I'd like to think I have some domain knowledge here[0]. So here's my two cents: there is no performance-based stat that will tell you if you're good. Rank is the only thing that matters.
There are however a lot stats that can tell you if you're good amongst your peers in your current rank.
The idea is that once your rank converges with your skill, your stats will trend towards the median.
But the good news is, most players are interested in exactly that. They want to know how they're performing compared to players of a similar rank.
This was the basis for our[1] Tracker Score performance rating, which grades player performance relative to players at their level (so both an Iron and a Radiant player can have similar ratings.)
Is this score meant to be like a derivative of one's rank? How fast one's expected to climb from their current rank? Or, more like a game-agnostic aggregate ranking?
Really only winning matters at your current rank, other metrics can be deceiving.
For instance according to the tool my league account is D-tier in most metrics and gets an overall ~400/1000, except that it has a near 60% winrate over 200 games on my main role while playing against the top-5% of players in my region. As a toplaner I generally want to be on the opposite side of the map from the current objective and I'm not going to get to participate in many kills either - so my objective participation and kill participation are both D-tier. If I made improving those metrics my goal I'd be playing incorrectly and I'd likely lose more games.
In a shooter a high KDA in [current rank] is going to correlate better with winrate, since killing your opponent is the main avenue to victory, however you could still imagine a hypothetical player who gets a lot of kills but makes the game so hard for the rest of their team, they lose anyways.
At the end of the day, only your current winrate is a reliable metric for how fast you're going to climb from your current rank.
Scoring for League presents several challenges as you mentioned, and I’m not particularly happy with the state of our League rating. In theory top should still be picking up standing objectives, which are accounted for, but yes KP will be lower for top. The goal is to use stats which can work for all roles but of course that is particularly challenging for support and top.
Author here. You're absolutely correct in that there are two levels of skill indicator.
Something like a rank or rating is the gold standard for determining how good a player is against the rest of the group rated by the same system. It's the gold standard because it's verifiable: it translates to a probability of winning and this can be calibrated over time to increase faith in the rating. But by design, it's a fairly slow-moving indicator.
Faster feedback is interesting for figuring out in-game decision rules or improving more effectively, and that's where these per-match ratings like the Leetify rating comes into play. They are naturally more difficult to verify because they are working on a very small, noisy sample, and since the outcome of the match (win/loss) is attributed to an entire team, it can be hard to detangle the individual contributions to that, without resorting to subjective judgment. Leetify have, from what I understand, tried to take a quantitative/verifiable approach to it, but they haven't shared any details around exactly how they verified it – if at all.
I'll take a look at tracker.gg and see what else I can learn about this. Thanks for sharing!
This is sad that the article goes on to explaining the crucial importance of smokes, flashbangs, etc while not focusing on them for analyzing games. In my own experience (nowhere near a great player), people that were the most valuable to a team where those knowing when not to shoot: peek to take info, throw some stuff to mess with the enemy, etc.
With some friends we used another CS match analysis tool, scope.gg, and it took many of those things in action. It knew how to estimate your grenade/smoke/flash/molotov throws on how effective they were, your risk-taking (e.g. peek at a common place to get insta-headshotted), even your footsteps being heard by the opponent in stealth parts.
Really, this tool was much more accurate than anything based on the metrics presented here. And it showed, because it consistently nominated one friend as the best one, which matched our experience, and myself as the worst one of the team, which also matched our experience :D
Cool article nonetheless, and really show the issue with current counterstrike (and why I don't play it much anymore): the skill level is so high that you basically need tools to tell you your mistakes. Otherwise you're in for a truckload of frustration as a new player.
> the skill level is so high that you basically need tools to tell you your mistakes. Otherwise you're in for a truckload of frustration as a new player.
The onboarding experience is also non-existent. I've played a few hours of CS and stopped because I couldn't figure out what was going on and I wasn't invested enough to find out. When I say I couldn't figure out what was going on I don't mean I couldn't understand the angles I should take or other tactical things like that, but, once I went beyond quick play where I just shoot everyone, what team I was on and what I was actually trying to do.
Maybe I've got this all wrong, but back when I was an active player, mostly in the CS:S days, you didn't need onboarding. You could just play a game and not think about strategizing. Really good players would do this, but for most, it was some skill shooter that was still mostly run and gun.
Exactly. In my opinion, casual CS/TF2 is the “real” game and competitive is its weird mutant offshoot. But I guess the majority of gamers today prefer hardcore competition and ranking over just dicking around and having fun.
Competitiveness and skill-based matchmaking is what killed multiplayer gaming for me. It used to be you were good at a game or bad at a game, and you could always tell yourself you lost matches because too many really good players happened to be on the other team. Now there are only two options: you play better than you did yesterday and you win, or you play worse than you did yesterday and you lose. I despise that constant pressure. I want days when I get rofl-stomped by a team of demi-gods, and, dammit, I want to occasionally be the one doing the stomping. That just doesn't exist anymore. You're playing with a bunch of people who are about as good as you and if you don't try your hardest you're going to lose every time.
Seems like you want exactly the opposite of what everybody else wants. The biggest reason why people get frustrated with games is when challange is eithet too big too small.
I like the element of randomness. I hate knowing that every time I lose it isn't because they were better than me, it's because I didn't try hard enough. I play games to have fun, not to push myself to the limit every time. What precipitated my mostly giving up multiplayer was when I noticed I was forming my habits around winning and it was working. If my room lighting was good, if I drank coffee before a match, if I played during the right hours of the day, I won. When I started avoiding using the bathroom because having to pee made me more alert, I knew it wasn't worth it anymore. I like games, not sports.
Well if you chill a bit in the system your rank will naturally go down and you will get worse opponents. Many people do that.
Its much better system than one out of 5 games being balanced. Not sure where the problem is… maybe you dont like about yourself that you are actually competitive? I mean i played esports on pretty high level and i am pretty sure not going to a bathroom to keep you alert is not a thing. Sounds like pretty serious urge to win.
Good thing is i think one can learn to be less competitive, enjoy it more and it makes people even better. You can work with this and not make it all out competition.
During TF2’s massive popularity in its first decade, “everyone” wanted fun, casual shooting. The audience for social shooters is vast, but for some reason competitive play gets all the focus today.
In general, I would recommend spending some time playing casual to get an overall feel for the game - game/round format, weapon types, popular maps and their layouts, using sound, using the minimap, among others. Combine this with some YouTube videos and you’ll be in a much better place.
CS has such a long history, yet never really a good tutorial mode. I think partly because the core game loop is so simple, but also because of the wealth of knowledge and material out there. Of course, Valve’s online games are notorious for not having a decent tutorial mode, so that might be the main reason lol. I am a huge TF2 fan, and man does the “tutorial” sell the game short.
Author here: I really appreciate the nuanced and constructive discussion in these comments -- this is the reason I post to HN. Thanks!
> This is sad that the article goes on to explaining the crucial importance of smokes, flashbangs, etc while not focusing on them for analyzing games.
There are several layers here and I should write a follow-up article to discuss that because I realise now I was stuck in my mental context too much to explain the bigger picture in the article.
This article is not, strictly speaking, trying to explain performance; it tries to explain the Leetify rating which is hopefully a fair proxy for performance. Nor is it really about giving details on how to achieve a high rating -- that requires the things you mention to put oneself in position for a favourable engagement. (In fact, one can think of the kill itself as a proxy for several other things, such as successful flashbang usage.)
This article focuses entirely on one specific kill-or-get-killed decision. It took that much analysis to provide a useful guide for that decision alone! A more complicated set of decisions I don't even know how to approach quantitatively.
Like, how would you define a successful smoke, if you had to do it based on anything that does not boil down to community feels?
I'll look into scope.gg (and tracker.gg mentioned elsewhere in this discussion) and see if I can be inspired to analyse other decisions. Thanks again!
Thanks for the clarification, I see that I misunderstood your article as "I’m dissatisfied with existing tools so I’ll make my own", while it’s more along the lines "let’s reverse-engineer leetify". And for the latter it’s an excellent one :) Thanks!
Indeed k/d is an overall good proxy when taken with a step back, eg for a whole game. When looking at a single round, it ends up being noisy. Which is not to say it’s useless! In the end this is a shooter game, and to shoot is in the name (and not to die but that’s another matter) :D Now to the matter to evaluating odds of winning a combat.
To develop a bit on smokes: with some domain-specific knowledge you could approximate its usefulness. Smoke in catwalk on D2 allowing T to go through mid while a CT stands on shirt unable to take info nor shoot? Now that’s a good smoke. But this would require very tedious work an be non exhaustive. You could make a rating of "how many people walked in front of an opponent thanks to a smoke". Or how many bullets missed a CT defusing in the smoke. Possibilities are endless but not perfect.
For example scope.gg rates flashbangs by how many people were blinded. And I’d not be surprised if they gave a negative ranking to flashbangs that were heard but did not blind anyone: those give info to the opponent without helping you.
Now if you made a follow up that’d be great but you’ll end up meta-gaming instead of gaming. But that’s fun too!
> For example scope.gg rates flashbangs by how many people were blinded. And I’d not be surprised if they gave a negative ranking to flashbangs that were heard but did not blind anyone: those give info to the opponent without helping you.
Not sure if you'll read this, but I've been trying to figure out how the scope.gg "Rating 2.0" is computed, and all they say is that it's similar to the HLTV 2.0 rating, which in turn basically decomposes into kills and deaths and proxies thereof, i.e. no contribution from e.g. flashbangs. This would be in contrast to Leetify which at least counts flashbang assists.
I haven't used it in over a year so my memory may be failing me.
And to be fair I used their match explainer that focuses on rating in a single round, and to show you your mistakes. They rate pretty much everything, including the flashbangs I described. I must admit, I was not interested in my ranking, only in my mistakes. It is very possible I assumed all ratings in a match analysis counted towards your overall rate, but appears to be wrong.
So yeah, in the end we are not talking about the same thing. Sorry for bothering you with this!
(although it would be pretty interesting to have all those fine details be at least a signal for a player's overall rank)
The standards have gone up a lot over time. Your average player now knows who the tops teams are and watches them play. Vastly different from the early 2000s when just knowing to aim at heads set you apart.
Even to be a top 10% player these days you need to either be a veteran or to practice for long hours.
Must be hell for anyone new who wants to try the game!
On the other hand, there are a lot more competitive video games than there used to be, so the best competitive gamers are spread all around instead of just in Counter-Strike.
Yeah, as someone who grew up in ye old days of even pre-Steam CS, the skill level the "average" player has now is incredible. Hoping into a casual match of Counter-Strike you'd think you accidentally ended up in someone's scrim match.
I don't have the time or passion these days to practice, but even if I did, I'd imagine it'd still be a very difficult environment to just be somewhat decent.
I tried to get good at fpses back in the day but never did.
I was however making decent progress at learning an instrument at the time (the French horn, if you must know) and thought there should be some sort of tool to learn the fundamentals of first-person video game shooting using time honored music practice techniques.
The key to getting really killer music technique is to go as smoothly and calmly as possible, over and over again, even if that means going at a fraction of the target speed.
I would have liked to see an fps trainer that lets you practice shots at anywhere from 1/5th to 1x game speed, so you could master it at very slow speeds and gradually up the "tempo".
(There are some very good trainer programs now in AD 2024, but none that I know of that let you manipulate speed like that.)
I‘d say I‘m very proficient with FPS games. I also got to Global Elite in CS:GO. My take is that you mainly need two things, map awareness and aim. The first can only be acquired by playing A LOT. The latter too but the time it takes by playing gamemodes line Free For All Deathmatch where you get to practice the shooting aspect very frequently.
If taken seriously I‘m confident a player can get proficient within 500-800 hours of playtime. That sounds like a lot but with daily practice/playing ofr 1-2 hours its not that big of a deal.
And talent. CS is one of the first times in my life I saw people try very hard at getting better. But it just didn't work. Like some people put in a fourth of their effort and they would still run circles around them.
As someone who played professional CS in the 1.6-Source era this write up gave me a hearty chuckle. It must be what it feels like when a pro footballer reads a really intense parents write up on u12 tactics.
I'll second this. The upper echelon of 5v5 is an entirely different game. It's similar to what was alluded in the post[1] about chess in another thread. The level of complexity is so high and the depth of refinement is so technical that generic statistics just don't quantify the conclusions we draw from simply watching someone's FPV. One would need to start by analyzing data at the most basic level: did the player move between two points as efficiently as possible without penalty or mistakes? How far away was their crosshair from where it needed to be at the point of contact for each engagement? How often did they misjudge where everyone else was? How accurate are they at timing enemy movement to another location without seeing them? What is their level of consistency at making certain well-known plays at certain locations? These are just the basics. The game-world data is available though for analysis at this level of detail.
Counterstrike is identical to chess, there's tiers of players. I never made it to CAL-I but I was the tier below. When playing against CAL-I tier people, we'd get stomped on.
Flipside, if I went to a pub, it's quite likely that I would stomp on everyone else. I still can do this in TF2 and tryharding.
The other difference is that chess is a 1v1 game: the only person you can blame if you lose is yourself.
Counterstrike is a team game, where (if the matchmaker was competetently designed) on average you shouldn't be able to carry your team. I think this contributes to a lot of toxicity, because it's very easy to find a cause for the loss that isn't your own skill.
Imagine a team match of chess where everyone controls a couple pieces. At low brackets, you could know "tech" like en passant or underpromotion. You might have vague ideas like "control the center" and might know a few lines of theory. You might be screaming at the screen when a teammate castles into mate-in-3 and blame them for your loss - even though you were the one who moved a pawn to h3 on turn 1.
Ladder or ranking is the best overall performance indicator (given ideal conditions, no cheating).
It is a system of averages. I was a top 80 player in Age of Empires: AoK and easily beat the crap out of everything from below the top 100.
Above me, I could compete with most players on some maps on some days, top 50 would beat me in a best of five, but top 20 usually eat me under competitive conditions.
Most of them were already pro gamers back then and trained hard to maintain their ranking.
It piles up. Micro management especially. Being constantly a couple of milliseconds faster in organizing stuff and decision making is like a formula 1 driver who is 0,5 faster per lap. Worlds apart with 60 rounds to take.
Back then I got the intellimouse explorer from microsoft, a nice precise mouse. Shot everyone in the head and kicked from all servers. Even Valve sent me a message that one more ban from a server and my account is frozen. End of CS for me that day. Thanks God
> Counter-Strike is played by two teams with five players in each.
As someone who played CS from the first beta, this is a convention that started with certain competitive leagues but was never a hard and fast rule. Most maps don't play very well 5v5, but that format took off because it was harder to organize more people at once on IRC.
This could have been fixed with matchmaking, but I suppose the toothpaste was already out of the tube.
Surprised at the small size of the dataset here. Surely it’s possible to record thousands if not millions of results… has someone been able to do this? Seems like the simplest thing to do here would just be to develop some sort of plus/minus system and not try to overthink things. I work in sports analytics, and mostly in low scoring sports so we don’t get to do that very often.
> Surprised at the small size of the dataset here.
Author here. For answering the research question that spawned the article (engage or stay away from a do-or-die encounter), the data set actually appears to be "too large" – see the significance values of the parameters. I got carried away when copy-pasting match result tables, but the effect is so clear that a much smaller data set would have sufficed for showing it.
I think the thing you're asking for is what the Leetify rating tries to encode, but people in this discussion have suggested alternative rating systems which I have yet to look into.
Somewhat related and tangential to the thread, does anyone have an idea why CS feels so different compared to other FPS games? Despite playing the game since my school days, I’m still about as good/bad as ever. Encountering an enemy gives roughly a 50% chance of me dying. However both in single player fps games (COD/far cry/GTA etc) as well as other multiplayer games (only COD to be honest) the game feels much more forgiving.
It’s a weird thing from a game design perspective for me. One possible reason could be health regeneration but that ought to affect enemies as strongly too.
In most FPS games, a single shot does relatively little damage. A typical weapon in Apex Legends or CoD takes about 0.8 seconds to take a player down to zero health. If you're losing a firefight, you'll often have the chance to retreat, heal and re-engage unless your positioning is very poor.
In CS, many weapons are capable of killing in a single shot and most are capable of a kill in less than 0.1 seconds. If you carelessly blunder into the crosshairs of a player with good aim, you'll probably die before you've had time to react. This is a deliberate design decision, because it raises the skill ceiling. You can get away with errors in positioning and game sense at a lower level, because your opponents are likely to have poor aim; as the standard of aim improves at higher levels of play, tactical errors are punished with increasing severity.
So you don't need much actual skill at aiming. Plus weapons generally have huge magazines relative to the number of shots needed to kill people, so you can spam fire and will usually get a kill.
More casual shooters might have auto-aim, especially on console. CS is very fast and twitchy compared to more mainstream shooters.
Also, CS has recoil that moves you cursor / aim point. Other games might only have bullet spread. It's an interesting decision whether or not recoil should extend to UI elements like cursor position, in CS it does. One result of this is skilled players can move their cursor just right and counteract the recoil, potential putting every bullet on target like a laser beam, but I don't think any human can fully realize that potential.
Besides what everybody says there is also slight difference between engines in terms how they behave.
I used to be “progaming” going on tournaments in quake 3 and then COD1 and 2. They are all on same quake engine and feel very similar with different timing.
The way of the counterstrike source engine feels is different in a way i never fully absorbed. Also the gunplay is different you might have the pure aim (aim at target and click) but im CS you have to compensate for the each gun and really know them. In quake it basically shoots where you point no matter if you are running or in middle of a jump. Other games have aiming down the sights. That is a lot more forgiving and transferable skill than in CS.
Wat? CS is crawling with cheaters. Totally sucks. I can’t wait for the day when we have enough bandwidth that the entire game can be rendered on the server side and you just get a video feed. That would end most of this crap.
I’d rather read about why
are russians bad. Or more like why are they are playing the way it is. When you need to eco > they rush B. When you need
to save the round >
they go back 1v3. But that not just in CS, they do the same in Dota too. They just play and see these games
differently.
The argument I've heard is that Russian players have relatively high ping to EU servers, so are more prone to becoming frustrated and playing in a silly but fun way.
I'm not entirely sure I buy that, because the 2022 invasion of Ukraine was basically "Rush Kyiv, don't stop blyat"; an overwhelming advantage in manpower and equipment was squandered due to poor planning. Russia seems to have a very long martial tradition of just charging ahead and trying to win through sheer force of will.
fun thought exercise, but the fact that a relatively new player is exposed to services such as leetify and learning that way nowadays blows my mind.
trying to quantify performance this way is stuck somewhere between extremely abstract rating like elo-rating (that cs internal rating is originally based on) and the moneyball realm in modern sports.
this is all well and good inside a controlled environment of professional sports, but should not be taken too seriously outside of that scene.
i think the only thing one should care about is a bayesian measure of how you impact the odds of winning the round/game in the server/lobby. regardless, i would definitely love to work for a team that is passionated about cs outside of daily tasks!
> This means that a player, just by existing, is providing negative value to their team.
> A player has to actively do something good to reach break-even.
> Thus, if we think we won’t contribute positively to the match (and our goal is to improve our team’s chances of winning), we should not play.
> My Leetify rating in recent games have been around -4, which must mean my success rate in firefights has been about 40 %.
> That seems reasonable, but also highlights how devilish this game can be.
> 40 % isn’t a bad success rate. It can even feel fairly successful thanks to natural variance around that.
> But it most definitely means I’ve been contributing negatively – and I know I have.
That's brutal... Implies that when you have bad teammates you need to be even better to compensate for their suckiness. I suppose that explains why a lot of people get very angry (to say the least) during matches.
There are however a lot stats that can tell you if you're good amongst your peers in your current rank.
The idea is that once your rank converges with your skill, your stats will trend towards the median.
But the good news is, most players are interested in exactly that. They want to know how they're performing compared to players of a similar rank.
This was the basis for our[1] Tracker Score performance rating, which grades player performance relative to players at their level (so both an Iron and a Radiant player can have similar ratings.)
[0] https://tracker.gg [1] https://tracker.gg/articles/tracker-score-our-new-performanc...