http://dbd.game/killswitch
Devs learn how to do stats
So in this forum post page 2 https://forum.deadbydaylight.com/en/discussion/68011/the-reason-why-this-game-is-survivor-sided#latest Peanits"There's definitely some maps that are higher than others, but the overall range is between ~39.22% and ~53.60%. Not an insignificant difference, but not massively unfair for one side or the other." Normally this will be true IF they did right the math devs NEVER count the DC or the suicide on the hook with mean those numbers can be even higher and only there to confuse players in think that the game is a balance. So what i ask from the devs is you release stats that are correct and count only the thing they say the count.
@Peanits Will you do that ?
Comments
-
Where did you get the stat picture from? I know there was a thread with stats, but I don't remember this particular one.
Regarding your post, I agree that "clean" stats would be cool as there was also a missunderstanding in the survival rate of rank 1 PC users.
0 -
I post it where I get the picture
0 -
Sorry, didn't scroll down and the link took me to a few posts above the picture
Thx
0 -
Stats will never not be skewed. Generally, outliers are not included in the final recordings since it unfairly skews the data (especially if it's minimal).
I do agree that more than the chart should be given when stats are being used as part of a discussion like the parameters and any limitations, as someone who has done research. It doesn't have to be an academic article (actually I would be intrigued to see this if the devs do have it), but there needs to be a little bit more information given.
0 -
It's safe to say the those stats were influenced by DCs if they didn't exclude them.
You have to remember that if someone DCs then they leave it a 3v1 with 5 gens, In most of these cases the killer will end up killing the remaining survivors. With DCs being almost every game this is a common occurrence.
So then you have not only the person DCing but all the people that are left behind from that person DCing counting towards the kill count.
If they want true stats they need to exclude those games completely not just the person DCing, Take matches that everyone involved play trough to the end to draw conclusions from.
1 -
I agree that their stats are flawed, but for other reasons.
Specially one, I must say: they show individual Survival Rate per Rank, most played killers per rank and most used perks per rank, but no other statistic has rank specifications.
And individual survivor rate doesn't say a lot, really. It's your INDIVIDUAL chance to survive a match, but survivors play on teams. Even if your chance to die is high, it might be a match where only you die and all the other 3 escape. Killer's kill rate tells a lot more, because it is about how many got killed and how many escaped.
But look, they don't show kill rates per rank, only per platform:
So, it's the average on all ranks. What are we supposed to get from this, since there are BIG differences between rank 1-2 and rank 16-20?
I guess they don't show this stats because some killer's kill rates are freaking high on low ranks while freaking low on high ranks. It would show the big balance problems the game still has. Instead, they show inaccurate stats that can be interpreted on different ways so the community keeps arguing over them with two separate sides.
Funny thing is: When the stats from October 2018 had shown survival rate around 40% at high ranks, Mathieu Cote used them to argue the balance was fine. He even dismissed and laughed at a player coment complaining about killers being undepowered, saying "survivals are being killed most of the time".
Of course he didn't say those were only individual stats and didn't say much about who wins and who loses. And of course he didn't use the same logic later or said anything about the stats from April 2019, showing survival rate at 60% at ranks 2-4 and 80% at rank 1.
4 -
It reminds me of when they said Self Care only has a 24% usage rating, without mentioning the fact that you can, yknow. Use 4 perks lol.
This meant Self Cares usage rate was actually over 80%.
Their stats are garbage. Sorry to be rude BHVR, but you really don't understand how stats work, and/or you're being facetious as to placate the fact you don't have to take the balance seriously.
4 -
Yeah there are many, many things that their stats do not take into account making them near meaningless, much less actually being used for determining balance (which they do).
Show kill rates (not individual survivor rates) at each rank excluding any games with DC's or someone killing themselves on hook.
Also, how many kills are added to each game because the gates were opened and the survivors decided to just troll (which happens a lot) or the killer just camped the last guy on hook because there was no reason to leave. I'm not blaming the killer for camping in that scenario, but it certainly is not an indicator of "balance".
These stats should not be used for balance as they are just flawed from so many angles. The fact that they are using them to determine balance explains why 3/4 the killers are just not viable.
I want to assume they are not releasing faulty stats to intentionally mislead players about the games actual balance, but that is indeed what the effect becomes.
Post edited by Blueberry on1 -
Here's what we wanna see @Peanits
1. Kill rates in games without any DCs. An early DC often leads to a snowball game, or Survivors giving up. These games should NOT count at all.
2. Kills that occur after an exit gate has been opened, including EGC kills.
3. Survivor Pip Up Rate in regards to whether a Survivor is killed and still Pips up, or manages to escape and still fails to Pip.
2 -
Giving us the specific stats we want would probably show us that we're right and throw Behavior in a bad light.
That's a no-no. Hence all of these half-arsed statistics.
1 -
There's a huge issues with the first one: That is going to massively skew the stats. People aren't going to disconnect if they doing well, but as we all know, there are many people who will disconnect the instant they go down. If you exclude matches where anybody disconnects, you're excluding matches where someone does poorly and rage quits and only focusing on ones where the survivors do well.
We can only do so much when providing stats, and any time you exclude things the stats get more skewed. This is why we give general stats rather than picking and choosing what to include. This is why we recommend not drawing any conclusions based on stats alone.
The other two we could possibly look into if we release stats again, though.
5 -
Exactly.
0 -
"If you exclude matches where anybody disconnects, you're excluding matches where someone does poorly and rage quits and only focusing on ones where the survivors do well."
That is a very good point to bring up and makes sense. Thank you for clarifying that.
However, if that is the case, then these stats are meaningless and should not be used for determining balance, but they are. Why?
"The other two we could possibly look into if we release stats again, though."
I appreciate you looking into adding those, thank you.
3 -
So you're essentially balancing killers around data where survivors are disconnecting so it looks like the killer is really good and kills everyone, as janick said plague is a high tier killer (really isn't), going off these stats, you should NOT balance around skewed data but instead what the playerbase suggest, who know the game. @Peanits @Janick
7 -
+1
Basically what I'm thinking as well.
2 -
He really said that Plague is a high tier killer? XD Facepalm
0 -
I'd have to look back but I believe it was actually "highest" not just "high".
1 -
Thank you, I greatly appreciate that.
As I've said for years, I think the Pip Up rate will be the most accurate in determining whether or not the game is somewhat balanced, as dying doesn't really mean anything when you still get a Pip in my eyes, you still "won" in terms of Ranking, know what I'm saying?
I definitely can see how taking out DCs can still skew it, it crossed my mind after I made the post. At least from my experiences with the game, however, including them skews it MORE. I think showing both of them side by side would allow a little bit of comparison, wouldn't you agree?
I appreciate you considering these parameters @Peanits
1 -
lol even better haha
0 -
Correction, I looked back and was mistaken.
It was "Top tier (4k) at rank 1."
2 -
Well, devs didn't learned anything from the past 3 years.
1 -
This is what I mean, they're balancing ass killers around data that shouldn't be the case. Balancing their entire killer base on stats, that aren't even accurate. That's not a good idea from any point of view, BHVR, you need to balance around what full time players say. Look at the forums, how many posts say ghostface, plague, clown is ass yet the "data" says otherwise so according to you they're fine. BHVR this needs to stop it's not a healthy way to go about things.
1 -
This is interesting because we've been told now that these stats are heavily flawed, yet when the community tells them certain killers are weak they say those same stats (that they admit are heavily flawed) show otherwise...
There is a lot of contradiction going on with this.
2 -
The topest of tiers. Apex chadis Killeris.
By this logic Plague can down Survivors faster than Nurse because they're all in the broken state right? lol.
1 -
People are looking for certain things to be excluded because they are not a normal match. The game defines a match as 4v1 to the end. Not 4v1 -> 2v1 due to dcs. Some of these stats also included games that were not fully prepared from the start ie 2v1 1v1 etc. They are outliers in the statistics which skew them to be more balanced than they really are. For example the mean of the following 5 numbers: 1, 5, 3, 2, 200 the average of the number is 42.2 with a standard deviation of ~88.225 which is extremely high in value due to the 200 as an outlier. Similar to how games with instant dcs or suicides heavily skew the game in favor of the killer. These games are not in anyway accurate to be involved with actual intended matches. You guys ban for multiple dcs when ban waves go out. So this is clearly not 'normal' as far as the game developers/team/whatever defines it. Which means it should not be counted in statistics anyway.
0 -
I also noticed this. Good point out.
0 -
"There are three kinds of lies: lies, damned lies, and statistics." - Mark Twain
It's almost impossible to have good stats in this game that aren't up to personal interpretation.
Take DC's as a single example. Survival rates would skew lower if they include games with DC's and skew higher if they don't include those games. Both are irretrievably altered by the fact that DC's occured, unless DC's only occur in a statistically small sample of all games. Anyone who plays this game KNOWS FOR A FACT that that is not the case, so it's impossible to have good stats with or without DC's for DBD at the current moment.
And this is just a single issue that is just the tip of the iceberg:
- How many survivors are trying their best to escape at all ranks?
- How many killers are being as ruthless as possible at all ranks?
- Do the players consider ranking up or getting blood points to be more important?
- How many hours does each person at each rank have playing that particular killer? Should a rank 1 Nurse with 10 hours of Nurse play total be treated the same as a rank 1 Nurse with 1000 hours?
- How do events affect the stats?
- How do ultra rare add-ons, items, and offerings affect the stats?
- How many players on each side at each rank have access to fully optimal builds?
- What the flying fudge do they do to account for 2 person, 3 person, and 4 person SWF games? Leave them in? Leave them out? Have separate stats for solo and each tier of SWF? Try to separate good SWF from bad SWF? Only care about games with SWF that don't have a major rank imbalance? The resulting stats are basically bollocks no matter what you choose to go with.
- How do specific perks specifically affect the stats isolated from other problems mentinoed? People with Mettle of Man probably have a higher escape rate right now than people who don't, but how in the world would they even prove or measure that in order to make a fair decision about how powerful MoM is pre and post rework/nerf given all of the other factors already listed?
- Etc. Etc. Etc.
Throw in any biases of the people either gathering/presenting or reading the statistics given and almost anything can be used to say almost anything given enough spin.
At the end of the day, what's really important is whether you believe the devs are doing their best to use these statistics to balance the game (my personal belief) or you believe something more sinister is going on. As a humble armchair astrophysicist, I don't have enough data to prove either one is the case.
0 -
Why would you exclude EGC kills? So for kill rate stats, a 4K that featured a hatch close /entity kill would count as a 3K game?That would blatantly skew the stats. Maybe I’m misunderstanding cause I’m not following at all why this should be the case
0 -
You're probably right EGC should be included now that I think about it. Thanks.
0 -
So let me see if i get it you chose to give us stats that are wrong, incorrect and give a false idea about the balance of the game instead of giving us stats that are correct or match the idea of your game ( 4v1). So why don't you give us the % of kills that killer get per game and rank of how many DC we have per month and rank?
P.S. Don't you say again that you balance the game around the stats you collect because even that is false. @Peanits
0 -
I have to admit that I'm quite curious with kill rates per killer as well. I feel like this would give a far more proper depiction of the game.
0 -
Why not showing both? Matches including disconnects and matches excluding DCs? Would also show what a big difference it can be to dc and screw your mates.
1 -
@Peanits @not_Queen still waiting
0
