The actual Survivor win rate question
What do you think it is when it's defined as 3E or 4E Survivor win, 2E draw, 0 to 1E Killer win?
Do you think the win rate is equal to escape rate? Higher? Lower? 50% or more? 30% or less?
Comments
-
I'll start with a qualification of my answer: in most games win and loss are simple statements. Whether you win by one point or fifty, you won.
DbD is different. Both sides have a goal to accomplish and you can accomplish varying degrees of that goal. A 3k is a win for the killer, but not as much of a win as a 4k.
Survivor is a little more nuanced. A killer has a clear objective, kill the survivors. Survivor has the question of whether they care about the other survivors or not.
For me, personally, surviving is my priority objective. Getting other survivors out is a bonus objective. So an absolute win feels like a 4e. As a soloq survivor, finishing the gate and getting any survivors out feels at least like a partial win.
Edit to add: In terms of how many games do survivors have a 3e+, I'd guess it's around 35% if averaged across the killers.
2 -
If you count Survivors as a team, forcing 3E/4E only for Survivor wins, then sets of 5 games comes out to 22% (with all permutations I could find), with 28% draws/2Es. If you throw out draws, you get ~31% winrate. Technically as you include more sets of games, you get a more and more clearer picture, and personally I'm sure you would close in to a ~40% winrate with law of large numbers pushing things that way (again, if you throw out draws).
Another factor to consider is the rate of DCs. BHVR throws out DCs in their stats, so if Killers DC at 25% the rate of Survivors it is fair and fine. However, if Killer DCs over 25% of Survivor DCs, then Kill rates are inflated artificially, or in other words lets just say 55% real instead of 60% fake (no math for this napkin 'sake of argument' estimate). Also, if the Killer DCs under 25% of Survivor DCs, then the opposite happens, and Kill rates are deflated artificially, or lets say 65% real kill rate instead of 60% fake (again, I didn't do the math for this napkin 'sake of argument' estimate). Personally (as in from my matches on both sides) I think Killer DCs are about 10% of Survivor DCs, so real kill rates are even higher than 60% by however much that actually shakes out to.
There also is an argument that a DC results in a 4K/4E regardless (depending on which side had the DC), so you need 100% Killer to Survivor DCs to break even (or even 60SDC-40KDC rates to match the intended kill rates to go further in the weeds). Regardless of the specifics, the Survivor DC rate IME is far far far greater than the Killer DC rate, which would mean higher kill rates aren't being counted as a result.
2 -
If you don't care about MMR at all, a win can be whatever you want, so the win rate is subjective.
If you care about that invisible number that is only useful for the backend of the game right now, survivors are not a team, and a win is only when a survivor is not killed. Ergo, if you escaped but your other 3 team mates are dead, you won and they lost.
There is an exception to this, as being in a SWF will affect the amount of MMR you gain or lose depending on whether your SWF mates also survive, but only the quantity. If you gain or lose that MMR will still be decided by you escaping or not:
In other words: MMR wise, Escape rate = Win rate.
1 -
I’ve had many survivors claim they won and beat me by getting hatch so I think that for most survivors as long as they escape it’s counts as a win for them.
I don’t care if a survivor gets hatch because I killed the other 3 which is a win for me and the hatch RNG favored them, there’s no skill in it. Either I stumble upon it and close it or they stumble upon it and escape.
And I couldn’t be bothered or learn the spawns as a 4k is honestly not that serious to me. But if I’m the only one who escapes I consider it a loss.
0 -
I agree a 3K is a win for the Killer. My question, though, is if a win for the survivors is defined as a 3E or 4E what percentage of the time do you think survivors win? All responses are appreciated.
0 -
I actually have data on this from my own games, which I recorded (can post the dataset on request).
Below stats only include games where I was a survivor myself (solo queue only):
Number of Games: 155
Survivor Win Rate: 31.6%
Survival Rate: 39.52%
Draw Rate: 11.6%
Rage Quit Rate: 21.3%What constitutes as a win is exactly what you described.
3-4 escaping = survivor win
2 escaping = draw
1 or less escaping = survivor loss—
I also have stats for my killers games too though, if interested:
Number of Games: 129
Killer Win Rate: 69.0%
Kill Rate: 71.3%Draw Rate: 11.6%
Rage Quit Rate: 17.83%You can do some math to get survivor win rates (19.4%) and survival rates (28.7) for these.
And no it's not a typo, the draw rate for both my survivor and killer games are the exact same, which I find both shocking and hilarious.
—
I do feel the game has the potential to be very survivor sided at the very highest level (SWF, co-ordinated, using strongest items/perks). But that's not the majority of games. In solo queue? Pretty killer sided, and I don't think there's much argument against that.
3 -
2 escape through exit gate is all that killer needs to consider the game not a win. Survivor's win condition is 2 Escape. 3 and 4 man escape is killer humiliation. Just extra salt on wound and extra T-bag ego.
0 -
2 is a tie/draw, which is separate from a null (which is Hatch, which throws out the results/has no impact on MMR). A tie is a equal win/loss for both sides. So if a Killer loses in a 2K, Survivors as a team lose in a 2E, and if a Killer wins in a 2K, Survivors as a team win in a 2E.
You can't just say draws count as Survivor wins since they aren't Killer wins. I mean I guess you could, but they you would have to remain logically consistent in saying draws are a Survivor loss, because it isn't a Killer loss. That would make it so 0E/1E/2E are Survivor losses, and 0K/1K/2K are Killer losses. Thus making 2K/2E the worst case scenario for everyone, since all players lose.
2 -
I know I'm talking to a wall...but survivor is literally unplayable without a SWF team . So it really doesn't matter ....third game in a row where the survivors are caught, downed and hooked before my gen is even halfway done. I'm happy to play killer. So far nobody has survived tonight. But it's sad when you can't play with a friend anymore. Bring back 2v8
2 -
Depends on the MMR. At low MMR I would say survivors probably have a 10-20% win rate going off of 3E criteria.
Average MMR probably sits around 20-30%High MMR I would guess is around 35%
I’d be remiss to point out that the devs want a 40% escape rate and not win rate. So that means that 1-2 survivors should escape every match with 1 being more common.
There are also way too many factors. If it’s Nurse every time then the win rate for survivors is probably in the 10% or lower range.
1 -
I don't believe in draws. the objective of survivor is to escape through exit gate. Killers have concept called win-streaks. You can define win-streak as 4k's only but I would define winstreak's as killer as 3k or 4k mainly because end game for killer is rng based unless killer runs Whispers and slug 3rd survivor to get 4th survivor. For a killer to not consider a win, At least 2 survivor must escape through gate. When a killer does not consider their match as win, they conversely believe the opposite result. A loss.
A draw would mean that neither survivor escape and killer did not complete their kill objective. It is a result that is not conclusive. How you would achieve this as survivor is hiding for 60 minutes until server closes. For killer, in order achieve this result, you have to defend a 3 gen for 60 minutes preventing the survivor from escaping. Both of these rare and almost never occur.
0 -
If you want fun statistics, then you can believe whatever number BHVR tells you, but it's not data based on real skill. Most people are doing challenges or dailies and/or using non-meta perks. Meaning: they're not playing for real. Now they may want to win, but they're not playing 100. Y'know? Most games should be thrown out of that data pool, but they're not. Like, all those people that consider this a casual game. Yeah, throw those games out because they're not playing for real.
I think the number BHVR is counting is a bit relaxed and isn't a true indication of how people would really play, unless there was a real reason to do it.
0 -
So, in what percentage of games would you estimate at least 3 survivors escape?
0 -
Whatever BHVR says. When my team plays for fun? Maybe 60 percent of the time. When we play for real, a lot higher than that. But we haven't done that in a long time because it's boring as #########.
I don't know how much survivors were escaping when I still played killer a couple years ago, but it low. I know that it would be very different if I went against my own team and they played for real.
0 -
A 3-man or 4-man escape. Personally, I just like to escape myself. I don't agree with this, "If I got good feelings, that's a certified win."
0 -
I mean when you play so fast and loose with definitions, I can make winstreaks include numerous prior matches, so as long as you win more than you lose, it still counts. "I have 100 kills and 50 escapes as Killer, so my winstreak stands unless I get 51 escapes in a row without a single kill." That is a 'valid' winstreak with this framework.
I just think when we have such a clear metric of Kill/Escape, the side with more of their desired result, wins. If neither side has more, then neither side wins, thus a draw.
2 -
I can make winstreaks include numerous prior matches, so as long as you win more than you lose, it still counts. "I have 100 kills and 50 escapes as Killer, so my winstreak stands unless I get 51 escapes in a row without a single kill.
No. That is called a Kill-rate.
I just think when we have such a clear metric of Kill/Escape, the side with more of their desired result, wins. If neither side has more, then neither side wins, thus a draw.
Every game has a final goal. a mission, an objective. I don't think dbd is about counting score. there is like counting score like…hooks but I would consider the game more about Escaping as team VS killer killing the team.
The only reason why MMR is not team based because it creates a problem called Elo hell. This is where a player takes longer to climb out of MMR ranks if they're good player because the team isn't as strong as the player. For example, you could be a master survivor player but your team is full of bronze players and killer is bronze player. If killer goes after all bronze players, you might not be able to win as survivor player because team aspect of the game. You will eventually climb out of Elo hell because you will eventually get bronze teammates half competent and your master skill will carry you out of bronze but it takes longer for system to place you in the correct spot. absolute team systems are better but they have a bit downside in unorganized ranking environments.
Tournament use absolute team win systems where as unorganized ranking environment tend to prefer solo ranking system. This also has deep philosophy type ideas like… which team is stronger, a team 4 super stars or team with 4 good player that work together to win. this should answer question for why MMR wins does not reflect the absolute win condition of the game.
0 -
According to Nightlight, the overall average of 0Ks and 1Ks is roughly 35.5%. These would necessarily be 3Es or 4Es, so that's the 'winrate' you'd be looking for, I think.
By contrast, the killer winrate in the same stat block is roughly 51.5%. (That's 3Ks and 4Ks)
5 -
You already established a winstreak can be using any definition someone wants, even unconnected to anything hard connected to the game's definitions, so if they want to have 'more kills than escapes net over a day/week/month' as their winstreak, its still a winstreak.
The game already has the metric of whether you win or lose, in the form of MMR. It used to be Emblems, what I believe to be a system better for the game health. If someone were to hypothetically datamine their MMR between each match, a '+' would be a win, and a '-' would be a loss. Since that type of behavior is not allowed, we make up 3K or 4K or net positive K/E for our 'wins', hoping the kills or escapes actually gave enough MMR to actually 'win'.
Elo Hell already exists when people are ratting the second things get heated, and aiming for Hatch/Wake Up+Sole Survivor plays. If you play with a baby Surv, them dying makes it far far far more likely for you to die as well. Team MMR (when using only Kill and Escape for MMR metrics) would actually help prevent elo hell, because people would know that ratting isn't winning, and they need to actually help the team to win. Emblems, again, I think are far more better at measuring independent impact on a match from everyone.
2 -
Elo Hell already exists when people are ratting the second things get heated, and aiming for Hatch/Wake Up+Sole Survivor plays.
Not really. If your way better than your opponent as survivor, you will solo carry/5 gen chase the killer. If your team dies due to incompetency, you can run sole survivor+wake up and escape through exit gate. it won't count as big MMR win but you won't lose mmr. on average, you will have positive escape-rate thus pushing you out of wrong MMR faster. In a team based system, you escaping through exit gate as solo survivor = loss.
if they want to have 'more kills than escapes net over a day/week/month' as their winstreak, its still a winstreak.
Yes you can make up your own win streak. What i am talking about is general perceived team win-rate for survivor compared to killer win-rate.
what counts as win is what killer perceives as loss. For most people that play killer, I would wager that most people playing killer would not consider 0 kill, 1 kill and 2 kill as winning. I mean you could make a poll that asks, "do you consider a 2k a win? Yes/No. ".. I am pretty sure that most killer would answer no.
0 -
You can only solo carry against Killers who left their brain in another pocket. An early Killer skill is when to drop chase, so your ability to solo carry is directly connected to whether or not the opposing Killer has learned this skill or not. The second you go against Killers who have learned this skill, they will gobble up all the baby Survs and you will still lose. Even then your solution to allied rats is to rat harder than them? I'm sorry but I don't think that is a good system.
"… general percieved team win-rate for survivor compared to killer win-rate." - People on the (team-based*) Survivor side of this equation would consider a 2E 'not a win' in the same way Killers would consider a 2K 'not a win'. *(If they were only worried about if a specific person escapes then that is a separate condition that we haven't been talking about thus far.) That just goes back to my previous issue of a 2K/2E is a loss for all players in the case where Draws/Ties don't exist. A win for one side doesn't force a loss on the other side when you ignore Draws/Ties. It just either upgrades or downgrades all Draws/Ties to both sides win or both sides lose.
In Magic the Gathering tourneys (last I played), they consider a win to be worth 3 points, a draw worth 1, and a loss worth 0. That means the game integrity can be compromised when it is better for 2 equally skilled players to take turns surrendering to each other, giving each other 3 points across 2 games instead of 2 points each with draws. You could consider a Draw a loss here, but both sides are losing by not getting the 3 points, not only one side. Same for DBD, if a 2K is a loss for Killer (and not a Draw), and a 2E is a loss for Survivor (because again Draws don't exist), both sides lose.
I guess what I'm trying to say is you are assuming a zero-sum game, where if one side wins the other side loses. Then you create the conditions where draws don't exist, which makes draws losses (or wins) for both sides. Then instead of checking did X side win, you check 1-Y side for side X's winrate. It would have been much simpler to ask 'did the Survs win', but you are instead deciding to take the roundabout method of checking and falsely assuming "If Killer loses, Survivor wins".
1 -
You can only solo carry against Killers who left their brain in another pocket. An early Killer skill is when to drop chase, so your ability to solo carry is directly connected to whether or not the opposing Killer has learned this skill or not. The second you go against Killers who have learned this skill, they will gobble up all the baby Survs and you will still lose. Even then your solution to allied rats is to rat harder than them? I'm sorry but I don't think that is a good system.
at low level of play, the killers are not skilled enough to make good in form tactical decisions on their gameplay because they lack skills to make those plays. that is why they're at low rank killer. Just the same reason why survivors are low rank because… they lack certain skills to play better to escape their opponent. that is why when your clearly better survivor than the killer, it isn't that hard to defeat the killer however the team elements of the game require your teammates to have certain level competency for the team as a whole to escape. Sometimes that competency is not really there. that is why there is this hatch & exit gate stuff in end game for free escapes. these systems allow you to get out of poor mmr faster IF you are significantly better than your current MMR bracket suggests. In a team system, you would still get out of low MMR if your high MMR, but it would take longer.
People on the (team-based*) Survivor side of this equation would consider a 2E 'not a win' in the same way Killers would consider a 2K 'not a win'
Why would they not consider it a win? If killer agree that he didn't win then you must won. Some player won't admit they ever lose and saying that they're drew is a way of softing the blow. in my opinion, draw concept probably comes from literal MMR concept that dev are using for matching which is the killer kills two and 2 survivors escape, say each survivor = 10 mmr. You'd gain 20 MMR but lose 20 MMR = 0 MMR. They're counting the final MMR gain or Final MMR loss as win or loss. The number for going up or down is just a number to understand what level of player the matchmaker looks to pair with one another. It is matchmaking tool, not a win or loss condition. The win or loss condition is what game is about.
I guess what I'm trying to say is you are assuming a zero-sum game
All games follow zero-sum game. The goal of magic of the gathering is to deplete enemy life pool. to my knowledge, they have 20 health and when your health goes to 0. you lose. the other side wins although magic of the gathering also has bunch of alternative win conditions but let's stick with basic. that is zero-sum rule for magic.
You could consider a Draw a loss here, but both sides are losing by not getting the 3 points, not only one side. Same for DBD, if a 2K is a loss for Killer (and not a Draw), and a 2E is a loss for Survivor (because again Draws don't exist), both sides lose.
Dbd has scoring but it is only related to hooks. Hook in my opinion would be tie-break type statistic. it would be only useful if holding tournament.
For example if you have 6 teams and 8 killers. each killer is forced to play wraith. You want figure out which of 8 killer's is best wraith player vs these 6 teams and which team is best survivor team to facing wraith. Let's say each killer play each team 2 times. This means that every player would have 12 games. Suppose one killer has 5 wins and 7 losses and another killer also has 5 wins and 7 losses. when placing killer player in standing, you would look at their hook count difference. the killer with more hooks would be placed higher in standing. It would be the same for survivor teams where the team with least hooks with same win/loss ratio would be placed higher. If both teams have exact same stats, then you would look at total hook stages, than if it is still the same then maybe tiebreaker match. this is how I would go about it.
killer would always be compared to other killer players and survivors would always be compared to other survivor teams.
0 -
If I were to sum up the end of your first paragraph's point, it is that you can still win with bad teammates, but it may take longer. That is exactly my point with team-based MMR/Elo Hell, you can still win with bad teammates, and it may take longer, but you will rise. That would be better for the game than individual MMR, and Emblems would be better than both of those.
Also to address the first sentence, there are many facets to Killer that contribute to your kill rate. There are macro skills, like which gens to protect or sacrifice, which pallets to chase a weak link into in order to create a deadzone, and how long to chase each person for pallet drops or map pressure. There are also micro skills, like I know I get the lunge at the window here, or with Coup de Grace, or I get the hit with a Bamboozle window hop, or I get a max charge Demo lunge hit here, etc. The combination of both micro and macro skills determines someone's overall skill, or overall kill rate. You could go against a god Demo, but they have a babies understanding of macro, thus will chase anyone anywhere. You could go against a god macro gamer, but sucks at chase, and keeps lunging when they can't hit, but still gets pallet drops and prevents gens from completing and wins through attrition. My point was it requires just enough macro skill in one category to nullify solo hard carrying, and you can solo win against anyone that lacks that skill. You can't solo hard carry against anyone with that skill high enough though, even if they are garbage at many other facets of the game.
Why would Survs not consider a 2E a win? - Maybe because they consider it a tie. Maybe they use your definition of needing to have their side have more sub-wins than sub-losses (aka 3E is the minimum for a Survivor win). Maybe they only count their personal sub-win/loss, or in other words if the singular person escaped or got killed, and they were among the 2 people killed in a 2E. All of these are normal points of view that would mean 2Es are 'not a win', or a loss when draws don't exist.
All games are zero-sum games - Back when we used Emblem based matchmaking, you could have multiple people (even on opposite sides) 'winning'/'pipping', or multiple 'losing'/'depipping'. So that already is contradicted without checking other games. Even then, multiplayer magic isn't zero-sum, because in a 4-way free for all, there is 1 winner and 3 losers. Zero-sum would need an equal number of winners and losers. Heck, the game is asymmetrical to start with, so you can only make each individual Survivor compared against Killer be zero-sum. In a 3K and a gate escape, the Gate Escape person 'won' against the Killer, even if the Killer 'won' against the other 3 Survs. We have been mixing up using independent and team based 'wins' and 'losses' at different points of each of our posts, so hopefully we haven't been speaking past each other assuming independent when the other mean team, and vice-versa.
Hooks are better for granular measures - I agree here. I would much prefer the game be based on 8 hooks (and 0K) being the tie/draw condition, or even saying if that happens before all gens are popped, then it counts as a Killer win. I would also consider 9+ hooks post the final gen pop a Killer win (when Survivors are also playing under that knowledge, and don't go for endgame rescues for funsies). I would also consider camped/death on first hook 2 extra hook states as lesser, but I'm not sure by how. Although a 3K should (probably) be a guaranteed Killer win, regardless of hook states.
0 -
My escape rate as a survivor is 2 out of 10 games, honestly I think what makes this game bad is the tunneling, stupid map design and matchmaking (I'm a survivor with 2k hours of matchmaking by people who just installed dbd 1-3 months ago)
0 -
started tracking my games after seeing some users do the same, seemed interesting.
Tonight, I played 14 games, escaped 5 times. As a team, we won 6 games (3 out). I want to do a fairly large sample size but I expect the end results to be more or less the same since these 14 games felt just about right for an average dbd session
0 -
I do personally think that game events should be implemented into MMR calculations.
I've said for years that the 1vs1 chess ELO system that most MMR systems are based on is far too simple for games with a complex match making system like League of Legends or DBD.
These are games where your team can cost you a game through no fault of your own, but also a game where your teammates can carry you with no input from yourself at all. Simply boiling it down to escape vs. not escape is too simple.
That said an overly complex system based on scores creates issues with farming throwing off skill caluculations.
Boring Number part
If it were me the 2 simplest and fairest metrics to weigh MMR skills of players on are gens completed and hooks. E.g. with some example guesstimate numbers:
Killers:
- Kills grant MMR of say +2 per kill, -2 per gate escape.
- Hooks grant MMR of say +1.5 per hook and -1.5 per missed hook (hooks not hook stages).
- Gens grant MMR of say +1 per gen not repaired, -1 for each repair gen.
Then obviously apply a multiplier based on the skill difference vs. The Survivor team average.
This would mean killers who camp and tunnel players out are less likely to get all 12 hooks, and allow multiple gens to go by since they are not applying pressure to other members of the team. Some example results:
- Killer 0k (-8) with 2 hooks (3 - 15) and 5 gens completed (5) = -25
- Killer 0k (-8) with 8 hooks (12 - 6) and 5 gens completed (-5) = -7
- Killer 1k (2 - 6) with 8 hooks (12 - 6) and 5 gens completed (-5) = -3
- Killer 2k (4 - 4) with 6 hooks (9 - 9) and 5 gens completed (-5) = -5
- Killer 2k (4 - 4) with 9 hooks (13.5 - 4.5) and 5 gens completed (-5) = +4
- Killer 3k (6 - 2) with 7 hooks (10.5 - 7.5) and 4 gens completed (1 - 4) = -2
- Killer 3k (6 - 2) with 11 hooks (16.5 - 1.5) and 5 gens completed (-5) = +14
- Killer 4k (+12) with 0 hooks (-18) and 1 gens completed (4 - 1) = -3 (slugging is bad mmmkay)
- Killer 4k (+12) with 6 hooks (9 - 9) and 3 gens completed (2 - 3) = +11
- Killer 4k (+12) with 12 hooks (+18) and 4 gens completed (1 - 4) = +26
Survivors:
Survivors are little more tricky as they have to be calculated when the survivor dies/escapes. Since they may not be in the game as long with fewer "events" to track, they need bigger numbers...
- Escapes grant MMR of say -7 per death, +7 per gate escape.
- At point of death/escape, allies not on hook/dying/escaped +2, allies dead/dying -2.
- At point of death/escape, gens completed grant +3, gens not completed grant -3.
Again apply multiplier based on the difference between Killer and own MMR.
Survivors meanwhile enjoy bonuses based on their contribution to the team, and taking chase from the killer can actually result in net gains of MMR. There is also some incentive to not be the last man standing. Some example results:
- Survivor death (-7) with 0 allies alive (-6) and 1 gen done (3 - 12) = -22
- Survivor death (-7) with 1 ally alive (2 - 4) and 0 gens done (-15) = -24
- Suvivor death (-7) with 1 ally alive (2 - 4) and 4 gens done (12 -3) = 0
- Survivor death (-7) with 3 allies alive (+6) and 0 gens done (-15) = -16 (yes you SoH players suck)
- Survivor death (-7) with 3 allies alive (+6) and 3 gens done (9 - 6) = +2
- Survivor death (-7) with 3 allies alive (+6) and 5 gens done (+15) = +14
- Survivor escape (+7) with 0 allies alive (-6) and 2 gens done (6 - 9) = -2
- Suvivor escape (+7) with 0 allies alive (-6) and 5 gens done (+15) = +16
- Survivor escape (+7) with 2 allies alive (4 - 2) and 5 gens done (+15) = +23
This looks like it would work a lot better than just Kills vs. Escapes. There is the caveat that a survivor can dip before their ally is downed to get more MMR, bit that would be rare enough to not really matter (and it's only worth 2 points, gens repaired has the bigger effect).
Post edited by UndeddJester on0
