Book of Heroes

Improve RNG for PvP!

Please for the love of everything holy. Improve the Random number generator. The one now seems to pull numbers out of the sky, it should be fixed on the actual ratings of the two characters in the match.

For instance: (example)
If a player (P1) has a dodge percentage of a total of 35% dodge, an the other player(P2) has an accuracy percentage of 21% the percentage of dodge of (P1) would actually be 14% yet the RNG gets it's numbers from the sky an (P1) dodges 5attacks in a row.

Now say if (P2) has no accuracy it would render (P1) back at a dodge percentage of 35% he would still only dodge 3.5 hits out of 10 hits/ lets just say 6/20 per match (other matches shouldn't count). Yet the player can dodge 10/20 or more... Now with (P2) having the accurcy (P1) should barely dodge at all yet still dodges like a pro from some statistic that doesn't even exist.

I want everyone that reads this to take 4dice an at the same time roll them 20x an see how many times you can roll 17 or better, honestly post how many times you dodged. (Real rolls, dice must shake freely in hand before the throw an must tumble over 3 times no cheating)

I got
9, 23, 7, 11, 14, 10, 17, 11, 12, 10, 16, 4, 9, 10, 16, 16, 15, 13, 17, 13.

Ok now that 3 straight out dodges. Now every 3rd 16 counts as 1 dodge. So a total of 4/20 dodges.

Hit (H) Dodge (D)
H,D,H,H,H,H,D,H,H,H,H,(crit,H),H,H,H,D,H,H,D,H.
(These were all legit rolls) /crit hit because I rolled 4ones vice versa if you roll 4 sixs\

Now add back in (P2s) accuracy, every roll over 21is a hit an every 2nd 20 is a H.... How many times do you land a blow ROLL Yall.

Because I can tell you this. The RNG doesn't have that right statistically, mathematically, functionally... Over half of the outcomes aren't even in the stats to make it logically add up at all.
SüperSäiyan5 wrote:For instance: (example)If a player (P1) has a dodge percentage of a total of 35% dodge, an the other player(P2) has an accuracy percentage of 21% the percentage of dodge of (P1) would actually be 14% yet the RNG gets it's numbers from the sky an (P1) dodges 5attacks in a row.

While I overall agree that the RNG does a horrible job maintaining averages on a match to match basis and even over a short/intermediate period of matches, how you calculated the actual dodge is incorrect.

If dodge is 35% and the opponent has 21% accuracy it does not literally decrease(subtract) dodge by your accuracy, you have to multiply. For example:

35% dodge=.35 21% accuracy=.21

.35 x .21=0.0735 or 7.35%

Now take this new number and subtract it from original dodge stat and you get:

35-7.35= 27.65% adjusted dodge rate after accuracy is taken into account.

Also don't forget that this becomes even lower if slow or ensnare is active.

And finally, this calculation of how accuracy works has been confirmed by the Devs so please don't bother arguing whether it's actually calculated this way or not(not taking into account that I actually did my math right, lol) you'll just be wasting your time.
Well.... This current system is about as good as it gets now everyone does lose the occasional BS match which the other person dodges and Crits everytime. The thing is it goes both ways you get bashed generaly the same amount as you do the bashing (If you have any general skill) So before saying oh our PvP system sucks.... If you look at other cell-phone apps there aren't many other ways to do it

And P.S. You wouldn need a larger sample size for your theory of dice rolls to pvp to be even somewhat presentable
As someone who works in the field, I'm sure they are just using the built-in system RNG and those are as good as they get. With the millions of RNG calls they make, a run of bad/good luck is going to happen frequently. It's just math. You can't evaluate an RNG on 20 rolls. Come back with maybe 10,000 data points and then let's talk. Any sample size smaller than that is not valid for analysis.
First of all, accu enchant is very bad, even if 3rd enchat slot was available, you have to avoid it :)
Second thing is, you have to count all hits to the statistic, with 33,33% dodge chance you should get dodge on one out of three atacks, including the ones you don't care that much about. From JC view, dodge hurts the most on seismic and brutal attacks.
Let's asume a fight(in a perfect world where statictics will be accurate on one match) where you use these hits: wound, brutal, radi light, seismic, brutal, brutal, wound, radi, seismic.
Thats 9 hits, so in perfect world you will get dodged on 3 of them at random. If both yours radi lights and one brutal gets dodged, you dont care, if you get dodged on both seismics and one brutal, you will notice it and be mad about it. So with same RNG result, you get one good and one bad match.

RNG is about rolls, dont use dice, use random number generator(for example https://www.random.org/). Put range 1 to 10000. Now you have to realize that there is no roll history, every move/hit generates new number in that range.
Example:
Lets say you use wound vs opponent with 33,33% dodge and 21% status resist. First you need to hit, so roll 1 to 10000, if the result is under 3333, he will dodge it, so roll again for backswing(60% chance to backswing), if under 6000 you will backswing, if above, you won't.
If the original result was over 3333, you will hit the wound so now you roll for bleed. If under 2100, no debuff, if over, you apply bleed. More rolls are involved for crit/block but this is just a simple example how RNG works.
Now you go for another hit, lets say brutal, same rolls will be applied without looking at the result of the first hit. So every hit have the same rolls with same percentual chance and to see if it works right, you have to collect huge amount of data to compare, which is hard, because there are many debuffs that affects the dodge chance, or even deffing affects it...

Illy_the_Red

ModeratorMon Aug 10, 2015 7:28 am
Sàlád_ wrote:Well.... This current system is about as good as it gets now everyone does lose the occasional BS match which the other person dodges and Crits everytime. The thing is it goes both ways you get bashed generaly the same amount as you do the bashing (If you have any general skill) So before saying oh our PvP system sucks.... If you look at other cell-phone apps there aren't many other ways to do it

And P.S. You wouldn need a larger sample size for your theory of dice rolls to pvp to be even somewhat presentable

There are many tweaks that have been suggested to make things more even and less random, such as moving to different forms of stat resist (% reductions rather than Resist/No Resist) and changing Accuracy so it is actually worth enchanting it. Even at high (31-33) percentages of stat resist, I don't get the feeling I'm resisting every 1 in 3 effects. It's more like resist everything one match, nothing in another, then everything in a later one. It's the same with dodge - sometimes you dodge literally everything and sometimes nothing, even with a SW.

At the minute, unless you're at very high levels, you don't feel like you're doing much more than flipping a coin. Heads, you win. People want more out of pvp than certain classes having advantages over others and flipping coins to get results.
SüperSäiyan5 wrote:I want everyone that reads this to take 4dice an at the same time roll them 20x an see how many times you can roll 17 or better, honestly post how many times you dodged. (Real rolls, dice must shake freely in hand before the throw an must tumble over 3 times no cheating)

I don't understand what you're getting at. The odds of four six-sided dice coming up 17 or better is 23.91%, just shy of a quarter. We don't have to actually run the experiment.
Saraa wrote:
SüperSäiyan5 wrote:For instance: (example)If a player (P1) has a dodge percentage of a total of 35% dodge, an the other player(P2) has an accuracy percentage of 21% the percentage of dodge of (P1) would actually be 14% yet the RNG gets it's numbers from the sky an (P1) dodges 5attacks in a row.

While I overall agree that the RNG does a horrible job maintaining averages on a match to match basis and even over a short/intermediate period of matches, how you calculated the actual dodge is incorrect.

If dodge is 35% and the opponent has 21% accuracy it does not literally decrease(subtract) dodge by your accuracy, you have to multiply. For example:

35% dodge=.35 21% accuracy=.21

.35 x .21=0.0735 or 7.35%

Now take this new number and subtract it from original dodge stat and you get:

35-7.35= 27.65% adjusted dodge rate after accuracy is taken into account.

Also don't forget that this becomes even lower if slow or ensnare is active.

And finally, this calculation of how accuracy works has been confirmed by the Devs so please don't bother arguing whether it's actually calculated this way or not(not taking into account that I actually did my math right, lol) you'll just be wasting your time.

So basically you are telling me that the % offered by accuracy is a lie and that it is a complete joke to even have it in the game... Wonderful who knew this game wouldn't accurately present the odds of what you get out of a enchantment. It doesn't say at all the percentage off your opponents dodge% an than it makes it completely useless if the opponent has no dodge. Way to go programmers, you sure know your way around misleading people and not giving a proper description. The way the skill reads is it effects the ability to dodge.

Kinda like how *shadow veil* doesn't increase the damage of bombs only straight flat damage. Does it say that anywhere in the skill ""!!!!???DESCRIPTION???!!!!" Hell my bad I thought that's what a description was used for...


On the other guys response. Note that dice things was purely for a exaple on the percentage an no one posted their rolls, an seeing how there is less than 20 attacks done in one match you dont need 10000 or whatever number you explained roll neither should the RNG.
apLundell wrote:
SüperSäiyan5 wrote:I want everyone that reads this to take 4dice an at the same time roll them 20x an see how many times you can roll 17 or better, honestly post how many times you dodged. (Real rolls, dice must shake freely in hand before the throw an must tumble over 3 times no cheating)

I don't understand what you're getting at. The odds of four six-sided dice coming up 17 or better is 23.91%, just shy of a quarter. We don't have to actually run the experiment.


All I did was take the percentage of what the chance to dodge was off the total amount 4 dice would make what's left would be the number to beat to be able to dodge. I just wanted to see how close/far apart people's rolls would be and the average hits landed. It was just for curiosity not a serious issue.
For future postes, critiquing my initial post that was purely for rough reference. It is not the point of the post so please if that's all you want to do take it to a post that would care.

Now what this post an posting area is about is what would you like to see improved or changed or added into the game. Regardless how you feel or take in my post on requesting a improvement to the RNG.

Bottom line is it could be better an I would like to see it's improvement (future suggestions). If you don't agree (most would), don't post here it not a discussion. There is another area of the form for that labeled. I am not trying to be rude or disrespectful just straight forward.