Cinder, the average of your numbers is 10.14 and the median is 10. Each line avg and median:
It doesn't seem to be an overall average that's the issue, but consecutive rolled values being identical far more often than the odds would suggest. For example, the odds of rolling three numbers out of six the same is one in 400. (The first number being the target number and 1:1 odds). Out of 120 numbers, 14 shows up 13 times (10.1%), which is more than double the odds. Other significant appearances are 20 at 11 times, and 13,17, and 19 only showing up 3 times each, which is half the chance.
I have two issues to bring up.
There has been a consistent impression of bias in the dice roller towards extremes and repeated results, frequently supported by data like the above, leading several of my players to complain about their rolls not being really random. If it was just a few isolated cases, I would simply chalk it up to the whims of chance nd move on, but this has been a recurring issue that has been ongoing for years. I would request that any potential causes of this with the RNG code be looked at.
Secondly, the setup is rather clunky when it comes to complex rolls. I would request that it be replaced or supplemented by direct input. For instance one of my characters rolls 2d8+2d6+1d6+15 in damage for certain attacks. As it is, this ends up taking up four lines of the code generator (Rolling for multiple rounds causes some inputs to bleed over), kn has to add it together himself, and the static number must be noted and added seperately. If on the other hand I could simply type in the above sequence on a single line and select how often the roll is repeated, all that mess could be avoided.
Edited: daishain on 2nd Aug, 2018 - 5:03pm
It isn't the same because rolling multiple dice makes getting average results statistically more likely since there are more possible combinations of dice numbers that add up to numbers in the middle of the possible range. In Daishain's example of 2d8+3d6, the average result of these rolls is 17. But there are many, many more ways to reach the number 17 than there are to reach the maximum damage result of 34. In order to get a 34, you HAVE to roll (8,8,6,6,6) which is pretty unlikely. But if you want 17, there are many more combinations that could give you that result. You could have (4,5,3,2,3) or (8,1,2,5,1) and so on.
While raising the floor might eliminate the minimum from being an impossible value, it doesn't eliminate the fact that rolling more dice smoothes the average damage result to be closer to somewhere in the middle. I hope that makes sense. Khan Academy has a few great videos on dice probability. Also any statistics video or course on combinations and permutations etc.
Alright, you want better final results with your rolls. I do not know how many times these special combination rolls are required for Dungeons & Dragons 5e but what I can do is create a row like this (Pictures as row or in a line) as an example:
Amt of times roll D4 + bonus +
Amt of times roll D6 + bonus +
Amt of times roll D8 + bonus +
Amt of times roll D10 + bonus +
Amt of times roll D12 + bonus +
Amt of times roll D20 = Result
Or just have one bonus field at the end and you add all the bonuses since the bonus does not change.
Unless this is something more complex where for example each D6 roll must have its bonus added per roll and then added together - if so the current format works better for that.
The other issue of the resulting rolls can be a min and max check on each roll. For instance, if "1" is it really 1, let us roll a lower value than median again to see? If 20, is it really 20, let us roll a higher value than median?