50
#30
 | 
Lithuania CEOofwomen
no hooxi?
2022-09-26 14:47
2 replies
Hooxi is there, read the article
2022-09-26 15:16
1 reply
not enough
2022-09-26 16:06
great article tbf
2022-09-26 14:47
nafany on top lessss go
2022-09-26 14:51
3 replies
Goat igl after glaive and karrigan for sure
2022-09-26 14:51
2 replies
glaive>karrigan>nitro>nafany
2022-09-26 14:58
name checks out
2022-09-26 15:14
#44
 | 
Scotland Ludax
holy shit magixx consistently inconsistent
2022-09-26 14:52
nice hltv!!!
2022-09-26 14:55
Being inconsistent is also in itself a consistency because you're consistently inconsistent thus you are actually consistent (but not really)
2022-09-26 14:57
1 reply
Mindblowing
2022-09-26 19:06
#50
F1KU | 
Finland Dunu
Common Ner0 W
2022-09-26 14:57
#53
 | 
France ignorefAN
Pefecto such a god
2022-09-26 15:01
#54
 | 
United States 1Tuurtle
" Audric "⁠JACKZ⁠" Jug is the player with the lowest standard deviation at just 0.24 as well as a fairly low mean of 0.96; he was consistently below average statistically this year." You didnt have to do him like that HLTV
2022-09-26 15:02
6 replies
LOL at least he’s consistently average
2022-09-26 15:24
4 replies
#108
 | 
United States 1Tuurtle
They said he's consistently below average man...
2022-09-26 16:14
3 replies
.96 is barely below if avg is 1 no?
2022-09-26 16:15
2 replies
#110
 | 
United States 1Tuurtle
idk, just saying what the article said. They said he's below average consistently
2022-09-26 16:16
I don't think 1.0 is the average rating amongst players on top teams on HLTV - I would guess more like 1.05 or more
2022-09-27 09:27
hahahaha
2022-09-27 09:26
@Hltv Rating 3.0 when??
2022-09-26 15:03
1 reply
#90
nqz | 
Canada qui9
+1 this is a really cool article and i would love to see a "consistency rating" in the stats for example i think k0nfig being inconsistent is kind of a myth as shown by the article, we just like to remember his crazy games, but in reality he's kinda just washed and mid nowadays. makes me wonder if maybe he was more inconsistent in the past, as an aggressive rifler like nafany or mopoz.
2022-09-26 15:37
#56
 | 
Denmark Marvelw0w
Bit most overrated player to exist
2022-09-26 15:07
I'm a simple man. I see NER0cs, I click.
2022-09-26 15:07
only look at xyp, he should move to flameZ spot in 'rating v variance' graph and farlig clearly the worst tier1 AWPer
2022-09-26 15:07
#61
 | 
Denmark Flaeskefar
2022-09-26 15:12
#62
 | 
Italy MulaManca
Ok so... TACO is consistently the worst rated pro lmao
2022-09-26 15:12
1 reply
#148
 | 
United States charybids
TACO is bad, and consistently bad
2022-09-26 17:51
wow props to degster, NAF, and Twistzz. Consistency is everything in this game, gives a solid anchor for the team to rely on.
2022-09-26 15:21
#65
 | 
Argentina cheapdeed
whoever wrote this article is a real hater... i respect that
2022-09-26 15:14
2 replies
#206
 | 
Yugoslavia Yungletti
The only Farlig fan in the world up here ^
2022-09-28 00:14
1 reply
#211
 | 
Argentina cheapdeed
the cockwatching is crazy.... farlig is a beast just wait💯​
2022-09-28 01:40
degster Mr Consistent - expected from the GOAT
2022-09-26 15:14
lmao so niko is the 3-5th most inconsistent player. Basically confirming what everyone was thinking but his fan boys deny
2022-09-26 15:15
16 replies
He has a high floor too so you cant exactly say he is "inconsistent"
2022-09-26 15:24
11 replies
Isn’t that the definition of inconsistent? Having really good games then really bad games. It’s better to have a teammate that’s consistent with a slightly lower peak than a teammate with high peaks then shits the bed. Good comparison of this would be Twistzz and Niko. Niko has a higher peak but because of Twistzz consistency he’s a better teammate to have and probably a better player overall.
2022-09-26 15:32
10 replies
I just said he has a high floor, which literally means he does NOT have too bad games.
2022-09-26 15:36
1 reply
But he does, anyone who follows CS closely knows niko shits the bed more often than most top 20 players. He’s known for being inconsistent and choking regardless of this article
2022-09-26 15:38
#93
NAF | 
Canada Chordal
You didn't read the article. Niko is only shown as "inconsistent" because his highs are so ridiculous. Twistzz has not had pop-off games like him, thus lesser deviation from him, though their floors are not that far off. You can look at the final graph and see that NiKo is not there, because he still consistently posts ratings above 1.00
2022-09-26 15:38
1 reply
Ok that’s basically what I said aside from the fact that he’s a known choker. Everyone knows he’s inconsistent…
2022-09-26 15:47
What the chart explained it's that NiKo plays his f*cking ass for his team making space, doing entrys, clutching, multikilling, and that gives him a high 1.50 ceiling, but, when he can't, he has at least a 1.00 floor. Meanwhile Twistzz supports his team, having good overall ratings, his ceiling is 1.20 max, but with a -1.00 rating.
2022-09-26 16:16
2 replies
And one of them chokes every grand final and the other doesn’t. That’s all that matters at the end of the day. No one cares about 2nd place
2022-09-26 16:49
1 reply
Sure bro. I got baited by you in such a stupid fashion. GJ
2022-09-26 19:35
You'd rather have a passive rifler who is more dependent on the team's performance over an agro play maker (try to internalize the words 'play maker' here) with the highest peaks a rifler has ever had because the former is a little more consistent over the other while they both have the same floor rating. Niko can be just as consistent if you put him in twistzz's roles, not true for the inverse. Can't believe that has to be said
2022-09-26 17:31
2 replies
Serbian flag, tldr bye
2022-09-26 17:37
1 reply
hilarious that low effort baiters can actually make real impact into sentiments about players like we saw with impressionable minds like Pimp's, guess that's why some people do it inb4 you reply with an uninspired one liner how i'm the one baiting
2022-09-26 17:48
Yet again ruining Americans reputation. Negative IQ petard
2022-09-26 15:29
2 replies
Found the delusional niko fan
2022-09-26 15:29
1 reply
Not even a Choko fanboy Nt though cringefock
2022-09-26 22:35
Inconsistent because in some games he goes 1.7, and in others, he goes 1.2
2022-09-27 04:20
Stats are not very useful in CS, where you can win a round just by surviving and lots of other scenarios which stats don't take into account
2022-09-26 15:19
12 replies
Nah that’s massive cope, they analyze so many different data points to come to these conclusions. Also they make sense
2022-09-26 15:19
7 replies
I just stated facts but whatever, i dont care if u agree
2022-09-26 15:21
Lol
2022-09-26 15:28
And all data points rely on rating :)
2022-09-26 15:49
4 replies
And rating is based off of many data points
2022-09-26 15:49
3 replies
so what? does that mean that they are an infallible metric or something? ofc not
2022-09-26 17:25
2 replies
#166
 | 
Uzbekistan R_Sick
And so? Should we just listen to a mopoz coper then?
2022-09-26 19:54
1 reply
I like mopoz but that has nothing to fo with this, in fact i dont care if he has bad ratings or loses every game. I just like to watch him play. Can't see how my flair is relevant in any way. Fact is stats don't tell the whole story. That's my only point. Next time maybe use brain
2022-09-26 21:03
I disagree. I truly believe we can have statistics that make sense, pretty much as regular sport have statistics that make sense. Are these ones those statistics? Probably not. But if nobody tries different stuff out, we can never get such statistics ever.
2022-09-26 17:20
3 replies
hope we get statistics that reflect players impact and performance way better that the ones we have now. but for now, we'll just have to watch demos and analyse with our eyes and brains to know how good a player is or what does he bring to the table imo
2022-09-26 17:24
2 replies
Can agree. As for now, we can put players in some vague brackets but not really say which individual player is truly better than another by statistic alone.
2022-09-26 17:27
1 reply
+1, also it's not like if statistics are completely useless, it's just that they don't show the whole picture 'cause CS:GO is a complex game to analyse.
2022-09-26 17:30
Dude, these articles are brilliant!
2022-09-26 15:21
#75
sh1ro | 
Europe pix3w
banger
2022-09-26 15:23
I don't rate the rating system for rating consistency, too many things can influence kills like positions, the info you get, maps and opponents played. eye test + things like HS% ADR DPR better for judging current form
2022-09-26 15:27
Wow, this was top-tier analysis. I'm very happy to see people like this involved on the CSGO scene. Orgs desperately need more analytical people in their staff, it would be a huge leap in terms of professionalism in decision-making.
2022-09-26 15:27
#82
maga | 
Denmark sla1ve
S1mple no. 1 rat
2022-09-26 15:29
2 replies
Jealous that danish CS couldn’t keep up with the meta
2022-09-26 15:34
1 reply
What meta
2022-09-27 00:29
common NERO W
2022-09-26 15:29
how the fuck is syrson consistent????
2022-09-26 15:37
5 replies
#95
NAF | 
Canada Chordal
He is consistently good online and consistently shit on lan
2022-09-26 15:40
3 replies
#168
 | 
Uzbekistan R_Sick
Weren't these stats taken from LAN performances? syrson just consistently doesn't hit highs so we came up with the narrative that he is inconsistent which he isn't
2022-09-26 19:56
2 replies
Which means he isn’t inconsistent, he’s just average ot mediocre as an awper
2022-09-27 10:48
1 reply
#200
 | 
Uzbekistan R_Sick
He isn't a mediocre AWPer, but there is just something that is preventing him from becoming a true elite, but he is really well above most of the awpers on the scene
2022-09-27 14:24
Because syrson never bottomfrags he is always average. But online he always pops off
2022-09-27 10:07
rez must be by far the most inconsistent player on the scene
2022-09-26 15:38
1 reply
#113
 | 
United States 1Tuurtle
Didnt notice Magixxx....
2022-09-26 16:18
Quality read, NER0! Any chance we get this as a stats feature on the website?
2022-09-26 15:39
i feel bad for all the players who were in the low variance, low diviation group :P
2022-09-26 15:42
After updates from Valve at the end of 2017, inconsistency is very high. They done something to this game that year, somehow it sabotaged upcoming or present riflers back then and put sniper players on peek 🙂 Already in 2018-19 showed-up plenty of awp players like Zywoo, broky, degster, sunpayus zorte, el1an, woxic, sh1ro and many more, but sabotaged riflers like GeT_RiGhT, f0rest, Happy, fer, gla1ve, flusha, olofm, coldzera etc. The ones who arent believing to this what im saying, just check demos from the past and will conclude what im talking about. Just make comparison from 2015-16 demo to 2018-19 and u will got right answer 👍 Its clearly that VALVE somehow destroyed their game in aspect of rifling.
2022-09-26 15:44
16 replies
Tf ive just read hahaahha
2022-09-26 16:03
2 replies
Don't act like clown, you can check details if im right or not 🙂 Im in this game since 2013, i know what im talking
2022-09-26 16:19
#131
 | 
China RADNIKEY
You heard it here first The only reason why Happy and gtr don’t play tier 1 any more is because Valve sabotaged them
2022-09-26 17:01
#114
 | 
United States 1Tuurtle
Because no new riflers have been good since 2017... He wait a minute! dude the riflers you mentioned just got washed, it happens. They cant adapt/improve and so they get worse than current players
2022-09-26 16:20
10 replies
Write me 6-7 consistent riflers from 2019 to today, and i will back-off instantly saying you are completely right 👉
2022-09-26 16:35
9 replies
#119
 | 
United States 1Tuurtle
Niko, Elige, Naf, Twistzz, Ropz, BlameF, Axile Want more?
2022-09-26 16:37
8 replies
Nah bruh, blamef & ax1le cant be consistent. others are fine 🙂 Dont ever try to compare blamef or ax1le with list i typed above. 1st. blamef is hell of baiter, 0 impact. 2nd ax1le is not delivering numbers or impact same as list above.
2022-09-26 16:53
7 replies
#128
 | 
United States 1Tuurtle
You asked for consistent riflers, BlameF is sort of the definition of consistency. Axile sort of I guess.. How about guys like Krimz and Kscerato?
2022-09-26 16:58
6 replies
#129
 | 
United States 1Tuurtle
Or Stavn or Magisk also work
2022-09-26 16:58
KRIMZ was, but not anymore because of reasons i mentioned. The way KSCERATO played @ ESL Pro League is KSCERATO used to be, if he is playing like that in future there is not doubt that FURIA will be top3 team. But i know thats impossible, so my answer is no. stavn or magisK not even close haha 🙂 If i need to mention consistent rifle players here i go : YEKINDAR, ropz, mir, EliGE, NAF, Spinx, FASHR, NiKo, XANTARES, b1t
2022-09-29 14:32
4 replies
#216
 | 
United States 1Tuurtle
I think Krimz is more consistent than Fashr. He made top 20 like twice since 2017 so he's still been consistently good. Fashr has been insane on nuke, and average on everything else. KSCERATO has been one of the most consistently good riflers since at least 2020. Stavn has been consistently good and carried Heroic for the past few years. Magisk has been a top 20 player in 2018 and 2019 and he's still consistently good. calling Yekindar and Fashr consistent but not the others I've mentioned is kinda weird imo. both are way more inconsistent than Krimz, Kscerato, BlameF, Stavn, Magisk Yekindar and Fashr are similar to Axile, only Axile is a better player on average
2022-09-29 14:37
3 replies
KSCERATO eliminated from ESL Pro League yesterday. stavn eliminated from ESL Pro League yesterday. Both of them were nowhere close compared to groupstage performances, thats why im talking or better said ; inconsistent
2022-09-29 15:12
2 replies
#218
 | 
United States 1Tuurtle
So one match is the best way to judge consistency? In that case even back in 2015-2016 players like Get_Right, Happy, F0rest were never consistent
2022-09-29 15:23
1 reply
Do you kidding me ? stavn fucked up everywhere ESL Pro League S15 ESL One Cologne BLAST Premier PGL Major Antwerp ESL Pro League S16 same as KSCERATO ESL Valencia IEM Cologne ROOBET Cup. IEM Dallas PGL Major Antwerp IEM Katowice Please dont compare GeT_RiGhT, Happy, f0rest or anyone else to this generation full with bunch of puggers
2022-09-29 15:45
Because new awpers arent bots with rifles
2022-09-26 16:25
AWP conspiracy you heard it here first
2022-09-26 18:53
#98
 | 
Belarus vrub0n
SLAPSTER IS THE BOSS
2022-09-26 15:45
A follow up to this great analysis could look into how any inconsistency is handled within the team. Does the team absorb the inconsistent level of one player?
2022-09-26 15:47
#105
 | 
Sweden Hasklon
great article, time to kick rez now
2022-09-26 15:54
Finally apEX top 10 of something
2022-09-26 16:24
#117
 | 
Finland bliplbob
GJ Love these articles that challenge some common narratives
2022-09-26 16:34
how is tiziaN one time an aggressive rifler and then a passive one (passive is right imo)
2022-09-26 16:38
1 reply
#126
MSL | 
United Kingdom NER0cs
He should be passive, yeah. Think on one of the charts his dot is hiding behind someone else and I missed it 😅 good catch
2022-09-26 16:55
Asger "⁠Farlig⁠" Jensen is the AWPer with the lowest floor in our sample, which fits with the narrative around the Dane. FARTLIG SHINES AS CONSISTENLY BAD PLAYER
2022-09-26 16:48
would be cool to also analyze the most "inconsistent" players of the past
2022-09-26 16:51
The biggest challenge to overcome :(
2022-09-26 16:51
literally me
2022-09-26 16:59
Great article, absolutely brilliant. Adn somewhat surprising. I extremely happy that finally Axile is getting the recognition he deserves. He is not just a "good rifler" he definitely is one of the best and most reliable, very close to Niko. Also, now that I think about it, couldn't this be used to find out what players really are onliners and who are not? Comparing consistency online to consistency lan.
2022-09-26 17:23
1 reply
Finally? Wasn't he in hltv top 5 players 2021 iirc?
2022-09-27 10:36
2 anime weebs on top of inconsistency and absolute chads are most consistant. think about it kids
2022-09-26 17:19
good article, now do one determining bait kills, exit kills, eco kills, vs entry and high impact kills, *laughs in anti-astralis*
2022-09-26 17:26
#139
 | 
Portugal 7sson
Just dont pick Nafany in Fantasy.
2022-09-26 17:27
Did nobody found the standard deviation of the players to be fucking massive? Like 0.3 SD means roughly that there is a 32% chance that the player 0.3 below or above his mean. If a dude has 1.2 mean rating, then there is 32% chance that he scores 0.9 or 1.5. This is fucking insane.
2022-09-26 17:43
3 replies
I mean a .3 SD and the mean of 1.2 means that 68% of the time they will have between .9 and 1.5. Yeah that does seem pretty wide even like that
2022-09-26 18:52
2 replies
Oh yeah, i wrote it incorrectly, its 32% to score <0.9 or >1.5 and 68% to score between .9 and 1.5. Its huge
2022-09-26 18:53
1 reply
It makes sense though. Many of these teams (especially the top ones) play wildly different levels of opponents throughout tournaments. If they had a large enough sample, they should have found a way to normalize by team ranking differential between the player's team and their opponent also.
2022-09-26 18:56
Absolutely love this approach. Would love to see more statistical analyses like this in the future. Great job. I am wondering though why JACKZ is considered a passive rifler.
2022-09-26 18:27
9 replies
Frozen, too. Weird
2022-09-26 18:49
8 replies
#177
MSL | 
United Kingdom NER0cs
Both of them have <20% opening kill attempts on T-side
2022-09-26 21:32
7 replies
Opening kills isn't the only measure for aggressiveness, its incomplete and sometimes flawed. Karrigan is a very aggressive player, because he is the first one to enter the site, the "bait", has a low % of opk attempts, however if you simply watch the games, he is the one going in first with rain, same thing for gla1ve.
2022-09-26 21:34
5 replies
#183
MSL | 
United Kingdom NER0cs
All three of the players you mentioned have high OpK attempts on T-Side
2022-09-27 00:14
4 replies
Bro i just told you that the OPK rating does not capture overall aggression and you mention the stat again? Syrson has as much opk attempts as ax1le on T-side yet you can clearly see who's more aggressive. But syrson's role is much more passive than ax1le's. Again, the stat does not capture the actual "entries" the one that go in first into bombsites. Hampus goes for a fuck ton of opening duels but he is almost always alone, while plopski is the one that often actually goes in first into the bombsites. Opening kills is just one side of the aggression, you also need to consider space takers.
2022-09-27 00:49
3 replies
#186
MSL | 
United Kingdom NER0cs
It's the best stat we have for it right now, by far. And over a large dataset like this do you have a better solution?
2022-09-27 02:30
#189
FASHR | 
France Ansi
+1 this is not the first time HLTV is trying to make a players intangible assets measurable. but you can never forget were talking about individuals that play customized games with specific tactics/roles etc that cant be measured in just opk's and rating. the same goes for example this article: hltv.org/news/34460/should-rosters-be-gi.. For years now i have seen people trying to tweak human performance measurements as if we're talking about robots that have a certain range of fixed capabilities that allows data to be processed in some sort of pivot table. I really appreciate it when HLTV users like you take the time to open the conversation on what really is presented in articles like this, and question whether the goal (measuring consistency) really is achieved. IMO we're talking about human input, there are dozens of unmeasurable influences imaginable, therefore it is inadvisable to make a stat out of this. with all respect for the effort, but this article should never have been published, i consider it incomplete, faulty and misleading.
2022-09-27 05:19
"Bro i just told you that the OPK rating does not capture overall aggression and you mention the stat again?" To be fair: You use certain players as examples to show that aggressiveness cannot be tied to OpK attempts. However, if precisely these players turn out to be aggressive in the sense of this interpretation, your counter-argument has become proof of the author's thesis/approach. So it's clear that he's referring to that statistic again... But of course: I think it is clear to all of us that the aggressiveness of a player cannot be measured by this indicator alone! But with such statistics it's all about identifying those parameters (or to create new ones using the available data) from among all the existing parameters that make the performance differences measurable in the best possible way. It can therefore only ever be an approximation with no claim to completeness. So I partly disagree with the previous poster. Yes, there are many non-measurable influencing factors on a player's performance, but especially in e-sports, where all in-game activities can be digitally recorded and therefore evaluated, statistics are a useful method to support(!) the evaluation of players and their performance - even more than in conventional sports - and a further development and discussion of this "research area" is therefore very helpful.
2022-09-28 00:04
It doesnt have to be black and white though. Theres a middle ground between being a hyperaggressive player and a passive one. Niko often goes for opening kills on g2, that doesnt however mean that all the other players on his team are either passive riflers or awpers. If you have a pack of two riflers always attacking together, with the same one going in first (so going for opening kills), it doesnt mean the second one is a passive rifler. Opening kill % only talks about the first kill of the round iirc. Just tells us who is the first to be aggressive.
2022-09-27 18:54
Interesting artivle. I'd like more of these.
2022-09-26 19:09
this is an excellent article
2022-09-26 19:12
poor bymas
2022-09-26 19:31
This is the best post in this websites history, I love it. The typography of the title is clever and gorgeous. Kudos
2022-09-26 19:54
#171
 | 
Germany at0myq
unluko 4 magixxbossu<3
2022-09-26 20:09
Whoever read this till the end is geekier than Raj from Big Bang Theory
2022-09-26 20:18
#173
 | 
Uzbekistan R_Sick
Nice one
2022-09-26 20:19
interesting
2022-09-26 20:42
another banger ner0 article
2022-09-26 21:26
Roj mine boy🙂why inconstant🙂
2022-09-26 22:22
good article htlv confirmed
2022-09-26 22:47
must be bogus. Syrson is very inconsistent, yet is top 10 least inconsistent. maybe put the calculator away and watch the games?
2022-09-27 03:32
Incredible article Nero - you're a gift to the community!
2022-09-27 09:31
banger for sure
2022-09-27 13:27
Whatever those numbers are, I don't understand any of it kekw
2022-09-27 14:46
#207
 | 
Yugoslavia Yungletti
Great analysis, NER0. Please do an analysis on the inconsistency of players when they play in group stage vs. playoffs.
2022-09-28 00:19
Choko strikes again, what a choking inconsistent bot!
2022-09-28 01:07
First of all: thank you for this article! I'd love to see more of this type of content, because as I mentioned above, I consider such analyses as very helpful for the further development and discussion of statistics in order to support(!) the evaluation of players and their performances. And the lively discussion here in the comments section seems to confirm the need... @NEROcs: Nevertheless, two requests to the author: 1. Please add the consideration of the coefficient of variation already submitted via Twitter here as well. From my point of view, it deserves a more detailed consideration/appreciation and would address or replace at least parts of your newly created (but unfortunately not described in more detail) formula for measuring inconsistency (e.g. taking into account high mean rating values). 2. Please explain the classification of the player types or playstyles you are using in your analysis. As you can see, this also gives plenty of reason for discussion... Especially for a common understanding - especially for a consistent use of these terms - it would be very helpful if these basics are clearly derived and not only mentioned in the comments when asked. Because, as you yourself point out very well, the role has a significant influence on the interpretation of the results and should therefore be precisely defined. (Of course you will never be able to please everyone with these definitions.) For example, is a passive rifler here synonymous with a lurker or can they be separated and if so, how should they or why did you limit your analysis to exactly these three types?
2022-09-28 01:21
2 replies
I aren't read that
2022-09-30 14:07
1 reply
Me neither. I assume it's written for the HLTV staff.
2022-10-04 06:26
This is pretty cool, Kudos
2022-09-28 01:25
The stats, use of different statistic methods (like standard deviation )and understanding of player metrics by hltv is truly amazing. if only our math teachers taught us this way:)
2022-10-04 06:49
Hmm some players that made the consistent list does not make sense. Maybe constantly bad also counts? And inconsistenly bad
2022-10-05 12:12
Login or register to add your comment to the discussion.
Now playing
Thumbnail for stream
United Kingdom
ESL TV B
1119 viewers
Top streams
All(24)
Casters(12)
Streamers(9)
Organizers(3)
United Kingdom
ESL TV B
(1119)
Argentina
forg1
(833)
United States
Cooper
(569)
Brazil
lucas
(519)
Brazil
kNgV-
(519)
Argentina
infocsarg
(431)
United States
FishTheHusky
(335)
Canada
jasonR
(325)
United Kingdom
ESL TV
(293)
United States
freakazoid
(251)
Brazil
BTSBrasilFPS
(245)
Other
IESF
(132)
Israel
shushan
(126)
Brazil
Jogando Junto
(122)
Argentina
nikoz
(121)
United States
Laski
(67)
Canada
Strife
(61)
United States
Limitless
(48)
Argentina
NikoM
(27)
Australia
viridian
(22)
Brazil
JokerBR
(22)
United States
tawaAlx
(19)
United States
Eros
(13)
Brazil
BTSBrasilCSGO
(1)