Exploring headshot percentage

Taking a look at three ways of calculating the old standing stat.

Jul 11, 2025

Headshot percentage has been around for a very long time. The number of kills gotten with headshots appears on the earliest CS:GO box scores, even predating ADR (Average Damage per Round). In this article, I’m hoping to provide a general overview of what this stat conveys, and introduce alternate ways of calculating it, now that we have the means of doing so.

For this piece, I’ll be taking a look at stats from the HLTV labeled Big Events from the first half of 2025. To get the advanced metrics, which are unable to be found on HLTV’s stats pages, I’ve collated a database of stats extracted from the demos (generously provided for free by HLTV on their match result pages).

The Classic - Kill Headshot Percentage

Let’s start off with what has been the standard way of calculating this metric. If you go to HLTV’s leaderboard section, and take a look at the Headshot percentage section, you’ll see players sorted by the number of headshot kills divided by the total number of kills gotten by a player over a given span of time. Does this metric correlate with on the server performance? Let’s find out by plotting HS% against kills per round:

We can clearly see something which might be obvious for the more experienced reader but is still worth pointing out just in case - the green cluster of nodes representing AWPers near the left of the chart. Snipers tend to have a lower headshot percentage since, well, you don’t need to aim for the head to get a kill with the AWP. I think we’ll have a clearer picture, if we filter them out.

We would have a clearer picture, if it wasn’t for one pesky outlier. Get out of here, Danil “donk” Kryshkovets. Take two, this time with only regular mortals included:

If we take a big step back from the screen, and look at this chart from afar, we can see that sort of, kind of with more headshot% come more kills, but that correlation isn’t strong at all. I plugged the numbers into Excel and weighted them according to rounds played, getting an 0.30 R-score between KPR and kill HS% (note: R-score operates on a scale, on which 0 is no correlation whatsoever and 1 is a linear or full correlation - as one metric rises, the other does too).

This means, that players like David “frozen” Čerňanský (low HS%, but high KPR) or Kaue “kauez” Kaschuk (extremely high HS% but very low KPR) shouldn’t be a surprise to us. I think this proves, that headshot percentage isn’t really a production stat, it’s a stat which is descriptive of a player’s playstyle. This was also proven way by HLTV’s very own Harry “NER0” Richards back in 2021. But does it do the best job that it possibly can at telling us which performers aim for the head?

The True - Bullet Headshot Percentage

Another way to calculate headshot percentage, would be to treat every bullet the same - either hit the head or it hit another part of the opponent. Not counting missed bullets is crucial here, as it would introduce a lot of noise into the data without a sure way to filter out all the random smoke spams. I felt like this would be a poor way to treat things, since my intuition suggested that this percentage would depend a lot on the gun a player’s using.

But alas that was not the case. Pistols led the way in bulletHS% due to the fact you cannot spray them, yet all the other weapons hovered right around the 20% mark. Sure enough, Leetify defines Headshot accuracy as ‘all shots that hit the enemy in the head divided by all shots that hit the enemy’, with the exclusion of AWP shots, so this metric is already in use. Thanks to that exclusion, AWPers can once again come back into the fold.

Crikey. Come back they did. Ilya “m0NESY” Osipov took the top spot on the leaderboard on a sample of shots comparable to the struggling Keith “NAF” Markovic, even overtaking former teammate and headshot king Mario “malbsMd” Samayoa. Out of curiosity, I checked m0NESY’s kill headshot percentage with AWP kills excluded, and he would outrank malbsMd there as well, with a ridiculous 68%. Roland “ultimate” Tomkowiak also entered the top 10, while his countryman, Janusz “Snax” Pogorzelski had a notable drop compared with his killHS%, which placed him in the bottom 5.

To underline just how much NAF struggled - in 2024 he had a 20% bulletHS% which dropped by 4% percentage points this year, according to nomiun. I feel like the more analytically inclined teams should probably start monitoring similar drops in HS%, I’d equate it to a baseball pitcher’s fastball velocity dropping before an injury.

While the changes in bulletHS% aren’t very dramatic from weapon to weapon, I feel like it would be good to level the playing field and take a look at the metric only while using the best rifle in the game, AK-47.

The undisputed AK headshot king for the first season of 2025 has been crowned, and of course, it is kauez, who after being cut from paiN Gaming currently resides in RED Canids (alongside André “drop” Abreu and Marcelo “coldzera” David). Another high placement is the recently benched (with 3.5 years left on his contract) Timur “FL4MUS” Marev. His former IGL, Denis “electronic” Sharipov has never been a headshot machine, but is now 7th from last, placing below Casper “cadiaN” Møller.

One more notable placement is Azbayar “Senzu” Munkhbold. He ranks 7th (9th if you include kauez and Marcelo “chelo” Cespedes) on HLTVs killHS% leaderboard and does so with 7.5% of his kills coming from the AWP. It is usually Ayush “mzinho” Batbold who gets the plaudits for his head-clicking ability out of The MongolZ roster, but maybe it is time for us to reframe that conversation.

The New - Damage Headshot Percentage

A few weeks back, I proposed that we take a look at damage headshot percentage. It was calculated in a similar way to killHS% - it takes the damage done by a player to the opponent’s head and divides it by the entire damage done. That was a decent enough definition, but it failed to take into account that damage from grenades shouldn’t affect an aim describing stat. For the purposes of this article, I defined damageHS% as the amount of damage dealt in headshots, divided by the total amount of damage dealt.

I’d argue it does a better way of describing a player’s aim than killHS%, solely because of the sample. An average tier-1 player gets just 14.26 kills per map, each having a binary outcome - kill was either was a headshot or not. Looking at the damage dealt not only includes all the hit bullets (45.27 events per player per map, a 3x increase) but also introduces proper weighting to each bullet. As much as I love numbers, I do believe that they should just be used as support for things we notice while watching the game, and a one-deag on a full HP opponent just feels more important to me than an MP9 dink. This also means that AWPers will be excluded from the following section, as they deal a significant amount of damage aiming for the body.

Damage headshot% seems to rate FL4MUS and Leonid “chopper” Vishnyakov higher than both killHS% and bulletHS%, roughly meaning that most of their impact comes by getting headshots. Gabriel “FalleN” Toledo ranked in the top half of all riflers when it came to bulletHS% but got punished a lot by AWPing for the first half of the season.

Does it really make a difference?

In this final part of this article, I wanted underline the differences between the three metrics, which all in all aim to measure the same thing.

As you can see, some differences in the rankings between killHS% and bulletHS% aren’t all that cosmetic. As established before, FalleN was undervalued by killHS% due to his AWPing, while players like Johannes “tabseN” Wodarz and Justin “jks” Savage made jumps from down the bottom of the list, to being solidly average. Johnny “JT” Theodosiou, Rasmus “HooXi” Nielsen and, surprisingly, Oldřich “PR” Nový went from top half to bottom half head-aimers.

I looked at the aforementioned R-score, or in non-nerd language, how much the values between metrics really differ - differences between damageHS% and bulletHS% among riflers were mostly cosmetic. That combination of the metrics had the most correlation with a 0.96 R score (again, on a scale of 0 to 1, where 1 is strongly correlated).

Taking into account only the riflers, killHS% and bulletHS%, while being completely different in calculation, still had a 0.92 R score. DamageHS% correlated well with killHS% (0.94) but had the most correlation with bulletHS% (0.96). I’d argue it brings us the best of both worlds - both the raw frequency of head-clicks of the bullet metric and the impact of those clicks provided by killHS%.

The bombshell

As established before, we should pick and choose our metrics to best support our arguments, which brings me to one last thing I’d like to leave you with. It’s well established that Myroslav “zont1x” Plakhotia purposefully aims for the body, voo “cs2” csgo even made a great youtube video about it. During my research for this piece, it turned out, that he in fact does not have the lowest damageHS% on T-sides this year.

Far from me to imply Snax is anything but a legend of the game - I got my IEM Krakow tickets after all - but him ranking below zont1x in this metric, might just mean that his years as a top fragger are behind him, and it’s alright for him to step down as G2’s IGL.