May 2018 Tuning Discussion - PRELIMINARY

[Reeb99]

And I don't believe you. I've seen it too many times just like everyone else. I understand confirmation bias when it comes to drop rates and the like, but you aren't playing the same pvp teams we are.

[Yup]

Just fought a team with Dessert Marauder w/Echo.. it got 8 (EIGHT!) turns in a row. I've NEVER had more than 3. How does it hit that 35% chance 8 times in a row???? If it is all RNG, which I'll accept. The RNG REALLY seems to favor the AI. So if it is merely player perception.. then it seems to me some sort of "cap" on the AI would assist. Such as "no more than 4 perfect hits in a row", "No more than 3 extra turns in a row" that sort of thing. Because that's really what players experience.. a "max cap" on those sorts of abilities firing - even if that "cap" is due to the player missing a perfect hit or something.

And it's not just PvP, it's seen in PvE as well... running BNW GOF with the S Mayan Priestess and Echo talismans.. if I run it.. she might get 3 or 4 extra turns.. if I auto-play she OFTEN gets 6 -8 extra turns. (BTW S Mayan doesn't do ANY damage to the DK in GOD when she "revenges" - not sure if that's intentional. She'll hit the DK shield on revenge, but won't ever do any actual damage on revenge if the shield is down. She does do damage in GOF)

[Mizrael]

I’m gonna join in on the “cheating defense conspiracy”. I think everyone of us has mentioned seeing it...and not once. i would dare ta say i’ve seen it 20,000 times.

Plus the 50/50 coin toss:
I remember one episode of Derren Brown trying to do a 10 heads in a row. It took him way over 100 tries to get the footage....i’ve lost the coin toss 11! times in a row the other day. go figure....

[JJJ428]

Mine is more the invincibility buff anymore I swear I have how many toons that remove buffs and will remove every buff but invincibility

[Liebhild]

If I have a 35% chance I don't have to run 20000 samples to say with 99,9& that there is something wrong.

Number is much lower, and there is something totally wrong. Read here in the forum, lot's of examples.

F.e. my AoF NEVER, really never, stuns on 1st attack, I am observing this for weeks now.

[hold]

It's pretty straight forward in terms of implementation. The problem you are running into is small sample size and a bit of confirmation bias. If you were to run the same battle 20,000 times in a row, the results would average out. But as mentioned, it's quite possible that you can indeed run into unlucky streaks.

Again, this has nothing to do with defense getting favored in any way from a mechanics stand point. That's just not the case.

-Kaz

I'm sorry but I belive that is plain bollocks ... maybe it's a bug again but the reason I'm avoidig teams with Red Lilith is that most of them running Echo talismans and they always get at least 3 turns.

It's possible like the weird coincidences like pulling 2 WUR ater each other on 2 Legendary souls or any of the other weird dupe shit ... but highly unlikely.

Either you really don't have knowlegde about some of the code or you are talking bullshit - sorry for being blunt.

[Yup]

Red Lilith is that most of them running Echo talismans and they always get at least 3 turns.

I wholeheartedly agree with that. Red Lilith + Echo on defense gets 3 turns 95% of the time

[Nifelheim]

to those that suspect something else at play with unlucky streak, is your win streak above 50%? If so, consider yourself already ahead haha. Top players are in the upper 98% not withstanding crashes, so...

And the devs have said in the past the AI isn't that smart. Play pvp on auto and AI will prob play your team same

[scott-5496]

I wonder if in PVP defence test screen if you set the same attack team vs the same defence team if you run this on auto say 100 times then outcome should be about 50/50 attack to defence wins? Looking at statistical analysis (I do this in my job for sample surveys) 100 cases is not a really enough but something closer to say 500 would be pretty accurate - you should never really need 20,000 goes or anywhere near it to get an idea on accuracy of results to +/- 5%. The maths does not quite hold true for a game of course as the number of possibel battles is infinite (surveys are not if you are sampling from a fixed universe of something like households in a geographic area that you know the totals for). The logic should hold true though. Here in the UK when we are doing large surveys of a Local Government area we know that as sample of 1500 surveys wil do the trick for overall accuracy and that we can then look at smaller geographies within the Local Government area at a lower degree of accruacy. So the same will figure here give or take a % point of accuracy or two.

Anyway, who has the time to run these (well maybe I can over time!) but that would be interesting to do and just to see how the AI works out after several hundred runs on auto. I may have a think about it as I am a boring stats guy in my job so why not run this little experiment in the back ground to see what happens!

[Mizrael]

I wholeheartedly agree with that. Red Lilith + Echo on defense gets 3 turns 95% of the time

I’ve had a battle last night, where SSE gave echo to the team and EVERY single team member used up all their charges afterwards...seriously like 12 attacks in a row!!!