Elimination Rating: Players who do (and don't) rise to the occasion

Farming low-stakes upper bracket group stage games is one thing, but which players perform best when the loser is sent home?

Pressure is hard to quantify. A lot goes into the psychological makeup of a player before a match.

One easy way to apply pressure is to stage a playoff match in a stadium with bright lights, pyrotechnics, noisy fans, and the trophy in sight.

But this is very rare for most players. Sample size was the buzzword when we looked at big game players last year. Only the very cream of the crop at the top of the scene have played enough maps in arenas to determine the impact of that pressure.

Read more

Who are the modern day big game Counter-Strike players?

So what if we look at earlier stages in tournaments? Games where the loser is eliminated are explicitly higher-pressure scenarios given there are actual stakes at play.

Plenty of matches (too many, in the case of ESL Pro League and BLAST's seasonal groups) have too low a deterrence against losing. Most teams can drop a series or even two and still progress.

In low-stakes upper-bracket games, players are free to take risks they might not otherwise, and upsets are more likely.

But once you're in the elimination bracket, gloves are off. It is not a new concept, either. Stats in the NBA are frequently filtered by elimination matches to try and identify which players perform and, just as importantly, those who don't.

We'll start with a data dump with no context: the top ten highest-rated players in CS2's elimination matches with a minimum map count above 7. To keep it more familiar, we've excluded players who have not made a single arena match.

The top five are all stalwarts. Danil "⁠donk⁠" Kryshkovets is way ahead but has only played 8 elimination maps in CS2 so far, such is Spirit's dominance.

Then we have Ilya "⁠m0NESY⁠" Osipov and Mathieu "⁠ZywOo⁠" Herbaut to complete what is so far proving to be 2024's top trio.

Jonathan "⁠EliGE⁠" Jablonowski is a standout in fourth just below the three immortals at a whopping 1.27 rating, before a big gap to Helvijs "⁠broky⁠" Saukants' 1.21 in fifth.

The big shock is Ricky "⁠floppy⁠" Kemery — our eyebrows were raised, too. Upon a deeper look, it is more than just his incredible IEM Sydney playoff performance propping him this high up: He has performed far, far better in elimination games than his overall rating during this period.

He's the most improved, in fact, by far. Some of this is a natural consequence of lower brackets having, well, lower-tier teams. Elimination rating is not like playoff rating in this sense as opposition might be weaker under this system.

However, the average elimination game rating is lower than a player's overall rating, with a small but relevant delta of -0.02. This implies that most players find it harder in these higher-pressure situations.

EliGE and m0NESY re-appear from the top ten, and usual big game player Håvard "⁠rain⁠" Nygaard also crops up with a small improvement in elimination games. But, overall, these are quite small differences. More interest can be found in the list of players dropping off.

Dmitry "⁠sh1ro⁠" Sokolov farmed the group stage at IEM Katowice but was less impactful in playoffs, and also had a nightmare at IEM Sydney with Cloud9 and in Spirit's Major quarter-final defeat at the hands of FaZe.

With 5 of his 13 maps coming at Sydney (with a dead Cloud9), and his sky-high overall rating, his figure of -0.18 does paint sh1ro in a bad light and it will be interesting to see if he can undo the damage as his sample size grows to a more respectable rate.

Astralis and MOUZ are two teams that have reputations for showing strong levels in groups and dropping off later, and that means most of their elimination games come in playoffs.

Even still, these figures represent significant drops for Jimi "⁠Jimpphat⁠" Salo, Martin "⁠stavn⁠" Lund, and Jakob "⁠jabbi⁠" Nygaard. The same is true for Evgenii "⁠FL1T⁠" Lebedev, Ismailcan "⁠XANTARES⁠" Dörtkardeş, and Ali "⁠Wicadia⁠" Haydar Yalçın.

Vitality's Lotan "⁠Spinx⁠" Giladi rounds out the top ten and should raise eyebrows himself. He can match ZywOo's impact at times, but when times get tough Vitality increasingly look like the one-man show Spinx seemed to have remedied in 2023.

Ultimately, these examples are mostly outliers; most players hover quite close to their overall rating. Here's a chart that only highlights players who stray away from the 1:1 ratio in terms of rating across elimination games and overall.

donk, of course, breaks the scale in the top-right, but a lot of these outliers are lower-sample cases — and you can see just how many unlabelled dots cluster the 1:1 ratio line.

We must also bear in mind that top teams play the vast majority of their elimination maps in arenas. ESL Pro League, played in a studio environment, lowered the percentages for most teams except Spirit, who didn't attend, but you can see which teams are facing elimination later on in tournaments.

Players from the four top teams of CS2 (Spirit, MOUZ, FaZe, and Vitality) all reach upwards of 48%.

In contrast, FlyQuest and FURIA are yet to reach a CS2 arena. HEROIC only have two arena maps among their roster from when Guy "⁠NertZ⁠" Iluz was in ENCE. And, it is in these teams that elimination rating offers a sight into the players that do not even show up in arena rating comparisons.

As we said in the big game players article, sample size is the number one issue with looking at arena rating.

Olek "⁠hades⁠" Miskiewicz looks like a huge outlier here, but it is just his IEM Katowice quarter-final that is being considered here — one series. NertZ is an outlier in the opposite direction, due to one bad series against FaZe in Sydney.

Complexity players all fare well and have decent sample sizes (for big games, at least) of 10 maps each, but this is flawed, too; they only made the arena twice, a while ago at IEM Sydney and Fall Final.

When we flick the switch to a sample size of more than that figure of 10 maps, only 21 players remain: Vitality (14 arena maps), FaZe (40 (!)), G2 (14), MOUZ (13), and Russel "⁠Twistzz⁠" Van Dulken (15).

Here's the full list of players above that mark, and you can see the trend of MOUZ players dropping off while FaZe and Vitality remain strong. m0NESY is the outlier, outdoing a team where Nikola "⁠NiKo⁠" Kovač is second with 1.01.

Read more

xertioN: "I just want us to translate this into Katowice, Cologne, and Majors"

This is why Dorian "⁠xertioN⁠" Berman spoke of the need to translate MOUZ's studio form into arenas. A narrative that has been well established by now.

Going back to elimination games, we are left with many of the same conclusions that we had with arena rating.

Adding pressure, via arena or just the prospect of elimination, does make a game more important. Performing in more important games should be recognized and rewarded.

But, by and large, the same players crop up at the top of leaderboards in meaningless seeding matches and high-pressure RMR deciders.

Exceptions like floppy make this a worthwhile exercise, as does highlight those that do drop off like sh1ro, stavn, and Jimpphat where otherwise narrative does heavier lifting than hard facts.

Explore elimination game data in more detail with interactive charts and more here.