Encounter Damage Range (EDR) build analysis


Advice


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber

I calculated the encounter damage range (EDR) of LVL 2 ranger builds using ten thousand players over a bakers dozen encounters of four rounds.

% STR+1, DEX+4 flurry ranger agile finesse dagger
% draw, draw, hunt
% move, twin, strike,
% twin, strike, strike
% twin, strike, strike
ATK = [8 6 4 8 6 4 4 8 6 4 4];

vs.

% STR+4, DEX+1 outwit ranger longsword
% draw, hunt, move,
% strike, strike, strike
% strike, strike, strike
% strike, strike, strike
ATK = [8 3 -2 8 3 -2 8 3 -2];

What percentage of 1H longsword players are as bad as 2x dagger players?

This can be determined by plotting the encounter damage ranges vs. AC, and determining where the sigma bounds of the two builds intersect.

AC %

12 0.1
15 2.3
17 16
19 50
24 84

Despite the longsword always having the better average damage, fewer rolls of larger die causes more performance variation.

At AC19 encounters the longsword does 23+/-11 while the dagger does 17+/-6

So even though the longsword is third better average, its accuracy is nearly twice as bad. With fewer rolls hitting less you get more performance variation despite the larger die and bonus.

So at very easy AC you can certainly say the dagger is not a choice, but at very hard AC many if not most will find their 'better' longsword is just as bad. So for when I run out of arrows before we get to the big boss - I know my lighter daggers are likely to be doing just as good as your longsword - and it allows me to carry a lot more arrows. So don't talk to me about your 'better' DPR, I am using my more powerful longbow (need to implement deadly to simulate) for the easy warmup fights.

I actually calculated with percentile equivalents of sigma, I verifed histogram is normal gaussian so it rounds to same number for large simulations. But when you start missing badly, then the 0 damage clips the histogram from being normal - so percentile works better.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber

chart link

https://drive.google.com/open?id=1wOZZfZFMLpAYbkUuCTkL4YJdXQSsdNFH

It is interesting that if you view this as the boss sees you as the easy AC, you see how the critical successes really start piling on the damage, the lower the AC the more crits. You see a clear breakpoint in the DMG/AC slope as it switches from all hits being single damage to all hits being double damage.


2 people marked this as a favorite.
Pathfinder Lost Omens, Rulebook Subscriber

Why did you use the non-combat Hunter's Edge for the Longsword versus the Flurry for daggers? Flurry on the longsword or precision would be a much fairer comparison.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber

Because I was trying to test if the d8 could overcome the lack of twin and agile and flurry.

Yes closer comparison where only difference is agile for better MAP vs. the d8 is interesting- I would expect there to be a higher AC wher e there longsword is always better.

But first here is another close comparison. Here is the starknife vs. the dagger (both twin flurry agile) so the only difference is deadly d8 on the martial weapon.

https://drive.google.com/file/d/13iOQnIxUVua153alBVT5CZnSmq8ooi6s/view?usp= sharing

Above DC18 only the nat20 can be critical success, and I was very suprised how little difference extra crit damage made. Only the top few% of players will notice the few damage, everyone else their luck washes it out.

Critical effects really only show up at easy AC, at AC10 only 15% of the starknifes are not noticing the extra damage. But that is why getting crit with critical effects hurts, because the boss sees you as easy AC - looting the bosses martial weapon you will not do as well because you are fighting hard AC.


4 people marked this as a favorite.

I think people would find your arguments more relevant, if you compared two properly built characters, instead of contriving a scenario that's purely academic.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber

% STR+4, DEX+1 flurry ranger longsword
% draw, hunt, move
% twin, strike, strike
% twin, strike, strike
% twin, strike, strike

ATK = [8 5 2 2 8 5 2 2 8 5 2 2];

vs. the same dagger so only difference here is the d8, no agile, and extra strike.

This longsword build is always better until the harder DC19, then by the incredibly hard DC26-31 half the longsword players are doing as bad as all the dagger players (but this is 2nd level so that is anyways a deadly encounter)

Now the drawback of course is 2 bulk vs. 0.2 bulk but that should not be a problem for a strength ranger, it will be an issue for DEX ranger who already has to pack their bow and quivers plus they lack a STR bonus.

Though now I am interested is the deadly long bow compared to the dual wield longsword - previous post deadly only helps at easy AC, but it would get an extra twin (hunted) shot because you do not need to move into melee (ignoring blocked shots)


Pathfinder Lost Omens, Rulebook Subscriber
krazmuze wrote:

Because I was trying to test if the d8 could overcome the lack of twin and agile and flurry.

Well then it doesn't seem a fair comparison. If you are trying to disprove the idea of the long sword being the 'better' weapon then you need to compare the dagger user to an equally optimised longsword weilder. Otherwise all you've shown is that someone taking the skills ranger build with a longsword is still viable when compared to a combat build.

Always make the opposition arguement as strong as you can envision


Malk_Content wrote:
krazmuze wrote:

Because I was trying to test if the d8 could overcome the lack of twin and agile and flurry.

Well then it doesn't seem a fair comparison. If you are trying to disprove the idea of the long sword being the 'better' weapon then you need to compare the dagger user to an equally optimised longsword weilder. Otherwise all you've shown is that someone taking the skills ranger build with a longsword is still viable when compared to a combat build.

Always make the opposition arguement as strong as you can envision

I don't think he's trying to demonstrate that any actual build is better or worse. I think he's just trying to show that his statistical analysis, and how he can quantify the reliability of a given build.


krazmuze wrote:
So for when I run out of arrows before we get to the big boss - I know my lighter daggers are likely to be doing just as good as your longsword

On an 18 Strength Ranger, Longsword does 1d8+4=8.5 on average and Dagger does 1d4+4=6.5 on average. You're doing 30% more damage with a Longsword, whatever the combat conditions. If for you this 30% damage difference is ok, then it's ok. But there is no world where Daggers are competitive to Longswords on a Ranger.

To be accurate, you have 0.5% chance to deal more damage with Daggers than Longswords after 10 hits. If that's what you wanted to show.


SuperBidi wrote:
krazmuze wrote:
So for when I run out of arrows before we get to the big boss - I know my lighter daggers are likely to be doing just as good as your longsword

On an 18 Strength Ranger, Longsword does 1d8+4=8.5 on average and Dagger does 1d4+4=6.5 on average. You're doing 30% more damage with a Longsword, whatever the combat conditions. If for you this 30% damage difference is ok, then it's ok. But there is no world where Daggers are competitive to Longswords on a Ranger.

To be accurate, you have 0.5% chance to deal more damage with Daggers than Longswords after 10 hits. If that's what you wanted to show.

The point of Krazmuze's calculation is to show that variance is a big factor into calculating such things. Which is why this 30% you're mentioning is because of an infinite number of rolls, which doesn't happen in reality at all.

The point he's trying to make (coming from the other thread) is that variance is a huge factor that normal DPR calculations don't take into account. Which is why, with dice variance, this "30% more damage more" is subject to the variance and may overlap. If the issue is purely damage, then yes, the sword hits harder, but you still need to hit which is why the dagger can be a good option as well and not just an inferior choice.


3 people marked this as a favorite.
Lightning Raven wrote:
The point he's trying to make (coming from the other thread) is that variance is a huge factor that normal DPR calculations don't take into account. Which is why, with dice variance, this "30% more damage more" is subject to the variance and may overlap. If the issue is purely damage, then yes, the sword hits harder, but you still need to hit which is why the dagger can be a good option as well and not just an inferior choice.

He's adding variance where there is none.

If you compare the damage one person with a dagger does and the damage someone else with a longsword does, his calculation is correct.
But you don't calculate DPR to brag about your damage (well, I hope so). You calculate DPR to optimize YOUR efficiency. So you compare YOUR efficiency with a dagger to YOUR efficient with a longsword. As such, you don't care about hitting because the attack bonus with a dagger and with a longsword is the same, so if you hit with a longsword you hit with a dagger and if you don't hit with a longsword you don't hit with a dagger. The only variance between the two is the variance in damage, as one is using a d4 and the other one is using a d8. And after 10 damage rolls, you have 99.5% chance to deal more damage with the longsword, period. No need to make it more complex than necessary. The overlap is extremely small, and you don't need more than 3 fights to get to whatever sigma level you want.


The variance is coming from the dices themselves and how they can overlap, which makes the small increases and other optimizations not noticeable during the span of the campaign, which can make it easier to realized that the less optimized choice is not actually that bad and the dice rolls will influence as much as will your smaller dice or slightly lower bonus.

For instance, using agile weapons will make a difference if everything else is the same simply because they will hit more often after the first attack, making up for the fact that a longsword deals more damage.
It's just showing that DPR calculations isn't as accurate as people seem to evaluate it, specially now that this system have other factors to take into account.


1 person marked this as a favorite.
Lightning Raven wrote:
The variance is coming from the dices themselves and how they can overlap

Yes, but there are errors in his way of calculating. Events are not separated. You can't say:

- I roll a d20 and if I hit with my dagger I roll damage.
- I roll a d20 and if I hit with my longsword I roll damage.
- I compare both damage value.

The proper way of doing it is:
- I roll a d20.
- If I hit with my dagger I roll damage.
- If I hit with my longword I roll damage.
- I compare both damage value.

So, the variance is incredibly lowered, because the result on the d20 is the same for both weapons as we are calculating our own DPR, not comparing 2 different person's damage.


Why are you comparing a simple weapon to a martial weapon? For fair comparison it should be either Morningstar or shortswords for a more fair comparison, simple weapons are intentionally designed to be noticeably worse than martial. Or if you want to keep d4's you should probably be doing main gauche v. rapier as both have similar damage to trait trade-offs as opposed to doing longsword to main gauche when one is low-damage with a lot of traits and the other is high-damage with few traits within their respective niches.

Also, the 1-handed weapon user should absolutely have a combat edge or else the two-weapon user should not have a combat edge.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber
SuperBidi wrote:


To be accurate, you have 0.5% chance to deal more damage with Daggers than Longswords after 10 hits. If that's what you wanted to show.

I am doing histogram analysis but with more players to remove the noise from analysis, for a levels worth of actions comparing different build options and report that histogram over the average encounter. How costly is it to not use an action or not take that feat? Critical success and weapon effects, increase crit chances and bonus die so not as simple as the weapon die histogram. The more attack and dmg rolls you make, the narrower the histogram spread.

The first comparison was not intended to be fair, it was to point out the DPR average can mislead you - the DPR for the longsword is always better in this case - but histogram analysis showed actual encounter experience is only always better for easy AC. More crits from more hits (flurry agile) made the dagger viable for bosses.

Yes the later more fair comparison shows that the dagger having an extra draw and less dmg hurt it until you get into deadly bosses where those agile bonuses mattered for getting more hits and crits.


Lightning Raven wrote:
The variance is coming from the dices themselves and how they can overlap

No, it's coming from the fact that the dagger build has Agile and Flurry, while the longsword build doesn't, and both builds are attacking as much as possible. Therefore the longsword build is going to be less reliable because it's making a bunch of attacks at -10. I'd be curious what how the reliability would change if the longsword only made two attacks per round instead of 3.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber
SuperBidi wrote:


So, the variance is incredibly lowered, because the result on the d20 is the same for both weapons as we are calculating our own DPR, not comparing 2 different person's damage.

Sure if you want to look back at a past campaign and say this build would have performed better than that build for the same rolls. But to achieve the same number of rolls you can look at builds that only swap simple weapon and do not impact number of strike to claim this. Even just switching to a martial weapon with a critical die will change your dataset of rolls. This is the butterfly effect.

But that is not why people are reading build analysis. They want a guarantee that if they take this build that they will be better than the other player. They want to know is their build always better regardless of both players luck, or is there an even or slight chance the other player does better than them with a 'worse' build.

We know this because people take build advice, then get into arguments at the game store/forums that they do not see this or that build as viable as the DPR analysis claimed - that they have been doing better at their table with an 'inferior' build.

Keep in mind this first analysis showed that despite the none optimal longsword build DPR always being better, that this will not be all players experience if they are not having easy fights.

Statistics histograms show you the range of luck, it tells you if your are always, sometimes or rarely better than the other build.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber
Strill wrote:
No, it's coming from the fact that the dagger build has Agile and Flurry, while the longsword build doesn't, and both builds are attacking as much as possible. Therefore the longsword build is going to be less reliable because it's making a bunch of attacks at -10. I'd be curious what how the reliability would change if the longsword only made two attacks per round instead of 3.

I will get more hits with the dagger which changes my luck spread. It is a different dataset with more rolls that hit for the dagger because of the build. So future rolls do not quantify my build performance, because changing the build changes the number of rolls. This is the butterfly effect.

For high AC you can calculate from ATK vs. AC if it will ever hit without doing any histogram and can conclude it is a wasted attack. But it would be interesting to see the histogram for never crit hit but still can hit/miss - how much extra dmg does it do.

I certainly was surprised that statistically it shows the boss should take a martial weapon over simple weapon for those extra crit dmg on your easy AC, but it does little for you fighting the hard AC boss. It really demonstrates how lopsided critical success is.

I bought home matlab so I would stop taking up work time, so not responding with more data does not mean I am not interested in detailed fair comparisons....

Someone else is writing a full blown tool that uses histograms, and I have been giving feedback on how I was doing analysis.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber

So to make it less confusing to histogram analyze I am thinking of not doing the average +/-1,2,3 standard deviations. I do not have the percentiles memorized so it is hard to think the analysis.

Since I was already using percentile math because the histogram is non-gaussian for the edge case approaching zero encounter damage - I can easily switch to [5 50 95] percentile to correspond to the d20. This will be slightly less than two sigma bounds.

The range shows 90% of players performance.

if significant overlap half the players using this build are doing as bad as most of that other build

If minor overlap only those players that always crit hit will do as well as my 'better' build with their worse 'build'. Those that always crit miss will do as bad as the 'worse' build.

If the bound gap is far enough (~1.5x the bound) then you can estimate the 'better' build is always better.


krazmuze wrote:
We know this because people take build advice, then get into arguments at the game store/forums that they do not see this or that build as viable as the DPR analysis claimed - that they have been doing better at their table with an 'inferior' build.

If someone is going to take an anecdote into a DPR analysis discussion, and present it as meaningful information, then I highly doubt they have the background or wherewithal to understand or pay attention to anything involving statistics.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber
Paradozen wrote:

Why are you comparing a simple weapon to a martial weapon? For fair comparison it should be either Morningstar or shortswords for a more fair comparison, simple weapons are intentionally designed to be noticeably worse than martial. Or if you want to keep d4's you should probably be doing main gauche v. rapier as both have similar damage to trait trade-offs as opposed to doing longsword to main gauche when one is low-damage with a lot of traits and the other is high-damage with few traits within their respective niches.

Also, the 1-handed weapon user should absolutely have a combat edge or else the two-weapon user should not have a combat edge.

I proved that the good dagger build is as good as the bad longsword build, when DPR says the longsword is the better choice. The intent was to show how DPR is misleading, not to say this was the most optimal weapon build.

Yes it is more interesting to use the analysis to make more closely optimized comparisons - but then I would been analyzing the build choices rather than analyzing if histogram can show why DPR without deviations is misleading.

I personally found the starknife vs. dagger run very interesting, because you cannot conclude that the martial weapon is always deadlier.
Instead it shows that when you have a hard AC that only nat20 is a crit, that nearly all players are not going to see the benefit of it. That extra damage on the crit is being washed out by their luck on the other hit/miss. But that martial weapon on a boss used on your easy AC - that extra crit damage is going to put you down fast!

The other optimizations you discuss are interesting to me, but I cannot compare everything to everything!

I want to check how much the spread is for the +1 kukri vs. simple dagger because it is Plaguestone lvl1 boss loot. Same number of strikes for same build but +1 and d6 upgrade - will everyone that takes that upgrade do better or will it get washed out by luck by facing lvl2 severe that you suck at hitting anyways.


krazmuze wrote:
Paradozen wrote:

Why are you comparing a simple weapon to a martial weapon? For fair comparison it should be either Morningstar or shortswords for a more fair comparison, simple weapons are intentionally designed to be noticeably worse than martial. Or if you want to keep d4's you should probably be doing main gauche v. rapier as both have similar damage to trait trade-offs as opposed to doing longsword to main gauche when one is low-damage with a lot of traits and the other is high-damage with few traits within their respective niches.

Also, the 1-handed weapon user should absolutely have a combat edge or else the two-weapon user should not have a combat edge.

I proved that the good dagger build is as good as the bad longsword build, when DPR says the longsword is the better choice. The intent was to show how DPR is misleading, not to say this was the most optimal weapon build.

What do you even mean by "as good as"? How did you prove that both builds are as good as one another? You proved that the longsword is less reliable, but it's still better on average, which is all I care about, because I play for the long-term, and am indifferent to short-term variance.

Heck, your own numbers don't even support the idea that both builds are as good as one another. The longsword even has better bad-case performance, with a damage of 12, vs the dagger's damage of 11. Meanwhile the longsword has a much much better average-case, and best-case performance.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber
Strill wrote:
krazmuze wrote:
We know this because people take build advice, then get into arguments at the game store/forums that they do not see this or that build as viable as the DPR analysis claimed - that they have been doing better at their table with an 'inferior' build.
If someone is going to take an anecdote into a DPR analysis discussion, and present it as meaningful information, then I highly doubt they have the background or wherewithal to understand or pay attention to anything involving statistics.

Sure that is true for a table, because all those experiences are anecdotal with a sample of 4, but the stats can tell them why their good build is sucking compared to the bad build. Or they can blame it on the dice and put them in jail.

But when there are thousands of players on a forum talking about it - it is no longer anecdotal. At 100 players the histogram is still noisy, at 1000 it smooths out, at 10000 rounding error for reporting purposes becomes the noise, at a million I was getting percentiles matching sigma at ten thousandth place - which shows the histograms are normal gaussian.

So you can use statistics - I would certainly hope that the devs did this to make sure that if they intend the option to be better for everyone that will take it - that it is. Likewise if they intend most people can make flavor choices and not notice the small build differences, then that makes for a better game and the fact that someone who crit hit/miss all the time can do better/worse than nearly all - that just makes it more fun


krazmuze wrote:
This longsword build is always better until the harder DC19, then by the incredibly hard DC26-31 half the longsword players are doing as bad as all the dagger players (but this is 2nd level so that is anyways a deadly encounter)

I think you have bigger problems than build DPR if your enemy has an AC 30 at level 2 :p

So not sure what the takeaway is here. You reinforce that good builds are important, as you show that a bad longsword build loses a lot of damage potential, but other than that your conclusions seem to be the same as if you looked at average damage. The longsword usually does better against at-level enemies. Both weapons come closer together in damage against higher ACs. Average DPR against at-level VS higher ACs would show you the exact same thing, and this isn't to do with your variation being added, right? Or am I mistaken? (to be clear, I am genuinely interested in what I can learn from your analysis, I just don't see the lesson through the data myself)


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber
BellyBeard wrote:


Average DPR against at-level VS higher ACs would show you the exact same thing, and this isn't to do with your variation being added, right? Or am I mistaken? (to be clear, I am genuinely interested in what I can learn from your analysis, I just don't see the lesson through the data myself)

For that analysis it tracked that the longsword is always better for any reasonable encounter - the varation never reached the point that it could be question other than unreasonable AC.

For other builds that are closer together then that crossover AC where it is always better, halfway better, or only better for the top few %?

Look at the deadly vs. non deadly dagger for that example. It really shows how lopsided critical success is. The deadly dagger is not going to do anything for you taking on the boss, but the boss is going to take full advantage of that deadly dagger.

The reason is there are so few crits at higher AC that the variation of the crit die that do occur get washed out with the base die variations. It is the gambler fallacy - there is not enough crits for you to be reliable with it. This is because you never see anything but the few nat20 crits but the boss is getting a lot of AC+10 crits. Only the top few % of players are lucky enough to see the few extra points from deadly.

The DPR would say the deadly dagger is always better.

So getting an always better for desired AC range using histogram analysis is good because it says it is a robust build that anyone can use and always do better. Other builds maybe more subject to histogram spread overlaps that it depends on the target AC.

I think the charts are confusing with all the +/-1,2,3 sigma bounds. It will be easier to see the AC breakpoints if I just do the 5% outlier bounds and median.


krazmuze wrote:
So getting an always better for desired AC range using histogram analysis is good because it says it is a robust build that anyone can use and always do better. Other builds maybe more subject to histogram spread overlaps that it depends on the target AC.

OK, I think I am starting to understand. Thanks for explaining to me.

So the idea is, sometimes when people discuss DPR, they might say something like "the DPR of both choices is really close, so either option works". But then there's an ill-defined boundary where the DPR is no longer close enough to be equivalent. And your analysis helps with that, by showing at what point a difference in damage or modifiers actually starts to matter. Does that sound accurate to you?

If that's the case, then I like the idea and agree it adds a valuable dimension to the build analysis. It is what's needed to ultimately make decisions between builds. For example, it can let you decide whether giving up 10% average damage is worth a +1 bonus to AC, by telling you when that bonus damage would be most useful and when you could pass it up. I think we can intuitively say the bonus damage would be most useful if you hit often, but this can actually put numbers to it.


1 person marked this as a favorite.

I tickles me to read these types of threads.

I really wonder how many people lose sleep over this silliness(my opinion)


Ten10 wrote:

I tickles me to read these types of threads.

I really wonder how many people lose sleep over this silliness(my opinion)

The first step to making your own system or additions to a system is to understand such systems. Or at least it should be.

In other words, I like homebrew, and I like well thought out homebrew and systems even more. So understanding how things work, and how things compare to each other, is important.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber

For me understanding how critical success makes the system play very different has been worth it, coming from last several D&D which has simpler crits are natural 20. If you want heroic fantasy your GM should just trivial minions against you, if you want gritty danger every minion is instead a boss and they have a boss and they have a boss. In D&D 5e a minion army can TPK, while the boss needs stacked with all kind of legendary actions/resistances and lair actions to not be boring.

It should be noted that I am no longer simulating a levels worth of damage, as I realized that encounters are more prey to short run wide variances. Simulating a levels worth of damage the short run good and bad luck washes out, but in real encounters those turn into wins and losses.

You think you are doing fine at winning the luck game then the boss triple crits you with high constant damage because he won initiative. Does not matter that your good luck in previous rounds enabled to kill the boss on your next turn no matter what.

So I am now doing duel simulations to see who wins the most.


BellyBeard wrote:
Ten10 wrote:

I tickles me to read these types of threads.

I really wonder how many people lose sleep over this silliness(my opinion)

The first step to making your own system or additions to a system is to understand such systems. Or at least it should be.

In other words, I like homebrew, and I like well thought out homebrew and systems even more. So understanding how things work, and how things compare to each other, is important.

Of course, of course...

Wait a minute, if that is the case why does krazmuze use the wording "the longsword build is always better" for?

If you want to truly compare things that matter. Compare metal dice to resin to plastic to wood to elevation to temperament at the time of rolling to group make up to age of players to paper vs plastic miniatures

krazmuze wrote:

For me understanding how critical success makes the system play very different has been worth it, coming from last several D&D which has simpler crits are natural 20. If you want heroic fantasy your GM should just trivial minions against you, if you want gritty danger every minion is instead a boss and they have a boss and they have a boss. In D&D 5e a minion army can TPK, while the boss needs stacked with all kind of legendary actions/resistances and lair actions to not be boring.

It should be noted that I am no longer simulating a levels worth of damage, as I realized that encounters are more prey to short run wide variances. Simulating a levels worth of damage the short run good and bad luck washes out, but in real encounters those turn into wins and losses.

You think you are doing fine at winning the luck game then the boss triple crits you with high constant damage because he won initiative. Does not matter that your good luck in previous rounds enabled to kill the boss on your next turn no matter what.

So I am now doing duel simulations to see who wins the most.

Couple things.

1. You not realizing encounters are important is vastly telling....
2. The boss triple criting, tpking the party is a statistical not even rounding error but look at what happened to the group.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber

1. Tell that to DPR optimizers who have long ignored that the encounter is over before they achieved their average damage with reliability. All I was doing here was adding a deviation to their same average damage number.

2. Actually at low AC that happens all the time, because every roll is a crit. You would be referring to other editions of the game where only natural 20 is a crit.

Which was the point of my post was how much more impact the critical success mechanic has, it is why it is easy for the boss to TPK the party.


krazmuze wrote:
1. Tell that to DPR optimizers who have long ignored that the encounter is over before they achieved their average damage with reliability.

I think you don't understand that DPR optimizers (as you call them) don't care about achieving their average damage. What they do care about, on the other hand, is that their DPR difference actually influence battle. You don't need more than 3 fights to feel a difference of +1 on a d20 (which is the smallest difference you can get in PF2 system). So, optimizing DPR has immediate and visible impact.


Ten10 wrote:
BellyBeard wrote:
Ten10 wrote:

I tickles me to read these types of threads.

I really wonder how many people lose sleep over this silliness(my opinion)

The first step to making your own system or additions to a system is to understand such systems. Or at least it should be.

In other words, I like homebrew, and I like well thought out homebrew and systems even more. So understanding how things work, and how things compare to each other, is important.

Of course, of course...

Wait a minute, if that is the case why does krazmuze use the wording "the longsword build is always better" for?

I don't know. But that's what I'm in this thread for. :^)


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber

For lvl2 STR+4, WIS+3 d8 fighting 4x4 against elites, you can win 30 to 10% of the time from AC0 to AC18, which then improves to 35% of the time by AC28.

For lvl2 STR+3, WIS+4 d8 fighting 4x4 against elites, you can win 10 to 0% of the time from AC0 to AC18, which then improves to 40% of the time by AC28.

If both sides are heavily armored/shielded then prime stat initiative is better than secondary strength, a slog with few if no crits, and more misses than hits. The elite bonuses are not enough to keep you from winning.

If lightly armored then prime strength is better then secondary initiative, mostly because constant damage crits are going to win the battle.

Yes I know those AC0/18/28 are silly extremes it is used to understand the breakpoint trendlines for interpolating realistic results inbetween.

This is mixed initiative for 10k fights, still need to deviations for lower number of fights, and for side initiative. Replace DEX for WIS if doing stealth initiative.

Yes 4 elites is beyond deadly, but the point was to see if where you put the +1 made any difference.

But for 3 fights the only possible results is won 0, 1, 2, or 3 fights so your win records can only be observed in 33% steps - thus is it is not possible for you to observe the lesser difference between these options. If you had said over a level then I would say probably if lightly armored. (25% of the <AC18 the STR+3 did as good as STR+4 in a random sample for each AC),


If you're trying to adjust for encounter value, and doing so by having builds constantly only full attack...

I don't think that analysis will ever be relevant.

Especially because:

Quote:
But that is not why people are reading build analysis. They want a guarantee that if they take this build that they will be better than the other player. They want to know is their build always better regardless of both players luck, or is there an even or slight chance the other player does better than them with a 'worse' build.

The analysis that always full attacks will never interact with this philosophy closely enough to be reliable information.


1 person marked this as a favorite.
Pathfinder Lost Omens Subscriber

these are really unimportant statistics

i'm glad i just noticed this thread...

krazmuze wrote:
1. Tell that to DPR optimizers who have long ignored that the encounter is over before they achieved their average damage with reliability. All I was doing here was adding a deviation to their same average damage number.

DPR is a heuristic for determining if your build could be a bit better, it's not something used to know how a fight will go. you can use it when looking at live damage versus DPR(as a heuristic) to see if you need to pull out your big guns or not, but beyond that, it's not a "live" thing.

you can also use it to look at DPR in different situations to see how your character should try to drive the fight. like rogues who's DPR is needed from flatfooted, and so you should look into the easiest way to reliably get flatfooted.

like how is your DPR when chasing someone who's faster? how is your DPR if you go for a 2 move then grab versus 2 move strike, etc.


Bandw2 wrote:

you can also use it to look at DPR in different situations to see how your character should try to drive the fight. like rogues who's DPR is needed from flatfooted, and so you should look into the easiest way to reliably get flatfooted.

like how is your DPR when chasing someone who's faster? how is your DPR if you go for a 2 move then grab versus 2 move strike, etc.

One example I'm interested in, for example, is Sly Striker. If you have to spend 1 action, 2 actions, to get flat-footed, how does that compare in viability to standing there with Sly Striker? What if you have only 2 actions? What if it's due to you using Feint, with or without the scoundrel racket?


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber
Cyouni wrote:

If you're trying to adjust for encounter value, and doing so by having builds constantly only full attack...

I don't think that analysis will ever be relevant.
...
The analysis that always full attacks will never interact with this philosophy closely enough to be reliable information.

You can see with changing AC of the fight is changing the successful number of attacks, +1 STR means you got a 5% jump at attacking, whereas +1 INIT means you got a 5% jump at going first. Not much by itself but that 5% results in +1 on constant damage (+2 on crits). So yes DPR says exactly that that hit is improved damage without even rolling the d8.

But the fewer hits the fewer damage rolls, the more likely that you will fall out of DPR being observable and drifts back to random chance. This is what is happening when you start winning more against heavily armored when you are heavily armored, the modifier did not matter - luck did.


Pathfinder Adventure, Adventure Path, Lost Omens, Pathfinder Accessories, Pawns Subscriber
Cyouni wrote:
Bandw2 wrote:

you can also use it to look at DPR in different situations to see how your character should try to drive the fight. like rogues who's DPR is needed from flatfooted, and so you should look into the easiest way to reliably get flatfooted.

like how is your DPR when chasing someone who's faster? how is your DPR if you go for a 2 move then grab versus 2 move strike, etc.

One example I'm interested in, for example, is Sly Striker. If you have to spend 1 action, 2 actions, to get flat-footed, how does that compare in viability to standing there with Sly Striker? What if you have only 2 actions? What if it's due to you using Feint, with or without the scoundrel racket?

In the other thread I did 2 vs. 3 actions, since these are all MAP-10 there was no difference, but I would expect 2 vs. 4 actions with better MAP matters, otherwise not enough hits in that last attack for modifiers to matter, only luck matters. Now using that last action to shield, or flatfoot, or whatever else? But to get to that point I have to do the base case of all attacks just to makes sure things are working, debugging things when you change all the parameters is not fun!

That is where I think encounter based analysis is very useful to see what the tradeoffs vs. full attacks are - especially if the GM uses the same tactics against me.

But the point of doing the simulator was to recognize when the number of rolls is low that DPR vanishes and you trend towards uniform randomness.

To do it right you would have to write a video game worthy combat engine to deal with all the corners of the math and know all the rules.

So I will probably just check things that pique my interest for rules I am familiar with what I have played - that is why I did what happens if put strength point into initiative instead?. For my flurry twin ranger going first is more important to me because I know the next level is all about the bosses. Not worth dying over and reincarnating as a STR fighter to fix my strength.


Pathfinder Lost Omens Subscriber
krazmuze wrote:
Cyouni wrote:
Bandw2 wrote:

you can also use it to look at DPR in different situations to see how your character should try to drive the fight. like rogues who's DPR is needed from flatfooted, and so you should look into the easiest way to reliably get flatfooted.

like how is your DPR when chasing someone who's faster? how is your DPR if you go for a 2 move then grab versus 2 move strike, etc.

One example I'm interested in, for example, is Sly Striker. If you have to spend 1 action, 2 actions, to get flat-footed, how does that compare in viability to standing there with Sly Striker? What if you have only 2 actions? What if it's due to you using Feint, with or without the scoundrel racket?

In the other thread I did 2 vs. 3 actions, since these are all MAP-10 there was no difference, but I would expect 2 vs. 4 actions with better MAP matters, otherwise not enough hits in that last attack for modifiers to matter, only luck matters. Now using that last action to shield, or flatfoot, or whatever else? But to get to that point I have to do the base case of all attacks just to makes sure things are working, debugging things when you change all the parameters is not fun!

That is where I think encounter based analysis is very useful to see what the tradeoffs vs. full attacks are - especially if the GM uses the same tactics against me.

But the point of doing the simulator was to recognize when the number of rolls is low that DPR vanishes and you trend towards uniform randomness.

To do it right you would have to write a video game worthy combat engine to deal with all the corners of the math and know all the rules.

So I will probably just check things that pique my interest for rules I am familiar with what I have played - that is why I did what happens if put strength point into initiative instead?. For my flurry twin ranger going first is more important to me because I know the next level is all about the bosses. Not worth dying over and reincarnating as a STR fighter to...

he was talking about sly striker which gives 1d6 precision damage against people who aren't flatfooteded that you hit or critical hit. when would it be better to move for a full sneak attack versus just going to hit. should you move twice to sneak attack another enemy that you could flank and attack or just attack the guy you're at twice and move away? etc.


Bandw2 wrote:
he was talking about sly striker which gives 1d6 precision damage against people who aren't flatfooteded that you hit or critical hit. when would it be better to move for a full sneak attack versus just going to hit. should you move twice to sneak attack another enemy that you could flank and attack or just attack the guy you're at twice and move away? etc.

Here's a quick DPR analysis comparing 2 attacks with no flat-footed bonus against 1 attack with flat-footed bonus, at level 1 and at 8 with sly striker. That 1 attack option can be either move to flank and attack, or feint and attack against a target already in melee. It shows that 2 attacks loses out at level 1, but with the Sly Striker feat it gets very slightly ahead at 8. This doesn't take into account the critical specialization of rapier, which makes the target flat-footed; that would further push the 2 attack option ahead for the Sly Striker, but only very marginally.

Looks like Sly Striker is a consolation prize more than a regular tactic, which is expected.


Pathfinder Lost Omens Subscriber

while i have no horse right now in sly striker i do wonder how the level 14 effect changes this (doing 2d6 on non-flatfooted)


Bandw2 wrote:
while i have no horse right now in sly striker i do wonder how the level 14 effect changes this (doing 2d6 on non-flatfooted)

Though I would have to run the numbers, I assume it probably puts it very slightly ahead in that analysis. Of course, as a rogue starting every turn adjacent to the enemy is probably not what you want, since that likely means they're attacking you. :) The move -> attack -> move one will be set up for that sequence much much more frequently, since the only requirement is that you start your turn within one Stride of the enemy.


1 person marked this as a favorite.
SuperBidi wrote:
krazmuze wrote:
1. Tell that to DPR optimizers who have long ignored that the encounter is over before they achieved their average damage with reliability.
I think you don't understand that DPR optimizers (as you call them) don't care about achieving their average damage. What they do care about, on the other hand, is that their DPR difference actually influence battle. You don't need more than 3 fights to feel a difference of +1 on a d20 (which is the smallest difference you can get in PF2 system). So, optimizing DPR has immediate and visible impact.

I call BS on that. That +1 is exactly one number on the d20, that means everyone at the table has to be rolling that exact number multiple times over every encounter in order for the (as you call them) 'feel the difference'.


Ten10 wrote:
SuperBidi wrote:
krazmuze wrote:
1. Tell that to DPR optimizers who have long ignored that the encounter is over before they achieved their average damage with reliability.
I think you don't understand that DPR optimizers (as you call them) don't care about achieving their average damage. What they do care about, on the other hand, is that their DPR difference actually influence battle. You don't need more than 3 fights to feel a difference of +1 on a d20 (which is the smallest difference you can get in PF2 system). So, optimizing DPR has immediate and visible impact.
I call BS on that. That +1 is exactly one number on the d20, that means everyone at the table has to be rolling that exact number multiple times over every encounter in order for the (as you call them) 'feel the difference'.

You're forgetting criticals,it affects 2 numbers on a d20


Ten10 wrote:
I call BS on that. That +1 is exactly one number on the d20, that means everyone at the table has to be rolling that exact number multiple times over every encounter in order for the (as you call them) 'feel the difference'.

Multiple times over every encounter? That's not feeling the difference, that's rolling a d3 for your attack rolls.

A +1 to hit is nearly equivalent to 15% overall damage. That's the level of difference you can feel without having to roll dice for ages. You can tell me you can't feel a 15% difference in damage, and I would agree that it's around the lowest difference a human can feel. But it's also the smallest bonus you can get, a +2 to hit is felt at each and every combat (+25% damage is equivalent to power attacking in PF1, you clearly feel the difference). So, you don't need big differences in DPR to have immediate and clear effects.


2 people marked this as a favorite.
SuperBidi wrote:
Ten10 wrote:
I call BS on that. That +1 is exactly one number on the d20, that means everyone at the table has to be rolling that exact number multiple times over every encounter in order for the (as you call them) 'feel the difference'.

Multiple times over every encounter? That's not feeling the difference, that's rolling a d3 for your attack rolls.

A +1 to hit is nearly equivalent to 15% overall damage. That's the level of difference you can feel without having to roll dice for ages. You can tell me you can't feel a 15% difference in damage, and I would agree that it's around the lowest difference a human can feel. But it's also the smallest bonus you can get, a +2 to hit is felt at each and every combat (+25% damage is equivalent to power attacking in PF1, you clearly feel the difference). So, you don't need big differences in DPR to have immediate and clear effects.

BS again.

Statistical spreadsheets are not actually rolling a d20.

Community / Forums / Pathfinder / Pathfinder Second Edition / Advice / Encounter Damage Range (EDR) build analysis All Messageboards

Want to post a reply? Sign in.