A question on the 48.30% escape rate
We saw the latest stats how 4 man swfs escape 48.30% of the time.
I was wondering about it and I want to know if anyone can help me understand.
Of all the 4 man swfs that play in high mmr they escape on average 48.30 percent. Is it possible that of this specific group of players (4-man swfs high mmr), that some of the 4 man groups are actually performing worse than 48.30 and that some are performing better than 48.30? Of course averaging out to 48.30.
If the answer is yes, is it possible that there are some swf teams that consistently get over a 50% escape rate? In a game that is designed for Survivor disadvantage and Killer advantage (60% kr) being able to still escape over 50% is pretty impressive in my opinion.
Comments
-
Yes its an average
meaning if little timmys 4 man group escapes 25% of the time
and big Jerrys 4 man group escapes 75% of the time
The average is 50% escape rate
Some people can be far above and far below the listed rates
5 -
I'm pretty sure Hens and his squad almost got to 200 wins in a row, so yeah in high mmr some teams are gonna be over 48% escape, just as many are gonna be below
1 -
MMR in general is a bell curve, and every individual MMR bracket is also going to have a bell curve in terms of skill. You have players that barely squeaked above the soft cap and struggle to hang on, and you also have the cream of the crop. There are players who struggle to win 50% of their killer games in "high MMR", and you have killers like Numinous who have 1000+ wins in a row on Billy.
4 -
Remember that stats don’t take what happened in a game into account either; so if a team did all 5 gens in 5 minutes but then decided to mess with the killer until they all died, or refused to leave despite the exits being 99’d, or didn’t try to do gens at all but just Head On Flashbang the killer for 20 minutes for content, so escape rate doesn’t really show anything of value. Much like MMR being based off escapes doesn’t really value what was done during the match, so if you ran the killer for a 5 gen chase but didn’t escape, was it really a skill full play?
4 -
Hens333s team had a 200 escape streak and I believe Eternal had a 380+ escape streak. Those streaks alone probably boosted them a few points.
1 -
Individual games don't matter in this context. It's an average of all games played by all players in a given MMR bracket. Escape rates in games with Head On and Flashbang would be something you could figure out from a given dataset, but it doesn't make aggregate escape rates irrelevant. Those are completely different user stories a data scientist would work on. Much of the community has a fundamental misunderstanding of data science, and they're upset that they got an answer to a question they didn't realize they were asking.
3 -
The soft cap makes win streaks (and loss streaks) possible and expected. Say, for example, you have a person with a 300 MMR point difference after the soft cap. So, assuming the soft cap is 1600 and the Elo system works like the other Elo systems, a player at 1600 MMR has less than a 1 percent chance to win when placed against a 1900 MMR player. The 1900 MMR player has less than a 1 percent chance to win against a 2200 MMR player.
As the number of players at higher MMR become even rarer the higher you go you have a situation where someone in the top percentile would win against 99 percent of the population but would also lose 99 percent of the time against someone in the top tenth of a percentile of players.
As the number of players at a soft cap would be higher the further down you go and rarer the higher you go up anyone at the soft cap probably has a lower win rate than the average and anyone significantly above the soft cap will have a much higher win rate than the average.
That's not a bug of having a soft cap; it's how the system is designed and what happens with a soft cap. The only alternative is a queue lasting up to an hour for the highest rated MMR players (as seen with Dowsey when MMR was first introduced) which would drive away all of the players at that level.
3 -
Sure they matter though as we don’t know how many games out of the ones the data is taken from these sort of things happen in which can massively skew the stats.
How many games could’ve ended in a 1k but turned into a 3/4K because the survivors wanted to get the last one out.
I’ve just been guilty in a match where 1 was on hook and both exits were opened with one survivor being pushed out the other gate while 2 of us healed up at the other. We could’ve just left and got 3 out but went back for the 1 on hook and all 3 of us got killed for it. That made the kill rate look better than it should have and although we dominated the game easily, our MMR won’t increase nor will the escape stats show how much of an easy time we had; it will look like the killer dominated and got a 3k. There needs to be more taken into account in both MMR and these statistics as they don’t mean much without the context of what else happened during the matches.
If everyone dies at the end to NOED despite the exits being open it doesn’t mean killers need to be nerfed, it could mean NOED needs to be looked at or the survivors should’ve just left.
How many matches did a survivor kill themselves on first hook 30 seconds into the game leaving it a 3v1 where the others are doomed. There are things that massively affect the outcome of a match not being considered in kill/escape statistics.
How many games are 2-4 survivors killed by EGC and is that because they didn’t want to escape and raise their MMR, a killer can’t make a survivor escape even if they’ve lost, only a survivor decides if they want to show as a kill or escape on the statistics or MMR.
There are lots of factors in matches that need to be considered and just looking at a binary kill/escape statistic doesn’t have enough accurate data to make meaningful decisions on, wether that be balance or MMR etc.
1 -
Those are fair questions, but they're more qualitative. All escape rate is asking is "how often?". That's all it's meant to do. I'm sure the devs look at those things too. It's why we get seemingly random buffs and nerfs despite escape rates being in line with their goals. Those things are factored in to a degree. Escape rate is measured on the outcome of all games, and all those things happen in games. But I'd caution against discarding escape rate entirely just because it's not meant to answer qualitative questions.
2 -
But the same is true of course for people putting too much stock in those statistics, as you say it’s not meant to answer qualitative questions either way, yet people quote them all too often when discussing things like balance and one side vs the other etc.
The developers themselves have always said to take the statistics they show us with a pinch of salt. It just hasn’t made things any better when we nowadays have an MMR system based on these binary statistics as the win condition.
1 -
I will be the first to say that match quality sucks more often than not, particularly in solo queue. I often feel like the game straight up doesn't respect my time. The trials are usually cutscenes in favor of one side or the other. Do I think that the escape rates are a bit contrived via matchmaking? Yes. I think the matchmaking algorithm absolutely tries to keep solo survivors from winning too often. I will die on that hill.
I don't put everything in escape rates, but they are what they are and every player experiences the same matchmaking algorithm.
3 -
Hook Kobes and other things affecting stats - While this is true, I think the DC rate being ignored when it comes to stats makes the stats far more dire. Think when people DC, when a Killer is acting like a complete and total scumbag, such as bleeding out for the 4k. Now these (predominantly) Killer wins are completely ignored by the stats, while people are still experiencing that annoyance/torment. In my matches in soloq, far far far more DCs negate losses in the stats than Kobes force a loss (especially since most of those Kobes are when someone ran the Killer for 2 minutes without a single gen pop, or sees basement campers, or locker rats, all of which would have been a loss anyways). Overall I think kill rates are deflated from DC removal far more than all the other factors inflating them, or at least among my matches on both sides (especially given that many of these were only accelerating a guaranteed loss like a Kobe attempt at 5 gens and 7 hooks).
3
