# Rolls got a lot worse

It is somehow impossible to make posts on FB app page regarding issues, that will be noticed therefore i must do it here. Did it occur to players that something happened recently to tales rolls? DB promised in the past to make RNG more... random, so here it is. We have a drastic increase of increased values, like 3, 99, 5 and my personal favorite 100. I had 5 games in the row where 100 came up. That, surrounded by high 90s of course. What happened to values between 30 and 70??? Was this an intentional move to make up for the introduction of colossal dragons that, it comes without saying, would be improving our ss strength? Is anyone looking at the outcome of fixes? yesterday night i had an epic run that i had to replay 2x (on first stage) because it had 4 rolls out of 6 over 92 and that of course included a 100. My question to DB is: was this intentional nerf or you just asked an intern to play with the code?

3

## Comments

Sometimes you get a very bad day with tales, sometimes you get a very good day with tales. I know people who had lots of luck last tale.

It's not personal. It's the R part of RNG. It's not supposed to be the same for everyone.

One problem is that many players do not understand the chance of success is often much higher or lower than the chance to win a single roll. For example, at 82% chance of success for your first roll, you only have about a 50% chance of winning all five rounds. Most people think it ought to be higher, but statistically, that is the number we are looking at. I'll post my Tales tools later.

Mature Dragon: Manannan mac Lir

Adolescent Dragons: Kutkh, Samartitxiki, Sun Wukong, Ti Malice, Veles, Wisakedjak

Baby: Anansi, Azeban, Exu, Guidgen, Huehuecoyotl, Itomi, Juha, Lilith, Loki, Maui, Mendacius, Nahash, Seokga, Set, Susanoo, Wahn, Yun Harla

Well DB is the one who likes to pretend the % remaining the same means the probability/chance is also the same (or as they kind of put it x% chance will always mean x% chance), so can't wholeheartedly just blame the players if they believe having 80 on stage 5 is the same as being on stage one and having 80%...........then you have when someone says something like their having a bunch of low rolls and someone else says they will get hit with high rolls later on, some people act like the 2nd person must be crazy but guess what the stats won't actually balance out if those high rolls don't show up will it, the more rolls that happened without particular numbers showing up the higher the probability of said numbers showing up on the next set of rolls.

Also I think I saw your tool on the previous forum, didn't really fully pay detailed attention to it (just skimmed), but while you need to avoid the probability of failing stage 1-5 in order to win it all, you don't need to win all 3 rounds in order to move on..........so like I said I only skimmed it and it was a month or two ago at that, but hopefully your probability accounts for the fact it doesn't take 3 wins in order to advance to the next stage (I believe it did and might have actually had the option to select 2 or 3 wins per stage, but not sure so just saying).

Mature Dragon: Manannan mac Lir

Adolescent Dragons: Kutkh, Samartitxiki, Sun Wukong, Ti Malice, Veles, Wisakedjak

Baby: Anansi, Azeban, Exu, Guidgen, Huehuecoyotl, Itomi, Juha, Lilith, Loki, Maui, Mendacius, Nahash, Seokga, Set, Susanoo, Wahn, Yun Harla

That being said, the "I do pay attention to those as well" portion of your comment gets at one of the major issues. If you do not record and chart every single roll for an extended period you will perceive a J-curve or U-curve because extreme rolls are more psychologically salient. Even people whose rolls are on a bell-curve (which is also not statistically normal) perceive a J-curve or U-curve.

Most people that posted on the previous forum that recorded their rolls over a period of three or more Tales found that they had a random distribution (equal chance of every number, or at least every decile). Some noticed clumping (numbers separated by 0-3 appearing above the statistical norm in "clumps" of three to ten). I have noticed that the time of day seems to have an effect on my rolls (may be a hint as to how they make random numbers). That being said, Nod posted all the rolls for all players for a period and it showed random distribution.

I'm not saying that it is not happening. I am just saying that there is a high probability of perceptual bias.

Mature Dragon: Manannan mac Lir

Adolescent Dragons: Kutkh, Samartitxiki, Sun Wukong, Ti Malice, Veles, Wisakedjak

Baby: Anansi, Azeban, Exu, Guidgen, Huehuecoyotl, Itomi, Juha, Lilith, Loki, Maui, Mendacius, Nahash, Seokga, Set, Susanoo, Wahn, Yun Harla

P.S I think it was also nod who once talked about how people see or find patterns in randomness (I am greatly paraphrasing, as I recall the general idea he was trying to say but not the exact words use), yet that statement is complete bullock when it comes to rng as it not only follows an algorithm but the stats likely wouldn't balance out if it was complete randomness, thus one could not go around claiming everything is working properly and the stats indicate a even distribution/balance.

They really should and hopefully it will speed up the results because that 5 second wait for it to show you what number was rolled is beyond annoying.

Seriously, though, every Tale I record all my rolls to observe this very thing. Oftentimes our perception indicates one win rate and the numbers show another, so by recording every roll throughout a Tale, I am able to see how bad my luck really is. It's usually a little worse than it should be by a couple percentage points. E.g., I'll have maybe 13% or 14% of my rolls being 90-100 by the end, instead of the expected 11%. That's within a margin of error and I attribute the consistency of almost always being in the red to my own bad luck.

This Tale is different. Granted, it's not over yet, but out of my first 100 rolls, 25% were 90-100. I've never seen it off by that much before. And I really doubt that later in this Tale I'll suddenly see 25% of my rolls being 1-11 (though as I mentioned I won't be racking up my usual total number of rolls, which average around 600-700 per Tale). The RNG seems to only get streaky with the high numbers, not the low ones.

Not true at all! If your SS are starting out with 80-85% for wins, you probably end up around 75% by the 5th stage and your chances of winning the whole thing are just above 50%. Take into consideration the extra dynamic of occasionally having several of one action in a row, and the chance of winning would likely be less than 50%. Even considering it was 50%, losing 6 times in a row is just 1 in 64! So If you've got an alliance with 64 people and you all do 6 tales with a similar %, the chances are that 1 of you will fail all of them. Of course, the chances are that 1 person will succeed with all of them.

1 in 64 is definitely not too low to write off to bad luck!

and whats the odds of that happening?

which is also 1.5625%, so not weird at all to have a low probability result of a specific outcome sequence (outcomes which themselves are based on the result of not simple one roll, but multiple rolls).

by your logic nothing is ever really too low to write off as more than just bad luck, because not only is there always going to be a probability (no matter how small) of a specific set of outcomes but one can always claim that with enough people chances are it will balance out (so basically it just a persons bad luck but the system is good)

Can the system really be considered to be working correctly, if balancing out the odds is a matter of taking one persons exceedingly negative results and combining it with a different persons exceedingly positive results.

P.S he/she did say that those 6 rolls now makes 10 fails out of their last 11 tries to complete all 5 stages, so you did the odds of the 6 in a row but didn't do the odds for the 1/11 success rate......better yet combine the two, what are the odds of failing 4/5 tries then failing the last 6 (i'm actually curious as to what it is, but having a brain fart as to how to figure it out)

P.P.S i'm interested in my odds, 5 fails in a row with 80% win rate (all

on stage one)Possible result combinations & chances of each combination of 3 results assuming

80%win rate:Win Win Win = win = .8 *.8 * .8 = .512 (51%)

Win Win Lose = win = .8 * .8 * .2 = .128 (13%)

Win Lose Lose = loss = .8 * .2 * .2 = .032 (3%)

Win Lose Win = win = .8 * .2 *.8 = .128 (13%)

Lose Lose Lose = loss = .2 * .2 * .2 = .008 (1%)

Lose Lose Win = loss = .2 * .2 * .8 = .032 (3%)

Lose Win Win = win = .2 * .8 * .8 = .128 (13%)

Lose Win Lose = loss = .2 * .8 * .2 = .032 (3%)

Total % Loss results: 10.4%

Total % Win results: 89.6% (at an 80% win rate, your chance of wining the first stage out of 5 =

90%)Total: 100%

Odds of losing on the first stage 5 times in a row on the next 5 attempts,(NOT 5 in any given string of any number of attempts): 10.4% to the power of 5 = 0.000012166 =0.001% or 1 in 10k.If you want to know the odds of this happening sometime during your entire tale and assuming you do 70 total attempts at the same difficulty, then there are 66 possible strings where you can lose 5 attempts in a row at the 1st stage = 66 * .000012166 = 0.000803 =

0.08% (or a little less than 1 in 1k). Assuming there are 5 thousand players that play Tales (hehe, lol), about 3 other players (4 total) will also experience the same extremely bad luck (of losing on the 1st stage 5 times in a row) that you have experienced this tale.The real question many players ask is what are the chances to advance on all 5 stages(assuming a win rate of 80% at the start):Assuming difficulty goes up each stage, you can fill in the above calculations 5x using declining win rates (80% for 1st satge, 79% for 2nd stage, 78% for third stage etc, 77% for 4th stage and 76% for 5th stage for example).

The calculation will then be 89,6% (stage 1) * (I dunno, 88% * 86% * 84% and * 82% ?) =

+/- 47%giving you the chance of successfully completing all 5 stages given a starting stage 1 win rate of 80%. I'm just too lazy to do the math but it's not complicated. Once you have that number you can calculate any strings of good or bad luck by using probability matrices within a given string of 60 or 70 tales runs.The final question is whether or not DB (the casino) is messing with it's RNG (fixing the tables):Once you have your average theoretical win rate on 1 attempt (5 stages) and assuming increasing difficulty (combined calculated for any easy, hard & epic runs), you can compare your own actual results using distribution statistics and calculate a standard deviation to the theoretical result.

If your actual total results (of 70 Tales runs) are beyond 5 sigma (using CERN threshold statistics for calculating the odds of random noise incorrectly showing the discovery of a new particle), you can then say that DB is probably messing around with it's RNG. Otherwise, we just have to chalk it up to bad luck.

Additionally, one has to consider non quantitative arguments. Does DB have the brainpower to realize additional profits by messing with its RNG (while not even being able to complete simple bug fixes to the core code for years)? And given the recent round of layoffs, what are the chances that a single whistleblower would not rise up and spill the beans if DB was indeed messing with its RNG? And are there not other areas of the game where one could profit far more by messing with the RNG (dragon/epic crafting, AvA etc?). Tales is an ATM for most players, not a large drain on their bank account and by fixing the tables, the casino doesn't necessarily profit more in the case of Tales as additional monies are usually only spent on maintaining Top 50 rankings which are not effected by artificially reducing the overall win rates for all players.

dude, that's harsh totally harsh

lol I'm a pessimistic realist so I put nothing past any corporation out to make money, that said in my case it's not really thinking they are messing with the odds, its considering the rng or at least the implementation of the rng to be flawed (of course I love when I get extremely lucky results, but hate it's guts when it's the opposite), there's simply too many people, doing too many things, all pulling numbers from the same place without any measures in place to ensure balanced results on a personal level.

a server/game wide balance does absolutely nothing to ensure a statistical balance will be reached on a specific players account.

P.S I do

thank you greatly for answering my question, and more so for all the show of work and breakdown explanation that you took your time to include..P.S I am not making that argument, just you know being a bit of a devils advocate.......and

I am not claiming db engages in or has engaged in cover ups(at least I'm not claiming it at this moment),a cover up is far different from not realize/being aware of an issue.P.P.S those aren't in bold because of you @Dutcher, I just wanted to ensure anyone who stumbles across my post don't miss those two points.

Also, I realize 100 is not a huge sample size. But when you take into consideration that I've been recording this data for what must be years' worth of Tales, each with an average of 600-700 rolls (which is not all that statistically different from 100, for comparison's sake), and the expected percentage has never been anywhere close to the corresponding number on the reverse (low) side, it makes you think.

If I usually get around 13-14% of my rolls falling in the 90-100 range instead of the expected 11%, and once or twice managed to get 10.5 or 10%, that's overall a consistent showing of bad luck. But if I get 25% and never once got, say, 5%, that's just unreasonably terrible "luck", if we can call it that.

My two cents on the issue.

And I understand getting only a 50% completion rate on hard isn't fun when you are expecting 75%. 25% of first 100 rolls at 90 or better, same, not fun and even more of an outlier statistically. Still, in the end a normal distribution and standard deviation has to be calculated. If you are expecting 11 rolls out of 100 at 90 or better with a standard deviation of 3, and you're getting 25, that's 4 to 5 times the standard deviation which I would indeed call close to impossible (5 sigma). If the standard deviation is 7 however, you're only at 2x the normal deviation which is really not all that unlikely. That's why you really need to run the numbers before calling it bad luck, really bad luck, downright worst luck ever or broken RNG.

Whatever the reason is in the end, bad luck or broken RNG, I calculated the probability of DB making any improvements in this area to be 0,0000000888% so there's really no point in wasting much time on this. It's all a gamble and we are gamblers, whether the fix is in or not. But I definitely do appreciate the complaints, without any losers there would be no place left to gamble!

I'm not sure I understand why standard deviation is the best way to analyze rolls based on chance. (I'm used to using it for data from real-world examples that have no hard theoretical expectations the way a die, coin, or random 1-100 integer selector do.) I ran my own numbers and got an SD of about 31.8. I checked my math four times and don't see any errors, but I suppose it's always possible. And then just for fun I calculated the SD of this scenario:

80x 100's

20x 20's

And the SD was 32.2. In that scenario, 80% of your rolls are technically in the 91-100 range, and 20% aren't. 80% actual - 10% expected = 70, which is still barely over 2 sigmas. Is this scenario still considered not that unlikely?

I'm no expert in probability analysis, so I'm probably missing something here, but I was just fiddling with the numbers and this just doesn't make sense to me. Maybe you (or someone) can run that same scenario and confirm/refute my results. Or just show me where I'm going wrong, if that's the case.

I'm not entirely sure what you used as the mean (average result) in order to calculate the sum of the squared variances of each possible result and then root the total (= 1x SD). The SD of any roll between 1-100 should indeed be around 34 (1% accounting for each possible result). The mean and SD in # rolls within a certain range however is a different calculation. The mean that should be used is 11 (this is the expected # of rolls of 90-100 if you roll 100 times. The SD from that mean could still be at 30 because of the very high squared variances on one side of the mean (89 squared plus 88 squared etc but again too lazy to calculate). Assuming your numbers and calculation method is correct and you're getting 80 actual rolls out of the 100 rolls at 90 or better and an SD of 2, then you're still in the bottom 5% range of expected results within a normal distribution. This is still a small possibility but not extremely unlikely. 5 out of 100 times you will see this happen. The actual number after doing the complete calculations is probably more like 2 or 3 times out of 100 attempts (x 100 rolls = 10,000 rolls) which to me still seems like a normal distribution of the probability of seeing 80/100 rolls at 90 or better with a mean of 11. I think your calculations are correct, the issue is with interpreting them or seeing the possibility of getting your result as being lower than it actually is.