What are the rarest situations in CS:GO?
When we added Achievements to our match pages, we didn't want to spam you with badges that were too easy to get. We solved that problem by gathering data (who could have guessed it!), and we didn't want to keep that data to ourselves.
We will start out with the expected frequency of different types of kills, but keep reading to find out if you are one of the lucky few who have received our rarest achievement!
Killfeed Modifiers
We want to award Achievements for getting a lot of blind kills, smoke kills, wallbangs, and noscopes in a single game. But how many of these do you have to get before it can be considered "a lot"? To answer this, we pulled data on how often the average player gets various amounts of these kinds of kills.
Here is how you read the table below: Each row identifies which kind of killfeed modifier we were counting. Each column tells you the minimum total that a single player had to achieve in a given match. That means that each table entry tells you the expected frequency of a player reaching a certain total in their match.
For example, the first box in the table tells you that you would expect an average player to get at least 1 wallbang kill in 40.5% of their matches. Moving all the way to the right in that same row, you would expect to see a single player get 5 or more wallbang kills in 0.25% of their matches, or 1 in 400.
| Match Total | 1 | 2 | 3 | 4 | 5 |
|---|---|---|---|---|---|
| Wallbang Kills | 40.5% | 12.9% | 3.56% | 0.93% | 0.25% |
| Smoke Kills | 28.9% | 7.85% | 2.07% | 0.55% | 0.15% |
| Blind Kills | 7.96% | 0.61% | 0.051% | 0.005% | 0.001% |
| Noscope Kills | 6.21% | 0.98% | 0.22% | 0.065% | 0.024% |
It makes sense that wallbangs would be the most common type of kill modifier, as during many wallbang kills you can still see part of the enemy. Even though they are the most common, you would still expect an average player to get zero wallbang kills in the majority of their matches. Noscopes being the rarest makes sense as well, since you cannot possibly get noscope kills if you are not using a sniper.
The drop-off on noscopes was way less harsh than the drop-off in blind kills. This suggests that someone who goes for at least one noscope in a match will likely go for many more. Blind kills are more a pure stroke of luck for the average player, so the expected frequency of even getting 3 in a single match already drops down to 1 in 20000.
Hopefully, now you will appreciate these little icons' rarity just a little bit more when they pop up in your killfeed.
Weapon-Specific Kills
We pulled the same stats for some of the more unique weapons as well! The next table is the exact same statistic, but with the knife, Zeus, and pistols:
| Match Total | 1 | 2 | 3 | 4 | 5 |
|---|---|---|---|---|---|
| Knife Kills | 3.79% | 0.314% | 0.013% | 0.005% | 0.002% |
| Zeus Kills | 0.901% | 0.119% | 0.035% | 0.007% | 0.003% |
| Pistol Kills | 87.0% | 69.1% | 49.4% | 32.3% | 19.5% |
Overall, knife and Zeus kills are both quite rare. The Zeus shows the same behavior as the noscope stat in the previous table, where the Zeus expected frequency drops less steeply than knife kills do. This suggests that fewer people are using the Zeus than the knife, but the people that do use the Zeus are more committed to the pursuit of zapping their opponents.
You would only expect the average player to get three knife kills in a single game with a frequency of about 1 in 7700, which is a super low expected frequency. Most players already have a sense that clips like this are insanely rare, so it is fun to put some numbers to the concept.
I honestly found it kinda shocking that the average player is expected to get zero pistol kills in 13% of their games. There are two dedicated rounds in the game where the only guns you are allowed to use are pistols, and many more where you are likely to use them due to economy troubles. Regardless, in about 1 in 8 games, the average player is expected to not get a single pistol kill. If you're curious, the pistol kill expected frequency drops to 1% when the threshold is at 10 kills, and drops to 1 in 100000 at 20 pistol kills in one match.
Curious on the details behind this calculation? Check out the Methodology section at the end of this post :)
What is the Rarest Achievement?
We used the numbers before this to decide the correct thresholds for new achievements to trigger often enough to be interesting, but not so frequently that they turn into spam. However, there are some achievements that cannot be calibrated, such as the "30-Bomb!" or "1v4 Clutch" Achievements. You can't pick the rarity for these achievements, because their criteria are built into the name. For a "30-Bomb!", you have to get 30 or more kills. It wouldn't make nearly as much sense to give out a "28-Bomb!" achievement. That means that some of these achievements turned out to be super rare.
Both of the Achievements that I just listed are fairly common:
30-Bomb expected frequency: 1 in 76 matches
1v4 Clutch expected frequency: 1 in 350 matches
What if we ratcheted this up a notch? Both of these have an easy upgrade, with 40 kills in a match being another impressive threshold, and a 1v5 clutch being one of the most hype feats in the game.
40-Bomb expected frequency: 1 in 1300 matches
1v5 expected frequency: 1 in 2400 matches
Note that this is the expected frequency of these Achievements occurring for the average player at any point during a single match! If you routinely 40-Bomb, you might wanna stop smurfing ;)
There is actually one Achievement in this category that we can calibrate, but we chose to pick the nice round number instead of selecting a number that would have a convenient frequency. Getting 10 MVP stars in one match is a super notable milestone, since you get to see the MVP number on the scoreboard tick over into double digits. However, it happens way less frequently than we anticipated.
10 MVPs expected frequency: 1 in 2900 matches
So next time you see yourself in double-digit MVPs, make sure to take a moment and enjoy the sheer unlikeliness of the situation.
Now, the moment you have been waiting for. The rarest achievement that we track is for getting zero deaths in a game!
The expected frequency of someone getting Zero Deaths in a...
Short Match: 1 in 12500 matches
Long Match: 1 in 91000 matches
This means that we only give out the Achievement for never dying in a complete Long Match just over once for every hundred-thousand opportunities, so if you have this achievement you are truly a legend. Send us your screenshots if you actually satisfy the criteria, you have to have played the entire game, and it can't have ended early due to a surrender!
What are your ideas for new Achievements that we should track? Tag @Leetify on Twitter or join the conversation on our Discord server!
The final section is a breakdown of the assumptions that went into the stats in this post, so keep reading if you have any questions about the limits of this data:
Methodology
The data in the "Killfeed Modifiers" and "Weapon-Specific Kills" sections is taken from a sample of 4.1 million matches analyzed by Leetify in April 2022. We divided the total number of times this situation occurred by the total number of opportunities to get the expected rate of the situation occurring. This total number of opportunities is the number of games analyzed multiplied by 10, since there are 10 players in each game and each player has the opportunity to qualify for the achievement. This means that this statistic is calculated from the standpoint of determining the expected frequency of an average player performing the stated feat. These values were not normalized for game length, so it would be easier to achieve these thresholds if the game goes a full 30 rounds as compared to a 16-3 stomp.
This data from the "What is the Rarest Achievement?" section was gathered in April 2022 from a sample of 850,000 recent matches.
For both sections, these matches were selected without bias from the population of all matches that Leetify processed after adding the Achievements parsing functionality. Given the extremely large sample sizes, we believe these statistics to be representative of the Leetify userbase. We expect these statistics to be fairly solid approximations of the CS:GO playerbase, but it is important to note that the kinds of players who have signed up for Leetify may not be an identical match to the CS:GO playerbase as a whole.
A possible inconsistency arises from the assumption that each of these matches were played 5v5. In the case of a player abandoning, part of the match would take place with fewer than the expected 10 players. However, the large sample sizes are also assumed to make this inconsistency negligible in the final analysis.