SBMM DOESN'T WORK. KILLER STATS FROM 100 MATCHES TO PROVE.
Comments
-
That isn't your data, this is your results. Whithout the data I cannot analyse the data. No offense but most people have no clue how to statistically correct analyse the data.
0 -
There.
That's the entire (continued) spreadsheet converted to a PDF.
0 -
thx for the effort, but I'd need a csv or excel. Like that I'd have to copy everything in myself and tbf I don't have time for that.
0 -
Fine. This should work. It's even more updated, but I can't verify all the data. It should be right, but there will probably be minor issues with it.
0 -
I do believe that if an average player's statistics start to align with someone who drastically spends more hours on the game - as a form of career, that presents a bigger fault.
1 -
Statistically, this isn't true though. He may have matches where he loses, but they are a drastically smaller percentage. This is not accounting for matches where he is being stream-sniped, or encountering cheaters/throwers.
I'm not saying that he steamrolls, because he has games which are close calls and where he actually has to put himself together to secure kills, but the man's statistics speak for him.
1 -
The simple fact is survivors lose even with this much of survivor favor.
2 -
Oh btw SBMM is proven by devs to NOT work, since there is extreme shortage of killers in most region, killers tend to throw onto already existing lobby... which can have excessively higher or lower MMR thanks to "priority".
0 -
I stand by what I said, the person I responded to spoke with hyperbole and misrepresented reality. Given that this is a thread about statistics, people should be culpable for that nonsense.
Edit: This doesn't mean SSBM isn't a ######### show. We know it is. My whole contention with the OP's nonsense is that he's not informed about what makes a statistically significant, or accurate, sample size. He makes 0 effort to control variables and his data is largely junk. The fact that he speaks in absolutisms about things he clearly misunderstood is just icing.
0 -
I think you're reading into the intentions of OP far too much.
OP recorded some stats, over his last 100 games, based on those statistics and then the comparison between them and someone else's statistics, he made his conclusion. Is it perfect? no, but you will never get a perfect conclusion on statistics that have far too many variables to begin with. So trying to reduce the argument down to that, is pretty redundant
1 -
I disagree. You can run some pretty cool data collection experiments in DBD. It hurts that they don't have an API that allows us to call data by user, but you can do it manually. You can eliminate SSBM bias by playing 50 games on every killer extremely sweaty before starting your data collection efforts, and then you can eliminate further variables by isolating and using high value, generic perks that address current metas across the entire roster. You can reduce feast or famine results by excluding NoED and certain iridescent add ons. You can exclude games where DC's tamper with the actual outcome of a match and you can VOD everything to gather extra statistics at the end that will allow you to sort and present the data more accurately.
You don't have to do something perfect for me to respect it. You just have to do something not utterly lazy that yields no valuable conclusions.
0 -
I dont see why you have to respect it.
The data is still data, you are free to criticise it, but it doesn't make it any lesser.
1 -
Then I don't see the need to continue this conversation. I did criticize it, it is junk data that yields nothing conclusive, despite how hard he pushed his bias. Which is really my only contention with what you originally said to me, that I'm reading too into it. Sounds like this conversation was completely unnecessary.
0 -
Yeah, try getting bullied by a team with 6k hours and then they get pissy after you waste their time when they three gen themselves for a half hour and then look up your stats.
0 -
Well, ONE of my teammates tonight had over 200 hours.
So that was cool
0 -
Thank you. I'm glad we have some level headed people on these forums :)
0 -
In Blackjack, a card counter can still lose their bankroll despite a 0.01% risk of ruin.
0 -
Well, if we are going to provide random gambling info unrelated to DBD, I can add one too:
The opposite sides of a die will always add up to seven. If you take a look, you'll see that one and six are on opposite sides of the cube (1+6=7), as are two and five (2+5=7), and three and four (3+4=7).
And did you know cats can't taste sweet things because of a genetic defect?
0 -
Technically card counting isn’t gambling. That’s interesting though.
No, I didn’t know that.
Did you know that the oceans of the world contain approximately 20 million tons of gold (that’s $771 Trillion Worth of Gold)
Not on the ocean floor—in the water itself, in tiny particles of approximately 13 billionths of a gram per litre.
0 -
Devs said that? When? Care to post a link?
Don't get me wrong, I believe you, but I just want to read or hear the exact words myself.
0 -
Ok, fine, card counting... But what does it have to do with DBD? Is there a new killer coming out that requires you to know how to count cards? Or are we getting a new map with ocean where side objective is to collect gold?
0 -
Almost.
TLDR: Your sample size is too small.
0 -
It's not: https://tools4dev.org/resources/how-to-choose-a-sample-size/
Prove me wrong.
0 -
Yes we should compare all killers that play this game to Otz. That's how balance should work /s
0 -
No problem.
For example, using a population size of 62,074 (last 30 days steam charts) and a generous 90% confidence level with a 5% margin of error you’d need a sample size of approximately 270, which is still on the low side considering we are dealing with a video game and not a survey.
Going for the minimum sample size of 100 is not going to give you an accurate result with the number of variables involved.
A goal closer to the maximum recommended sample size of 10% will give you more accurate results.
6,207.4 of course is too high, so you’d need to cap it out at 1000, which is 10 times more than the sample size you’re currently using.
Edit: I just want to add that I think it’s great you’re doing your own research/testing and I know it takes a long time, but if you want a reliable degree of accuracy, a larger sample size is required.
0 -
I see where there is a misunderstanding now. You might be confusing total population of players with total number of games played by individual player.
The population size would be relevant if I was trying to determine if matchmaking works for everybody. In that case we would need 10% of all players to take stats over multiple games each. But again, in this case people would not need to record more than 100 games each. But yes, we would need 10% of 62k people to participate for excellent results or 1000 for good results.
However, this is not what I'm determining from my stats. I'm determining that SBMM doesn't work for me specifically (as opposed to other people), therefore only the number of games that I played is relevant.
Just to make it clear: the conclusion that I'm drawing is that there is a category of people for whom SBMM doesn't work. This doesn't mean that it doesn't work for everybody, which is where total population size would become relevant.
2 -
Ive switched to playing killer a lot more since i changed to PC. I find killer im getting mostly 4k or 3k. Occasionally destroyed and only a couple hooks.
The biggest problem with matchmaking is it puts too much weight on my kills as a whole so when i pick a new killer my games are against fairly good players and i dont get a chance to even practice or play much while im trying to learn their abilities.
I think its harder to win consistently as survivor (mostly playing as a duo or 3), but depends mostly on RNG and how the killer decides to play.
One thing ive noticed is if im playing killer i can often change the entire way a game is playing by deciding to slug, proxy camp etc. Whereas for survivor youre forced into trying to fit around that which is a lot harder.
Saying that, killer is still the more stressful of the roles as youre not in a team so i do get it more from both sides now to be fair.
0 -
My bad mate, you’re absolutely correct and I got the wrong end of the stick.
*insert excuse here*
I need to work on my attention span lol.
0 -
I'm the opposite. I find killer relaxing, because I keep in mind a few important factors:
- I control how match goes, because I'm 100% of my team
- I get to play to the end of it.
- Statistically speaking, I'm probably going to win. (I win at least 3/4 of my matches)
As survivor I find it very stressful, therefore only play it for rift challenges:
- Not much in my control as I'm only 25% of team.
- I may get tunneled out quickly or left on hook or be ignored by killer, so I have no interesting gameplay (hanging on hook or M1 on gens is waste of time to me) to show for wasting time waiting for match.
- Statistically speaking, I'm probably going to lose. (I lose at least 3/4 of my matches)
0 -
Not a problem. I could have been more clear in my title by adding "for me".
1
