ฝัง
- เผยแพร่เมื่อ 27 ส.ค. 2024
- Analyzing clutch data from previous seasons to determine how much clutches actually affect outcomes of games and how clutches changed the entire course of VCT history.
Follow me on twitter: / platoonval
Support me & the channel: / @platoonval
Thanks to www.augment.gg/ for providing me data for insights.
#vct #valorant #vctanalysis
it might be worth mentioning that the impact of clutches is inversely proportional to how reliable these teams avoid getting into clutch situations, even if your clutch rates are high you would optimally want to avoid getting into 1v2 and 1v3 scenarios in general so i dont think this covers all, but it is still a great informational video.
Facts
you would essentially want a high clutch percentage with lower instances of 1v2 and 1v3s happening.
@@shankar19961 yes exactly, it was a good video but this incorporated would be awesome too, also when doing it you have to divide this by the total amount of rounds played or wherever you put "rounds" in the equations because ofc a team that played a lot of rounds will have more instances
@@shankar19961 yes exactly, it was a good video but this incorporated would be awesome too, also when doing it you have to divide this by the total amount of rounds played or wherever you put "rounds" in the equations because ofc a team that played a lot of rounds will have more instances
Pretty much exactly why loud are so low on the list but still the best team, they don’t rely on clutch plays
5:58 To better visualise the data and make it easier to understand I would add % of rounds won after previous round was won and % of rounds won after winning a round with 1 or 2 player alive left (to visualise similiar economy hit). Because right now the data might suggest 55% is a lot when the real % is not that far off from the norm
another video with a lot of interesting stats and ideas to think about. It makes me wonder how much clutch rounds are influenced by other factors, like attackers vs defenders, whether spike is planted or not, what the buys were like that round, etc.
I'd like to see some deeper analysis on some of these topics if possible.
How does the winrate of a round following a round win compare to the winrate of a round following a clutch? Is there a statistically significant difference (looking at the p-value)? How is this different based on the econ of the following round (rounds where a clutch leads to full buys from both teams the next round)?
Do some teams find themselves in favorable clutches more often? (flanker last alive, location unknown, key util left, health/guns on remaining players, etc - 1v2 definitely isn't the full story, and it's very possible some teams lose percentage points early in the rounds in exchange for higher clutch odds - comparing 100T clutches to C9 clutches on a more case-by-case basis could be enlightening)
How is this data affected when clutches with known locations are removed from the data? A 1v2 where you have to defuse on split planted for heaven is very different than a 1v2 where you are attacking with it planted for you. Is there a better way to define "winnable" clutches? I'd imagine some 1v3s should be classified as winnable, while several 1v2s should not be.
How does the data look when it is normalized by overall round winrate? It shouldn't be surprising to see a team with a 40% round winrate win fewer clutches - you alluded to this by mentioning KRU was higher than expected but didn't go into the data on it.
For the player-by-player stats - these should factor in the relative frequencies of 1v1 vs 1v2 clutches since 1v2 clutches are much harder to win. A player that ends up in many more 1v2 clutches will be unfairly placed lower on this list. Ideally, this graph would show the percentage of clutches won relative to the expected result for the average player. (e.g. since 1v1s are 50% and 1v2s are 16%, a player that has no 1v1 clutches but wins 40 1v2 clutches would be rated +86% where a player who wins 40 1v1 clutches and has no 1v2 clutches would be rated +50%). Ideally these would also be weighted by round winrate of the other team since a clutch against a poor performing team would be less impressive than one against FNC, but I doubt there is enough data for that to be statistically significant.
It would be really interesting to see a deeper dive into LOUD clutches vs. other teams. I suspect their low clutch rate has to do with their emphasis on structured, explosive retakes. They frequently win those retakes, but when they don't it frequently leaves a player not quite dead yet but with no util, possibly low health, and location known - a scenario with almost zero odds of winning the round. A big example of this mindset is round 21 of LOUD vs SEN Ascent. Tuyz has full control of A main with all of SEN coming through tree to A site, and rather than play the "clutcher" role of playing his life from main and making it really hard for SEN to plant, he uses Omen ult to regroup heaven for the retake. I don't believe their low clutch winrate is something they win "in spite of" - I believe it is their plan to win more rounds before clutches are necessary with the understanding that this will make them worse in the clutch.
It makes it really tough to do analysis, but clutches are all very different - and very strongly affected by player actions earlier in the round as well as team strategy.
These are all extremely valid and super interesting questions. I'm not well versed in statistics or have the time to do this, but I imagine this can be one of the jobs of an analyst for a pro team. Honestly statistical analysis of Valorant is pretty underrated imo, and even these surface level looks at the data are not enough for a good pro team to make any decisions, but it definitely seems like a niche worth exploring!
All great suggestions for deeper dives into clutch scenarios!
Not reading allat💀
YOOOOOOOO!!! Very smart questions!!
I do want to say that if you are winning by a large margin each time and losing by a narrow margin each time, then your clutch stat will be really low even if your team is really good.
LOUD's stats suggest to me that the team tends to win by bigger margins and avoids clutches if at all possible.
This jives with my personal observations of them, where a lot of their momentum really lives or dies off Aspas doing crazy shit. If Aspas gets away with it, then LOUD have the easiest wins of their lives, comfortably winning by margins of 3-4 players in most situations. If Aspas doesn't do well, then off their very structured executes they trade him and are more likely to get into those dangerous clutch situations and lose.
2023 C9 being the best team in the clutch does track very well with the eye test.
I remember hating to see any of their games because they were strategically terrible but would get away with their bad strats because they would just win the resulting unfavourable situations more often than anyone else.
This also does explain why they didn't make it to international tournaments though, since only relying on winning unfavourable situations is just not sustainable.
is a 1v1 even a clutch?
really cool video.definitely some data here that needs work. like post clutch effects. how much of those post clutch wins are just because of an economy swing forcing opponents to eco or halfbuy?
easy way to do this is just to compare the odds for a team to win a round after winning the previous one. use it as a control for comparison. replace the yellow threshold with that rather than just 50%.
Exactly.
yeah TS made a huge mistake dropping Ts Dubstep he hold the 3rd most clutches in 2023
it's unreal to see him up there with Boaster and Saya
It matters like if anyone watched EDG vs GiantX from last year, they were clutching rounds left and right and even 1v3 1v4 one after another
I wonder if there’s any correlation between clutches and types of agent. Like if the last player was more likely to clutch if it was an initiator or a controller. Guess boaster gets best of both being omen
I’m not sure if that is related more to agent or rather how the agent is played. I’d think lurks and recharge agents have more clutch chances since they are typically the last alive by how they’re played.
They don't usually use abilities in clutch, i only see them shooting people
I would have liked to see win rate vs expected in clutch situations. As you presented it, a team that got into lots of 1v1s would have likely had a higher clutch rate than a team that had most of their clutch situations be 1v2s
well clutchs are important to win but it's proves that your battle plan failed because the other team outsmart you or have a better read then you and then you see yourself in a 1v2, 1v3 or 1v4 situation, thats why Loud 2022 doesn't win to much clutches and win Champions because in very few rouds they put themselves in a clutch situation
I'm literally Johnny Cage.
This is a certified international love moment
Question, do you put it down like new york city?
Dragon Ranger Gaming won a lot of rounds with man disadvantage, i wonder how they stack up. Their game against 4am was really insane.
Can you make a video about paper rex playstyle when something use Gekko or any teams use gekko ? I think it might be interesting after Riots buff gekko and nerf skye 😅
Love the suggestion I'll consider it thanks!
The chosen one
Hey Platoon, first I want to say love the content. I love the angle you played at showing that without some of these clutches vct. But have you ever looked at what players are the best with specific guns? I am currently trying to get better with the phantom and want to know who is the best.
6:40 I expected as much from the inventor of Boaster's Law.
It's crazy to think 100t started as one of the best teams in NA with one of the highest clutch % to the worst. Guessed as much when you asked for who it was. What a fall from grace
When TS drop the 3rd clutches player last year. #feelsbadman 😔😔
Not the s0m shorty clip.
never forget mindfreak the clutchmaster of copenhagen 2022
Can you make a video on first kill win rate
I've played with first kill win rate stats before and it's definitely interesting! Will certainly consider it!
Is 1v1 considered a clutch? I always assumed a clutch was when the odds were less than 50% with the number of people alive
the timing of this video hits hard since 100T literally got eco'd after eeiu won a 1v3 today..
100T had some great clutches against Sentinels. You can win by mostly clutching, that’s means you are starting off on the wrong foot and have to come back. Not a sustainable winning formula
TS is 2nd?!?
Seeing this after TS lose to GenG, Ndg did ok but i miss dubstep
imo being in many clutch situations means that team doesn't hv a solid gameplan
Global Esports 💙❤📈
6:08
Bruh, it's just lucky kids
champions 2022 (for yay) lmao
1000th like
6:09 Here you might want to add the fact that Teams losing the next round after losing a clutch *that is NOT a save round for that team* because it is otherwise just unfair.
Also, I am sorry to say this but as a Data Scientist myself, most of the stuff you came up with is half baked. You forgot to take other variables that affect the graph like the one above. But hey, its unique content as far as Valorant goes.
Winning a clutch saves your team's econ. I think it's fair to attribute it to the win rate accordingly! Originally I was planning to see how much mental state alone affects the win rate but overall next round probability was too low so didn't think it was worth it to keep going.
Certainly I'm not trying to have this published or even advise a team to do something different. This is just a random YT video for VCT enjoyers 😂 but I always appreciate feedback from Data Scientists/Analysts like yourself!
This is an interesting topic, however, Cloud9 top 1 and EG top 3? This turned boring real quick
lmao boaster couldnt iwn those clutches this time HAHHAHHAHA
cool
A 1v1 is not a clutch btw... this clouds ur stats and makes ur point less impactful
a traded 2v2 becomes 1v1 and it is still a clutch
Smash
you are horrible at data analysis put some p and n values cuh
any null hypothesizers in the chat???? like are you actually braindead