I did address your issues.
That would take redesigning your simulation model to account for them. You haven't done that. I mean to myself but I haven't done it yet either.
On the hp totals - I used 5th level PCs, and the expected HP ranges I used are built around the hp totals you'd expect to see in CR 6 or lower. They were perfectly reasonable hp ranges to focus on, and they covered a broad spectrum due to the randomness.
In the HP distribution you provided what are your chances for seeing an average hp of each cr of monster? I say it distributes hp to much toward the extremely low CR's for level 5 PC's. But how can we actually discuss this properly without at least working out the answer to the question I proposed just now? So let's establish the answer to this question so we can talk reasonably and logically about your methods.
If I am right and your distribution distributes the hp range too much toward the lower ranges, fixing that will have a big impact on how much overkill damage you are seeing from the sim.
Also, if the last blow ENDS an encounter, overkill is meaningless, but only a small fraction of attacks end an encounter. By my records, the number can be anywhere from an average of 12 in some parties to 40 in others. Other than that, so long as you can get to another target to complete your attack sequence, you get to use it - and it is incredibly rare that PCs are unable to get to the next target.
I don't disagree with the high level point that few attacks end an encounter. I also agree with setting the pc's ability to get to the next enemy at 100% provided there is a next enemy to get to. How meaningful is breaking the fights into encounters though? If you assume your fighter gets all the killing blows (as you did in your sim) then the fighter will only have overkill apply to n-1 of the enemies instead of n enemies.
Which is to say in 1 enemy encounters the overkill effect is reduced to 0%.
In 2 enemy encounters it's reduced to 50% of whatever your sim would have calculated it as.
In 3 enemy encounters it's reduced to 67% of whatever your sim would have calculated it as.
In 4 enemy encounters it's reduced to 75% of whatever your sim would have calculated it as.
I consider 4 enemy encounters to be the average and so I'd estimate the actual overkill effect due to encounter style combat vs your endless simmed encounter to be 75% less. That's a significant reduction in the effect you are touting and it's not the only reductive effect either.
These two minor quibbles do not offset the MASSIVE differential I established with the simulation.
Not fully no. My position is that overkill has a minimal effect not no effect. These factors I'm discussing show your simulation estimated the impact of overkill damage to be significantly higher than it actually is.
You don't like the results because they contradict your position. However, they prove that overkill is a significant factor in the efficiency of a PC.
I'm willing to accept being wrong provided results produced from sound methodology reveal that to be so. I don't like your sim results because the methodology used in them skewed the results toward the conclusion that the overkill effect is more significant than it actually is.
And, once again, putting the simulations aside - I've recorded combats in 5E quite often and analyzed what went on to evaluate whether certain feats were actually overpowered, to look at how much of an impact overkill has, and for a variety of other reasons. My simulation and the recorded combats at the game table resulted in the same conclusions.
If we can show the sim methodology is biased toward inflating the impact of overkill and you get the same results with actual recorded combats then I'd suggest something is amiss with those results as well. Possibly small sample size. Possibly a DM that favors lower CR encounters. etc. All I know is that if the assumptions in the sim for ac, hp, etc were accurate enough that you would see a much lower effect of overkill in actual play because the sims numbers are significantly inflated as disucssed above.
