Calculating wasted time/pulls because of people with low survi

[kaminaris]

Nah
You aren’t looking for math skills
You are looking for validation

No, i am looking purely for math skills. Nothing else interests me.

[Schattenlied]

No, i am looking purely for math skills. Nothing else interests me.

That's clearly false, otherwise you'd just go find a better guild.

- - - Updated - - -

Not sure who you people raid with but when I raid with mature and logical people.

If the people you raid with are oh so mature and logical you shouldn't need "math" to convince them they are dying too much.

On the other hand if you dislike anyone who points your mistakes then well...

There's a difference between "we're wiping too much" and 'I WENT ON MMO-C AND FOUND SOMEONE TO MAKE AN ALGORITHM TO DETERMINE WE'RE DYING TOO MUCH!", the former is reasonable, the latter makes you look like a micromanaging [word I won't say here].

Maybe it was mistake thinking I find anyone with math skills here.

Plenty of people here are good at math, your mistake was thinking anyone here is interested in assisting you in your quest for validation... Go do it yourself.

[kaminaris]

That's clearly false, otherwise you'd just go find a better guild.

As I mentioned before, you have full right to think whatever you want, and I have full right to not care about your opinions.

Now getting back to topic.

I am aware in flaw in my algorithm, namely in function that calculates chance of wipe on one specific encounter.
penalties are being applied only on next players in queue. So if algorithm finds that last player in array dies, no penalty will be added.
It also means that order of players in a table/array actually changes the result.

Example 1 (normal order) - 10 000 iterations:

Average Run Time: 166.8755
Average Wipes: 6.526

Now I reordered players to be from best to worst:

Average Run Time: 159.93885
Average Wipes: 5.7751

Now order is from worst to best:

Average Run Time: 197.9096
Average Wipes: 9.9018

So as you can see, there is a massive difference between results based soely on player order. And this is something I can't wrap my head around.

Next issue would be actually calculating how many pulls more we will need compared to better guilds. I am still thinking about how to make it.
For example boss X takes 100 pulls for better guilds, if we have 3% less survi and Z% more failure rate how is that going to affect our number of pulls?

[pappaslop]

The problem is you cannot simply calculate that amount IF your raiders would shape up and stop dying, such data doesn't exist.
So your current team vs THE SAME team but with higher average survival rate.

It is simple.

You are making it complicated.

RECRUIT TO REPLACE.

Problem solved.

PS - you're not top 5 world (nor am I) - none of it matters.

[kaminaris]

I tried this variation:

function simulateEncounter(encounter) {
	let dpsDeaths = 0;
	let healerDeaths = 0;
	let tankDeaths = 0;
	let deaths = 0;

	for (const p of playerTable) {
		// if one dude die
		let realSurvi = p.survi;

		// player died
		if (roll() > realSurvi) {
			// if it was a healer
			switch (p.role) {
				case 'dps': dpsDeaths++; break;
				case 'healer': healerDeaths++; break;
				case 'tank': tankDeaths++; break;
			}

			deaths++;
		}
	}

	// additional chance of wipe
	const wipeChance = (dpsDeaths * encounter.dpsPenalty)
	+ (healerDeaths * encounter.healerPenalty)
	+ (tankDeaths * encounter.tankPenalty);

	if (roll() <= wipeChance) {
		return false;
	}

	return deaths < encounter.wipeCutOff;
}

But it does not seem to be close to reality. Additional chance of wipe is not relevant enough.

Here is full version, https://pastebin.com/701DEACa

According to this, we would be loosing around 30 minutes on 5 bosses.

[ShmooDude]

No, i am looking purely for math skills. Nothing else interests me.

Survival percentages aren't really that great in the first place (imo). Not how they're calculated on Warcraft Logs anyhow. They're based on the total encounter time. If all the encounter lengths were equal it wouldn't be a problem but they're not.

Say player A dies 45 seconds in to a 1 minute wipe. He has a 75% survival percentage for that wipe. Player B dies 45 seconds in to a 5 minute wipe. He has 15% survival percentage for that wipe. Warcraft logs then averages these percentages together with no consideration for the duration they were calculated off of. Lets say that there were 8 wipes total and these were their only early deaths (so 100% for the other 7 wipes). Player A has a 96.9% survivability rating. Player B has a 89.4% survivability rating. Player A's survivability rating is considered good while B's is bad, despite their making the same mistake (presumably).

Basically, a player's personal survivability metric is worse if the rest of the raid can (and does) continue on without you. The irony is, the more "important" a player is to the raid encounter (read probability of calling a wipe after their death), the better their survivability score will be despite those deaths. That's not to say these will always happen, but it throws enough uncertainty into the system to make me question it's validity as a metric if you don't have consistent pull times.

---

I'm not sure if this is really along the lines of what you're looking for but if you're any good at mining info off WCL, a more practical way to go about calculating "wasted" time would be to ignore the survivability percentage completely. In addition to what I wrote above, survivability already is a summary of a time metric so using it as an intermediary for a new calculated time metric isn't all that useful.

Instead, you could pull the following information from each wipe:
Start of pull
Players in the Raid
The first X deaths
End of pull
Wipe or Kill

Keep a running total of time spent by each player is in the raid (to compare to how much time they contribute to wasted time)

For each wipe, determine how long till the start of the next pull (to get an accurate wasted time metric)
Apply that amount of time to a running total of wasted time (for the entire raid as a whole).
Also track a running total of wasted time by player.

So quick and dirty example.
Wipe 1 7:00 - 7:05
Players A - U in the raid
First deaths Players A and B

Wipe 2 7:08 - 7:10
Players A - U in the raid
First deaths Players B and C

Kill 7:13 - 7:20
Players A - U in the raid

Players A - U all get 20 minutes of raid time (7:00 Wipe 1 start -> 7:20 Kill end)
Total wasted time gets 13 minutes (7:00 Wipe 1 start -> 7:08 Wipe 2 start and 7:08 Wipe 2 start -> 7:13 Kill start)
Player A gets 8 minutes of wasted time (7:00 Wipe 1 start -> 7:08 Wipe 2 start)
Player B gets 13 minutes of wasted time (7:00 Wipe 1 start -> 7:08 Wipe 2 start and 7:08 Wipe 2 start -> 7:13 Kill start)
Player C gets 5 minutes of wasted time (7:08 Wipe 2 start -> 7:13 Kill start).

It's not going to be perfect since first deaths aren't the end all and be all of what caused a wipe. It should show trends though and if certain people are dying first all the time, they should have significantly higher calculations of wasted time. You could also consider weighting the deaths. Say apply 2/3 of the time to the first and 1/3 to the second or whatever you feel is appropriate.

Anyhow, just a thought.

[Revamp Man]

The whole point of raiding is to get content done. The majority of people aren't actually friends unless they were before they started raiding together. At the end of the day all the majority of guilds care about is killing content within the allotted time given to them. Why would the majority of players not want to improve?

Because it’s not actually about improving it’s about what someone sees as an improvement who is to say why little Timmy dies on the transition phase in Rod Dan every two Pulls or so

Maybe it’s because he fucked up when it comes to the mechanic or maybe it’s because he’s colorblind

Like you said a majority of Raiders are not friends so if one raider pisses off the majority of the group I’m going to kick your ass to the curb because I cannot as a Raid leader progress into Raid if I have six people pissed off that someone comes in trying to act like a professional it would be much easier for me to kick the one dip shit and replace him with people from the group

- - - Updated - - -

No, i am looking purely for math skills. Nothing else interests me.

Then you should not have come to this forum because you don’t care about what this is going to do to your rating you only care about how to make it work so you would be better off going into a math section on Reddit or something
Then you should not have come to this forum because you don’t care about what this is going to do to your rating you only care about how to make it work so you would be better off going into a math section on Reddit or something

Do you want validation for quite possibly the dumbest idea and most if not all of the history of reading because we don’t know why times being wasted are you saying that time is wasted because your wife more often than not are you saying that it’s being wasted because you pull slowly instead of focusing on making an equation to show people how you can do this better just use your fucking words
Do you want validation for quite possibly the dumbest idea and most if not all of the history of reading because we don’t know why times being wasted are you saying that time is wasted because you wait more often than not are you saying that it’s being wasted because you pull slowly instead of focusing on making an equation to show people how you can do this better just use your fucking words

Bring up to the Raid leader hey we go for round two much we should pull faster so we can get more kills

Or say hey kills take too long we should make sure that people pre-pot on farm And don’t worry about padding meters and just focus on getting the job done

Maybe say hey we have people go FK just before the boss how about doing that on the trash pole if you have to make yourself a drink or go to the bathroom

Every single one of those suggestions that I just gave you will result in less waste of time but you do not care you want an algorithm so you can feel high and mighty and feel justification for when you’re an asshole so when your raid lead kick you from the group I want you to remember this

[Revamp Man]

That's clearly false, otherwise you'd just go find a better guild.

- - - Updated - - -

If the people you raid with are oh so mature and logical you shouldn't need "math" to convince them they are dying too much.

There's a difference between "we're wiping too much" and 'I WENT ON MMO-C AND FOUND SOMEONE TO MAKE AN ALGORITHM TO DETERMINE WE'RE DYING TOO MUCH!", the former is reasonable, the latter makes you look like a micromanaging [word I won't say here].




Plenty of people here are good at math, your mistake was thinking anyone here is interested in assisting you in your quest for validation... Go do it yourself.

Honestly I feel kind of bad for the people in this guys guilt because they don’t die and awful lot if the percentage of survivability is what he says

He’s obviously not the Raid leader or an officer so he’s just one of the Raiders which someone pointed out don’t even have to be friends with one another which is very true

And he’s going to go in and say that he made this magical equation that says people are dying to much this guy seems like the scientist that ran a bar and didn’t know anything about it

Honestly I’m back in my mythic Raid days if someone like him came into the guild group and said shit like that I would take them without hesitation or I would leave because you’re trying to change the guild to match your mentality which never ends well

Actually come to think of it my guild did have a random player that we grab and we were doing farm on her Roeck palace but we kept dying to lady and it was because our healers were slacking and we had a trial but the random guy decided to say that it was our majors fault because he was fire and everyone knows that fire is not the best spec. To post later we kill it and I happily mock the random person during the next few trash balls of the night but sadly we lost them each because a couple people in the group we’re dumb enough to believe what some random guy said simply because he sounded like he knew what he was talking about and that is what it’s going to happen to this guild

[deenman]

Right, it's not necessarily the individual survival rates that are the issue here. It's that on hard bosses you haven't beaten yet, you typically need the whole squad running at near 100% to get the victory.

Individually, each person might only get 1 death, but when 1 death nukes a run, and everyone is taking turns being the person who fucks up, THAT'S when you start to lose a lot of time.

oh yeah,this is extremly frustrating when this happens just like you said...a wipe mecanic where every raid member takes turns fucking it up is just the worst thing

[kaminaris]

Survival percentages aren't really that great in the first place (imo). Not how they're calculated on Warcraft Logs anyhow. They're based on the total encounter time. If all the encounter lengths were equal it wouldn't be a problem but they're not.

Say player A dies 45 seconds in to a 1 minute wipe. He has a 75% survival percentage for that wipe. Player B dies 45 seconds in to a 5 minute wipe. He has 15% survival percentage for that wipe. Warcraft logs then averages these percentages together with no consideration for the duration they were calculated off of. Lets say that there were 8 wipes total and these were their only early deaths (so 100% for the other 7 wipes). Player A has a 96.9% survivability rating. Player B has a 89.4% survivability rating. Player A's survivability rating is considered good while B's is bad, despite their making the same mistake (presumably).

Yes that is correct that is why 30 samples are not viable as test samples. However if you consider 300 or even 1000 pulls which is my first tool is doing, you will get pretty accurate results. I have also the ability to remove any pulls shorter than X seconds, more configurable than on WCL.

In 30 pulls, shit can happen, in 300 or 1000 pulls if shit happens 2% more often then its most likely players difference.

Basically, a player's personal survivability metric is worse if the rest of the raid can (and does) continue on without you. The irony is, the more "important" a player is to the raid encounter (read probability of calling a wipe after their death), the better their survivability score will be despite those deaths. That's not to say these will always happen, but it throws enough uncertainty into the system to make me question it's validity as a metric if you don't have consistent pull times.

I was wondering if having so much data, wouldn't it be better to use raw metrics instead of capped to 3 deaths.

---

I'm not sure if this is really along the lines of what you're looking for but if you're any good at mining info off WCL, a more practical way to go about calculating "wasted" time would be to ignore the survivability percentage completely. In addition to what I wrote above, survivability already is a summary of a time metric so using it as an intermediary for a new calculated time metric isn't all that useful.

Instead, you could pull the following information from each wipe:
Start of pull
Players in the Raid
The first X deaths
End of pull
Wipe or Kill

I mean, why would I want to have raw data parsed if I cannot compare that to anything? The whole point of this algorithm is to have:
What if our people would die less?

Sure, I can calcuate how much time we waste NOW. But I still won't know how much would it be different if people would shape up.
For example if farm of 5 bosses in ideal situation would take 1:30m while we are doing it in 2h (hypotetical) that means we are usually wasting at least 30 minutes per raid day. That accumulates into 1:30 weekly. 6h monthly.
During that time we could do about 60 additional pulls on some progress boss.

Keep a running total of time spent by each player is in the raid (to compare to how much time they contribute to wasted time)

For each wipe, determine how long till the start of the next pull (to get an accurate wasted time metric)
Apply that amount of time to a running total of wasted time (for the entire raid as a whole).
Also track a running total of wasted time by player.

So quick and dirty example.
Wipe 1 7:00 - 7:05
Players A - U in the raid
First deaths Players A and B

Wipe 2 7:08 - 7:10
Players A - U in the raid
First deaths Players B and C

Kill 7:13 - 7:20
Players A - U in the raid

Players A - U all get 20 minutes of raid time (7:00 Wipe 1 start -> 7:20 Kill end)
Total wasted time gets 13 minutes (7:00 Wipe 1 start -> 7:08 Wipe 2 start and 7:08 Wipe 2 start -> 7:13 Kill start)
Player A gets 8 minutes of wasted time (7:00 Wipe 1 start -> 7:08 Wipe 2 start)
Player B gets 13 minutes of wasted time (7:00 Wipe 1 start -> 7:08 Wipe 2 start and 7:08 Wipe 2 start -> 7:13 Kill start)
Player C gets 5 minutes of wasted time (7:08 Wipe 2 start -> 7:13 Kill start).

It's not going to be perfect since first deaths aren't the end all and be all of what caused a wipe. It should show trends though and if certain people are dying first all the time, they should have significantly higher calculations of wasted time. You could also consider weighting the deaths. Say apply 2/3 of the time to the first and 1/3 to the second or whatever you feel is appropriate.

Anyhow, just a thought.

But again, I could just take a stoper and note the times of each raid or write another WCL parser to handle that for me, but I still won't have anything to compare.

I don't need for example wipes caused by some random events like server lags or slack on trash. I wanted to see how much we could save time if our people wouldn't die so often.


EDIT
--------

I think what I also need to take into consideration is death time.
For example if first guy dies in first 30 seconds in a fight, probability of wipe increases by X% - lets say 20%
However if first guy dies in last 30 seconds, probability of wipe is only less by like 2%

so
initialWipeProbability = (1-(deathTime/fightDuration))*(20)

so death at 30s for 600s fight = 19% wipe probability
death at 570s = 1% wipe probability

[callipygoustp]

Shouldn't take a special formula to figure out who isn't pulling their weight. If you do need a special formula, then your raid team is shit and no formula is going to solve that problem.

[Revamp Man]

You're an awful person for calling a worse player a derogatory name. Does that make you feel superior?

Isn't the whole purpose of the game to roleplay? But past that point, why would you not want to do things as efficiently as possible? The game is a lot more in-depth than you think it is. The way you name-call people is disgusting. The whole point is leadership leading and shaping the guild in the direction they want to take it. If creating algorithms is what's considered professional for you then I feel sorry for you and your guild. I assume by now it would be a basic thing the majority of guilds do, if not simply vet warcraftlogs or input it into something that's already been created.

So this guy is not the Raid leader but he wants to make an algorithm that tells people why they suck and how the entire group can improve and that is the point right there


The same result can come from the Raid leader looking over a log or checking out white fest and talking to the players which is fine that is the easiest way to do it and the Raid leader should do it because for one reason or another they are the leader ship

High as a raider will respect what the Raid leader says but if someone else comes in and says hey we are in efficient here’s how we get better and uses some stupid equation I’m probably not going to respond very well to it

That’s called tact and it’s actually something used in leader ship. Little Jimmy might not like some random person who has no authority over him coming in and saying everyone else sucks let’s save 10 minutes and this magical equation will show us how.

I don’t want to play with some asshole like that so I’m just going to not show up to farm nights I’ll still do progression but I won’t do farm if that means getting kicked off the team cool there are other Raid teams out there


If you go into a group and you have a different mindset from that group then you have to either change your mindset or find a different group there is no third option.

Now from the point of the Raid leader if I have multiple people like Little Timmy who are upset that one of the other Raiders who seemingly has no position of authority is coming in and wanting to make changes to the group then I can either choose to get rid of one person or lose more than one person.

So you have to figure out which is better now is this guy wanting the equation a tank for the group because if so that’s half the problem he can speed up the night if he wants. Is he a healer which seems to be more likely or the possibility that he’s a random DPS.

None of those should really matter but if I am the Raid leader and I have little Timmy who is a DPS or a healer or a tank I have to weigh them against him and if it’s multiple people then I have to do the same thing.

You cannot progress if you lose a lot of people and as much as people say that you do not read to be friends with others chances are you have friends in your team.

So as a Raid leader the choice would be very obvious which is to kick this person who has a very high opinion of themselves from the group because maybe some people enjoy having fun on the farm maybe they don’t want to take farm too seriously because it’s not progression.

Personally I don’t give a shit about my guilds farm nights if we’re going to the boss I do everything I can to kill the boss as quickly as possible when it’s on progression that’s when I actually try but during farm if people want to tell a random story about someone at work while we’re killing the boss OK that’s cool just don’t do it on progression.

Heck I recently almost pissed off one of my guilds tanks because I told him to swap on the final boss so he stays up and I told him the reason was because it would be easier for him in my opinion and it would make more sense since he has higher mobility. He was upset for about 15 minutes but immediately saw what I meant and if I came off in a different way he told me that he was going to log out but he knows me he knows how I talk he understood my suggestion.

Now this guy wants to create an equation or an algorithm to tell an entire group of 19 other people who he is unsure how this will work and he’s unsure about the reception and he expects people to not criticize it and only help him with it because it’s a good thing.

Screw that you’re going to end up killing your guild unless you have a Raid leader with a back bone in which case he will tell you to sit down or kick you that’s the best choice that someone has.

[callipygoustp]

There's a why for everything.

Amen. In this case: why are some raid teams shit.

[Revamp Man]

Why did you not debate him when he made the dumb statement? Why did you feel the need to act immature toward him instead of bringing up other factors than just the individual flaw he pointed out in not the best manner so you could set an example of how to actually deal with this issue? Your guild sounds awful and immature to raid with.

- - - Updated - - -



There's a why for everything.

Because this random person came into my group and insulted someone who he didn’t know because he wasn’t playing the most efficient spec that’s not nice that’s not friendly that’s not any way to act when you are in a group who could not care about your opinion because you are a random person so I’m going to point out that they were wrong I’m not going to be nice about it because I don’t have to be nice about it. The tour ended up making our mage quit because when the tour spoke there were two other people in the group that said oh that kind of makes sense and that kind of sucks because the mage could’ve just taken it with a grain of salt considering the guy took more damage than a tank. So if you are a tool you deserve to be called out for being a tool this person was not a trial they were a random person that got invited because we listed for some reason and I did not expect them to return and if they are going to have that kind of mentality that drains the fun out of the group and blames deaths on a player not using the most efficient spec in a fight where it doesn’t matter then I’m not gonna worry about it.

Best part is the same guy came back like two weeks later and dropped bombs on our tank during radiance and our tank takes zero shit and told him to stop fucking doing it and do the mechanic correctly. The random guy‘s response was simply go ahead and kick me I’m your top DPS I bet you won’t do it LOL and in about 10 seconds he was kicked mid pull.

- - - Updated - - -

Amen. In this case: why are some raid teams shit.

I mean it’s not even about progression or performance apparently this guy just wants to figure out wasted time for polls which is fine but the numbers that he suggests for survivability is really weird.

If the group is shit then he should just leave as you said it doesn’t take a giant formula to figure that out and stuff like this never ends well

[callipygoustp]

I mean it’s not even about progression or performance apparently this guy just wants to figure out wasted time for polls which is fine but the numbers that he suggests for survivability is really weird.

If its not about progression or performance then why look at progression/peformance metrics?

If the group is shit then he should just leave as you said it doesn’t take a giant formula to figure that out and stuff like this never ends well

Yep, that is what I am saying. WoW is far from rocket science. If you have to complicate things.... either accept where you are(which this guy clearly isn't accepting) or, well, yeah... move on.

[callipygoustp]

The individual players bottom-up do not care enough to push themselves to improve. The raid leader is not good enough, does not care enough, or unable to recruit players that are as good as them and/or they have poor leadership skills. The players are decent but the guild has awful leadership so players constantly cycle through and leave as they improve to a much mentally healthier guild. etc

Not sure why you needed to quote me to state the obvious, but, good on ya.

[Revamp Man]

The individual players bottom-up do not care enough to push themselves to improve. The raid leader is not good enough, does not care enough, or unable to recruit players that are as good as them and/or they have poor leadership skills. The players are decent but the guild has awful leadership so players constantly cycle through and leave as they improve to a much mentally healthier guild. etc

- - - Updated - - -



yikes......

- - - Updated - - -



thank rng I don't play with people as childish as you in video games anymore.

Ditto my friend

[ShmooDude]

Yes that is correct that is why 30 samples are not viable as test samples. However if you consider 300 or even 1000 pulls which is my first tool is doing, you will get pretty accurate results. I have also the ability to remove any pulls shorter than X seconds, more configurable than on WCL.

In 30 pulls, shit can happen, in 300 or 1000 pulls if shit happens 2% more often then its most likely players difference.

Not necessarily, for example a tank dying has significantly higher odds of being a wipe so that's going to raise his survivability up compared to a DPS dying the same amount. Probably not enough to interfere with what you want to do with it though.

I was wondering if having so much data, wouldn't it be better to use raw metrics instead of capped to 3 deaths.

Given the way you're doing your calculations, definitely. Or at least increase it to say 6 since even by your own code 3 dps deaths is only a 30% chance to wipe.

But it does not seem to be close to reality. Additional chance of wipe is not relevant enough.

Here is full version, https://pastebin.com/701DEACa

According to this, we would be loosing around 30 minutes on 5 bosses.

Also (if I'm reading your program right) you're using 100-survi to calculate the chance of death, but it should be (100-survi)*2 because you never have 0% survivability rating for a death (it can be anything from 0-99; 2 makes it assume those deaths average out to 50%). That's probably why your numbers aren't working.

For example some log I pulled up of 17 wipes. There's a resto shaman with an 89% survivability rating. In your current program that's an 11% chance of failure. Over 17 wipes that's an average of 1.87 deaths. However the shamans actual deaths (with ignore after = 3) were 4, with a 64.8, 27.8, 93.3 and a 27.2.

Same logs, another resto shaman with 84.6% average. By your program that's 15.4% death chance or 2.62 deaths. Actual deaths were 5, with a 94.5, 52.9, 38.3, 34.8 and 17.5.

[kaminaris]

Not necessarily, for example a tank dying has significantly higher odds of being a wipe so that's going to raise his survivability up compared to a DPS dying the same amount. Probably not enough to interfere with what you want to do with it though.
Given the way you're doing your calculations, definitely. Or at least increase it to say 6 since even by your own code 3 dps deaths is only a 30% chance to wipe.
Also (if I'm reading your program right) you're using 100-survi to calculate the chance of death, but it should be (100-survi)*2 because you never have 0% survivability rating for a death (it can be anything from 0-99; 2 makes it assume those deaths average out to 50%). That's probably why your numbers aren't working.

For example some log I pulled up of 17 wipes. There's a resto shaman with an 89% survivability rating. In your current program that's an 11% chance of failure. Over 17 wipes that's an average of 1.87 deaths. However the shamans actual deaths (with ignore after = 3) were 4, with a 64.8, 27.8, 93.3 and a 27.2.

Same logs, another resto shaman with 84.6% average. By your program that's 15.4% death chance or 2.62 deaths. Actual deaths were 5, with a 94.5, 52.9, 38.3, 34.8 and 17.5.

Are you sure about that?

Technically you cannot have people who are perfect thus 100% survi doesnt exist, same goes for 0%.
But mathematically this is still possible.

My program calculates it like this.

Roll 0-100 -> if that roll was higher than someone survi -> death
So if you have 95% survi, the only rolls that kills you are 96, 97, 98, 99, 100 out of 101 possible outcomes.

[ShmooDude]

Are you sure about that?

Technically you cannot have people who are perfect thus 100% survi doesnt exist, same goes for 0%.
But mathematically this is still possible.

My program calculates it like this.

Roll 0-100 -> if that roll was higher than someone survi -> death
So if you have 95% survi, the only rolls that kills you are 96, 97, 98, 99, 100 out of 101 possible outcomes.

Yes, because the survivability rating is not a measure of the percentage of attempts you lived on. It's a measure of the percent of time you were alive (not counting a player being battle rezed). So a 95% survival rating doesn't mean you died on 5% of attempts. It means you died on approximately 10% of the attempts (might be a bit more or less depending on how early/late you died on the attempts you did die on). All of the following assumes an "ignore after = x" since we're trying to glean info from "relevant" deaths as opposed to the people dying after a wipe was called.

As an example, let's take a 49% survivability rating over two attempts. On how many attempts did that player die? It has to be two, because even if they died on pull for one of the attempts for a 0%, the second attempt could be no more than 98%.

Now, for a player with an 80% survivability. What percentage of time did they die? It's at least 20% although that would mean they died on pull for the attempts they died. However, it could theoretically be 100% of attempts, if that player always died late into the pull. Almost certainly neither of these is true, so the answer lies somewhere in the middle.

---

In terms of your program, the way it's calculated now results in you having 100% survivability on the attempts you win your roll and 0% survivability on the attempts you lose. The change I suggested still has you at 100% survivability on the attempts you win your roll, but 50% survivability on the attempts you lose. In other words, it assumes that on an attempt a player dies on, on average they died halfway through the pull. If you look at the log numbers from my previous post, this is roughly the case. The attempts the first player died on averaged to 53.3% and the attempts the second player died on averaged to 48.2%.