Let's discuss "Kill Rates" vs "Win Rates"

Nebula
Nebula Member Posts: 1,396

I see this brought up so often whenever kill rate is brought up, with many pointing out that kill rate doesn't equal win rate, and while yes, that's true, I don't think a lot of people actually understand what this means or look further into it. I wanted to clarify this topic that I see quite often when discussing statistics in this game.

Here's a breakdown of the actual statistics at play based off the stats we've received.

Now without getting too in depth about combinatorics, here is what the "win rate" for killer currently looks like based off the 60% kill rate benchmark that has been set by the devs:

  1. 0 kills (loss for killer): Probability = (1 - Survivor escape rate)^4 = (1 - 0.4)^4 = 0.1296
  2. 1 kill (loss for killer): Probability = 4 * (1 - Survivor escape rate)^3 * Kill rate = 4 * (1 - 0.4)^3 * 0.6 = 0.3456
  3. 2 kills (draw): Probability = 6 * (1 - Survivor escape rate)^2 * (Kill rate)^2 = 6 * (1 - 0.4)^2 * (0.6)^2 = 0.2304
  4. 3 kills (win for killer): Probability = 4 * (1 - Survivor escape rate) * (Kill rate)^3 = 4 * (1 - 0.4) * (0.6)^3 = 0.3456
  5. 4 kills (win for killer): Probability = (Kill rate)^4 = (0.6)^4 = 0.1296

Based off this data, currently killers are seeing a "win rate" of 0.4752 or 47.52%. This is where I feel so many people get lost. They see this win rate of less than 50% and assume that the games still not balanced for killers because true balance would be 50/50.

Draws are often forgotten and possibly purposefully overlooked in these scenarios.

The current outlook for killer currently has them at a 70.56(!)% chance that they achieve a 2k or better in a match. This data for escape rates and kill rates were taken over an average of hundreds of thousands of matches. Sure, it's not fully representative of what will actually happen in your given match, that's the nature of statistics, but this is what the entirety of the games outlook currently is.

Feel how you want about draws, but to completely invalidate them as a meaningful statistical outcome is misleading to this discussion. 2 kills, 2 escapes is a completely respectable match outcome for both sides and for it to be thrown out of the discussion for personal feeling is arbitrary.

TLDR: Killer currently has a 70.56% chance of achieving a 2k or higher in the average match.

Comments

  • xltechno
    xltechno Member Posts: 1,026

    I feel that the reason for the discrepancy between win rate and kill rate is that Survivor's winning pattern is often 3 people surviving, whereas Killer's winning pattern is simply 4k. There are very few patterns in which only one survivor escapes.

  • Firellius
    Firellius Member Posts: 4,296

    Sadly, the game is intrinsically inclined away from 2Ks, so chasing that is a pretty futile effort.

  • Bafugaboo
    Bafugaboo Member Posts: 406

    Teammates leaving you to die on first hook is a huge factor as well. 8 games in a row if you wonder.

  • Saiph
    Saiph Member Posts: 324

    This is a false debate. Killrate is the only objective data we have, and it is at 60%.

    Winrate is subjective, and prone to manipulation, because it depends of your definition of a win. But by association with other games, people often use the term "winrate" as "killrate", and it's a fine approximation, because if you define 4k as 1 win, 0k as a loss, and anything between as a fraction of a win (2k = 0.5 win for example), then the winrate becomes exactly 60%.

    However, some killer mains with very questionable mathematical honesty, have realized that if you twist the definition of a win to consider that 2k as "0 win" (instead of 0.5 win which would be much more natural), then you can pretend that the killer winrate is actually only around ~50%, and the game is "balanced". This is complete nonsense, as even with this definition the survivor winrate is around ~35% so still very far from balanced. The only difference is that instead of saying the winrate is 60-40, you say it's 50-35 but with ~15% matches where neither side wins. If you ignore these matches, then magic, the winrate becomes 60% for killer and 40% for survivor, again.

  • ChaosWam
    ChaosWam Member Posts: 1,805

    I really think 'winning' is something the community has agreed upon more than the developers, looking at how they balance the game, their released stats revolve around kills and not wins.

    I mean, escaping is a personal goal for survivors. Getting other survivors out isn't a scoring event, just bodyblocking is. There's no real incentive for sacrificing yourself to save your team, and there's no incentive to really assist your teammates aside possibly making some extra BP for yourself, or giving the killer other targets except you (as bad as that sounds.) Really it's up to survivors if it's worth helping your team, and naturally that extends to making SWF groups.

    And killers get praised by the entity for more kills, sure, but there's no indicator saying "Congrats, you won!" because you took out 3 or 4 survivors. It's very straight forward about 'There are 4 targets, kill some" and not much else more. It's also why a coordinated SWF is going to survive more often because they do have the benefit of cooperating to deny the killer as many hooks as possible.

  • Raptorrotas
    Raptorrotas Member Posts: 3,238
    edited March 15

    I think you made a mistake in your math. EDIT: or wrote it down wrong, which can happen easily.

    1. For 1 Kill its: 4 * (1-SR)^3 * SR = 4 * (1 - 0.4)^3 * 0.4

    You keep using the same rate twice in the mixed outcomes of 1k, 2k, 3k.

  • Spare_Them_Mori_Me
    Spare_Them_Mori_Me Member Posts: 1,478

    Killers winning platform is 3 or 4k, not just 4k.

    Survivors win if 3-4 get out.

    2 is a draw in either scenario. This is how I believe it has always been, buuuuut I'm not 100% confident atm lol.


    Anyway, 3k is often 'not good enough' or 'Not a win' to a lot of players, and thats totally fine! Holding oneself to 4k/4esc or I/we lose!' just makes things more stressful for whoever has that mindset. I can't imagine the frustration this game gives in that mindset. Dbd is a VERY frustrating game lol.

  • Aven_Fallen
    Aven_Fallen Member Posts: 16,106

    But this is just the way DBD plays out. It is basically a game of snowballing.

    If the Survivors do the Gens fast enough before the Killer can actually build pressure, you will most likely end up with a 4 man escape, maybe a 1K if the Killer gets someone in Endgame.

    And if the game turns into an early 3v1 (tunneling, Hook suicide, whatever), the game is guaranteed to be a 4K, unless the remaining 3 Survivors are far better than the Killer (obviously I count 3K + Hatch the same as a 4K, because in the end a 3K + Hatch would have been a 4K if the Killer would have slugged or the Survivor did not get lucky to find the Hatch first).

    But obviously, this does not mean that Draws should be ignored.

  • Lynxx
    Lynxx Member Posts: 510
    edited March 15

    Well, you know, there are five people in a match and we all see our own situation. So the 60% chance of not making it to endgame for most of the players takes forefront over 'win rate' cause the only way survivors win in the matchmaking is if they escape and that's four players affected..

    It wasn't so long ago that devs considered 2 kills - 2 escapes to be the perfect outcome. I don't know what changed.

  • Raptorrotas
    Raptorrotas Member Posts: 3,238
    edited March 15

    I'd say I see more survivors asking for the "fair" 50/50 they call the 50% escape chance.


    Quickly opened a graphic calculator webpage and wrote this down:

    • x = escape chance (the chance for a survivor to win)
    • (1-x)^4) + 4 * (1-x)^3 * x = chance for 4k and 3k combined (the chance for a killer to win)
    • 6 (1-x)^2 * x^2 = chance to 2k (the chance for the k
    • x^4 + 4 x^(3) * (1-x) = Chance for 0k and 1k combined (commonly refered to as survivor "team" win)

    With a 40% escape chance we get:

    • 47.25% chance for a killer win
    • 34.56% chance for a draw
    • 17.92% for a survivor "team" win
    • The Killer has a 1.188 times higher chance to win than each individual Survivor.

    With a 50% escape chance we get:

    • 31.25% chance for a killer win
    • 37.5% chance for a draw
    • 31.25% for a survivor "team" win
    • Each individual survivor has a 1.6 times higher chance to win than the killer.
    • Yes, this is a "fair" 50/50 or like 31.25/31.25 for the survivor "team" and the killer.


  • Nebula
    Nebula Member Posts: 1,396

    Thank you for this input and double checking my math!

    I believe the discrepancy comes from a different interpretation of survivors escape rates, mine coming from a 40% chance of an individual escaping, while yours comes from a 40% chance of the team escaping.

    An entirely valid way of interpreting the statistics as I'm not 100% sure if the devs clarified or not whether the 40% survival escape rate was for an individual or a team. If it is indeed a team escape rate, your version is 100% the correct way of interpreting the statistics. Thank you again for this!

  • xEa
    xEa Member Posts: 4,105

    This might not be a real discussion but rather the one side trying to instruct the other


    Lets go 🙄


    Step 1: Understand that it is not about "winrate" but rather win/loss ratio. It may sound aweful when there is only 40% winrate, but it is perfectly fine, when lossrate is 40% and drawrate is 20%. Thats ballanced.

    Step 2: There is a draw for killer. That part is always the most confusing to me, that some people somehow dont get this. The killer draw or tie is 2K. 0-1k = loss, 2k = draw, 3-4k = win. If you think a 2K is not a draw, then this is your personal take on how you define a win or draw and has nothing to do with the topic really. Its then more for "2K feels like a loss/win" topic.

    Step 3: Survivor dont play in a team, they play individually. They may (or may not) have the same goal of escaping, but if teammates escape or not is not relevant for their own statistical win/loss ratio. Neither in end outcome, BP, PIP or MMR. While in other games like CS you win or loose with your team, in DBD you simply dont.


    50% killrate is in fact not 50% winrate, because of the nature of 4 survivors and 5 different outcome.

    Thank you NL for helping out on the illustration. We pick ANY killer of any time with a certaint winrate. There is almost never any difference, so this is not a "special example"


    60% killrate translates into a very unballanced Win/Loss ratio. The killer wins 56% of the time and only looses 30%. Survivor simply loose 60% and only win 40%. Individually of course (Step 3).


    50% killrate is an almost ballanced Win/Loss ratio. The killer wins 41% of the time and looses 43%. Survivor win 50% and loose 50%. The outcome is ballanced.


    The only question is: Why are they doing it? And the answer is clear: Killers are DBD figureheads, and they are supposed to be threats. At least this is the reason. If i think this is good or not has nothing to do with the discussion here, so i leave it be.

    Questions/Objections? Just shoot.

  • Ryuhi
    Ryuhi Member Posts: 3,736
    edited March 15

    I'm curious where the survivor "team" mentality comes from in these discussions. If it looks like a duck (survivor dies) and it quacks like a duck ( MMR goes down when killed) then it might just be a duck (that survivor lost.) Each survivor has an individual outcome of win/lose/draw (if you count hatch as a draw, otherwise win/lose if simplified to ignore the MMR stipulation) but the killer has 5 possible outcomes (specified since lose/draw/win have a cadence of 2/1/2) and its possible for both a survivor and the killer to win simultaneously (3K/1E) as well as the killer and one survivor to lose (1K/3E.)

    I don't consider it a zero sum game, based on how the devs have structured MMR. It technically wasn't zero sum when emblems were the determining factor either, since pips could be awarded or removed on both sides. The only way it works as a zero sum game is if you consider the survivors a team and ignore both MMR and PIPs, which is why I'm so curious as to why its so many people's default view.

  • Saiph
    Saiph Member Posts: 324


    Sorry but that's completely wrong. The binomial formula works if X is an independent event which is absolutely not the case here. In DBD the chance of survivor 1 escaping depends heavily on survivors 2, 3 and 4 escaping too. Your calculator won't tell you anything for that case.

  • Raptorrotas
    Raptorrotas Member Posts: 3,238

    Wait no, my math is based on the individual survivor escape chance. You may need to read it again.

    • escape chance x
    • four survivors escaping x^4
    • 3 survivors escaping 4 * x^3 * (1-x)
  • Raptorrotas
    Raptorrotas Member Posts: 3,238
    edited March 15

    Huh, neat. To be fair it's been atleast 10years since I had anything to do with stochastics.

    Though I'd say thats to be expected when my numbers are based on a "chance" thats been derived from the "rate" the devs showed us. I'm aware that 40% escape rate doesnt mean you get a 40% chance to escape when you boot up dbd, but that after collecting tons of data there are ~40 survivors escaping for every ~100 survivors that play...

    EDIT: Just read it as the imaginary average escape chance x derived from the escape rates A B C D of the survivor participants a b c d. (I hope this is as horrible a joke as I think it is)

  • Raptorrotas
    Raptorrotas Member Posts: 3,238

    I've been beaten like a dead horse in another thread because I said killers cant "draw" in a 2k/2e cuz theres no singular team that also "draws". Theres nothing indicating survivors being a team, like you said, the 1e in a 3k still wins and the 1k in a 3k doesnt get points for the others escaping. I'm rather ... bruised ... and too tired of this topic to further elaborate, as I dont understand those people either.

    as for the hatch being a draw? it clearly gives the escape points *facepalm*

  • Shuma
    Shuma Member Posts: 55

    Your calculations are incorrect.

    Take your example of Unknown.

    Survivors have 16.77 4es and 14.19 3es. Add those together for a total of 30.96.

    killers have 35.46 4ks and 20.67 3ks for a total of 56.13.

    You then both of those numbers together for a total of 87.09. You get the killer win rate by dividing 56.13/87.09. The win rate for killer is 64.45%.

    for the survivor team, you divide 30.96/87.09 for a survivor team win rate of 35.54.

    You need to remember that draws aren’t counted so you don’t use the 12.91% in the calculation.

  • xEa
    xEa Member Posts: 4,105
    edited March 15

    These are not my calculations, these are statistics from real examples.

  • Spare_Them_Mori_Me
    Spare_Them_Mori_Me Member Posts: 1,478
    edited March 15

    The depth people get into this baffles me. People can set their own win/loss conditions, but they're just personal conditions. BHVR has their method. It sucks, but its there. I honestly wish they would change it, evolve the game a bit, and see how many problems melt away.

    I truly feel their 'Kills vs escapes' system is what most of the issues players complain about arises from. Might be weird, but I think it would even fix tunneling lol.

  • TheSubstitute
    TheSubstitute Member Posts: 2,438

    I agree because it's too simplistic and people did have other 'win' conditions that didn't lead to tunneling and camping. As an example, when BBQ tied its BP bonus to hooking each survivors once some Killers felt their personal win condition was hooking each survivor at least once. The Killer might see a 4E but still felt like they won because they fulfilled their personal win condition.

    That created less tunneling and camping because their goal was different than get a 4K by whatever means. It wasn't by any means the majority of Killers but it was enough that there was less tunneling than now.

    Incentives will modify behaviour.

  • Shuma
    Shuma Member Posts: 55

    Those are the kill rates, you have to do calculations with those stats to get win-rates. Draws aren’t counted in win-rates, so you have to remove those to get the win/loss rates.

  • xEa
    xEa Member Posts: 4,105

    We are talking about win/loss ratio here? then that is true, but since so many refere to "winrate", i did not put that into further account. Either way how we see it - winrate or killrate, it is all clearly in the killers favour.

  • crogers271
    crogers271 Member Posts: 1,741

    Okay, the numbers don't make any sense with the other sources of data we have.

    Nightlight currently has data for ~45k games over the last 28 days with the most common outcome being a 4k, then a 0k, then 1k and 3k pretty close together, with 2k being the rarest. This I think is rather uncontroversial that the ends of the spectrum are the most common occurrence, but you have them as the least common occurence.

    Basically, I don't get why you think those formulas apply to DbD.

  • Blueberry
    Blueberry Member Posts: 13,590
    edited March 18

    For SBMM reasons the two terms are basically synonymous. To me and an aim for objective balance though, they’re not. The reason people discredit kill stats is because there are a multitude of different things that lead to kills in DBD that have absolutely nothing to do with skill. The “with skill” part matters because the point of measuring these kills is for balancing matches based on skill. For just one example, a killer may get 3 kills in a match but that doesn’t tell me he played well. If a killer gets 8 hooks in a match though with zero kills, that does tell me he played well. However the kills would say he lost. While still flawed, hooks would be a significantly better criteria for measuring wins or game balance than hooks. As long as we’re measuring by kills, “wins” is a flawed term to use, even though I do understand that’s what they’re currently using, unfortunately.

  • C3Tooth
    C3Tooth Member Posts: 8,266

    2K count as lose on killer but 2 wins as survivor is a weird take for me.

    First, if count survivors as a team:

    • 1K is a lose for killer. Win for survivors, means that one survivor still have to see that match as winning.
    • 2K is a tie, side A (survivors) and side B (killer) have the same score: 2
    • 3K is a win for killer. Lose for survivors, that one able to escape doesnt matter, because they lose.


    Second, if count survivors escape as individual wins:

    • Those 2 survivors escaped still have their average win rate of 40% in general. Which still lose more than win.


    The math counting "there are 4 survivors so their win rate is multiple by 4" is flawed in both the math and logic itself. Why should I see Dwight and Meg escaped as 2 wins for me while I died?

  • Coffeecrashing
    Coffeecrashing Member Posts: 3,694

    1) Survivor wins count as individual wins.

    2) This means the survivor escape rate and survivor win rate are both 40%.

    3) 40% is the intended escape rate, that BHVR directly told us they want the game to have.

  • C3Tooth
    C3Tooth Member Posts: 8,266

    Then 60% kill rate is intended by Devs, why complain 60% kill rate is not 60% win rate?

    Suddenly 40% win rate matter as long as its on killer side?

  • Coffeecrashing
    Coffeecrashing Member Posts: 3,694

    60% kill rate isn’t a 60% win rate, because killers win on a 3k or 4k.

    In a perfect world, survivors would have a win condition that was team based. But we don’t live in a perfect world.

  • Spare_Them_Mori_Me
    Spare_Them_Mori_Me Member Posts: 1,478
    edited March 18

    Indeed. In a perfect world this would still be a party game and not some competitive cesspool like most games become. Dbd sort of sailed right in while waving to the toxicity.

    Hopefully BHVR remodels some things and the game can become a game again.

  • canonjack001
    canonjack001 Applicant Posts: 67

    Not good at math, but I am pretty sure 0k is 0.4^4 instead of (1 - 0.4)^4😂

  • canonjack001
    canonjack001 Applicant Posts: 67

    and the sum of all event should be 1😂

  • RpTheHotrod
    RpTheHotrod Member Posts: 1,863

    One thing to factor in is that the draw thing is ONLY for mmr adjustment. It's not related to wins or losses. MMR adjustments are designed to try to make future matches more fair - that's it. It's not a reward or penalty for winning or losing. Some games will have you lose MMR if you win, or gain MMR even if you lose. With that in mind, the devs stated that this is not a team vs team game. It's a 1 player vs 4 other individual players each with their own win condition. Survivors win by escaping and lose by not escaping. That means that you could have 3 winners and 2 losers in an example of 3 survivors escaping (killer and the 4th survivor lost). Essentially. The game is designed around the concept of killer vs 1 survivor situation for every survivor there is. Due to this, there is no actual draw that's possible. A survivor either wins or loses. The killer either wins or loses. A draw means neither side wins or loses, and that is impossible in a 1 v 1 (x4) situation in dbd. When it comes to MMR adjustments, 2 survivors escaping counts as MMR neutral, but wins and losses are at a per survivor/killer level. It's one killer vs 1-1-1-1, not a killer vs survivor team. Weird, I know, but that's how it's designed.