The "Please read this post first" link in the first of these links incorrectly. Players can replace the en in the link with forum or they can just go here.
The "Please read this post first" link in the first of these links incorrectly. Players can replace the en in the link with forum or they can just go here.
During this discussion there has been talk about monsters auto miss on 1 and auto hit on 20. And there has been reference to the MotU update where the current armor class and to hit system was introduced.
The discussion about the MotU changes is found here.
Important in that discussion is the following statement: Unlike monster attack rolls, player to hit rolls will be mapped to a d20 by rounding to the nearest 5% - if you hit on a 13, you’ll hit on a 13.
Read that carefully. What it says is that player attack numbers are mapped to d20 but monster attack numbers are not. This effectively means what I am observing -- that there are no auto misses or auto hits for monsters.
When dealing with monsters attacking characters the game determines if the monster misses. If it does not miss, then it hits.
This makes the game logic what I suspected earlier:
If dodge then miss
else if incorporeal then miss
else if concealment then miss
else if armor then miss
I really do think that the data I am accumulating supports this conclusion about how the monster to-hit is working.
There were 102 attacks recorded. If we pre-load our bins we would assign 5 misses and 5 hits before even looking to see it anything else happened. That leaves 92 attacks to resolve.
Dodge was 23%. We would now expect the number of dodge misses to be 92*.23=21. The actual number of dodge misses was 23.
Armor was 67%. We would now expect the number of armor misses to be 92*.77*.67=47. The actual number of armor misses was 51.
Which is more likely, that there are auto hits/misses and we undershot by 2 in one case and overshot by 4 in the other OR that there are no auto hits/misses and we were exactly on target in one case and just missed by 1 in the other?
Although this just addresses the second test, we can make the same comparisons using the first test's data as well.
We would expect dodge to give 97*.23=22 misses. Dodge gave 24. Like the second test, this is higher than expected.
We would expect incorporeal to give 97*.77*.1=7. Incorporeal gave 8.
We would expect concealment to give 97*.77*.9*.2=13. Concealment gave 14.
We would expect armor to give 97*.77*.9*.8*.67=36. Armor gave 39.
In both tests the number of all misses observed is consistently higher than expected if there are auto hits/misses but consistently on target with what is expected if there are no auto hits/misses.
It is an inescapable conclusion that there are no auto hits/misses for monsters.
I've tested 160 AC on EE, without any other defenses
I got hit 19/20 on an average
Let it rest Therigar, I know AC SHOULD do a bit, but on EE, it doesn't. I know you want it to work, but on EE, it doesn't
Let it go
Ainevek: 9/6/2 FvS/ranger/paladin (life 2/?) Shinweng: 8/5/2 monk/wizard/paladin (life 4/3?Abaranda: 18/2 FvS/monk (life 7/?) Kevenia: 6 sorcerer (life 2/?)
Two handed chruchers, Ghallanda
If we auto load 1 and 1 then we have 28 attacks to resolve. If we auto load 2 and 2 we have 26.
Expected dodge is 28*.23=6 or it is 26*.23=6. Actual results, 10.
Expected incorporeal is 28*.77*.25=5 or it is 26*.77*.25=5. Actual results, 9.
Which is more likely, that expected numbers are almost half the observed numbers OR that there is no auto hit/miss.
In this case the observed numbers are higher than expectations in any case. But, like the other examples, they are highest when we presume auto hit/miss.
This adds more weight to the conclusion that auto hit/miss does not apply.
This is the same thing that Teh_Troll challenged me with regarding AC. The difference is that he accused me of lying and manipulating the data. I won't do that. I accept that you did your test and that you saw those results.
Only, nobody else can see them.
So, if it is alright with you (and, really, even if it isn't) I am going to continue to run this test and to attempt to ferret out how this system is actually working.
What you can do, if you are already convinced, is stop reading the thread.
Personally, I think that would be a mistake because I think that we are near to demonstrating that various amounts of armor do result in misses on eElite content.
Nobody who has claimed to have tested has presented their tests for review by the community. We are accepting their claim of 5% miss simply on their say so. But in most of the cases these claims are not actually based on testing but on casual observation of on screen displays.
There is no comprehensive comparison of on screen displays with combat logs, no verification of the observations, nothing to demonstrate that those "tests" and conclusions have validity.
Keep in mind that it is very easy to make claims in an open forum like this. It is something entirely different to put material in front of the forums readers and to invite them to go through the process for themselves.
I've posted 3 videos thus far that you can look through for yourself. You can use pause/play and stop at each hit and each miss to see exactly what caused the hit or miss, you can count the exact numbers yourself.
I am spending 2 or 3 times looking over the same video doing the counts to make sure that the numbers I get are as accurate as I can.
Nobody else is offering this. It is a much bigger assumption on your part to take what they post in the forums as accurate than it is to take what I am posting and what you can verify for yourself by looking at the videos.
Again, I give credit to Teh_Troll for this. What is said in a post without providing a way for others to check it means nothing. If you can look for yourself then you can see if I am making this all up or if I am telling you what is going on.
The video evidence argues against the 5% miss. To the contrary, it argues for no auto misses and no auto hits for monsters at all.
I think that I will be able to discover if this is true. But, you are correct, it is just speculation at the moment. The inescapable conclusion right now is that there is no auto hit/miss. But, that conclusion might still be wrong. And that is because I don't have a character that I can take into the content and manipulate the AC up and down to such a degree that I can get a good test out of it. The conclusion is based on what I have to work with at the moment.
Think about what that means. I would have to level a character to L20 in order to get into epic content, I'd have to saddle it with no dodge, and I'd have to be able to fix it so that it went from 10 AC all the way through 150+ AC.
And we've been discussing this now for only a few days.
I don't even think it is possible to create such a character as I describe in order to really do that sort of testing. Neither do I think it is necessary.
Because I know the order in which defenses are checked -- and I am certain it is dodge, incorporeal, concealment, armor -- and because I can limit the types of defense to dodge and armor only that means I can test AC by having a character that will have a range of AC values.
First I have to get that AC to a high enough point that I observe additional misses. Then I have to back it down until I see that the number of misses is consistently the same -- at least on a % basis.
Then I'll know.
It could turn out that it is 5% miss regardless. But right now the data does not favor that conclusion. To the contrary, as tests are repeated the result should be that the number of dodge, incorporeal and concealment misses fluctuate above and below the expected numbers. If they remain consistently above those then it will argue that the auto hit/miss is not taking place.
This is the point of a statistical analysis. With standard deviation there should be data points above and below the expected result, and those should be within a known range. Any time the data point is outside that range we should question the result -- what anomaly occurred to explain this unexpected result?
I think we will get there. But it will take some time.
Last edited by Therigar; 09-15-2013 at 10:30 AM.
Can someone please take a character with <40 AC and no Concealment or Dodge or Displacement into an EE and screenshot a miss in the combat log and post it here?
I would, but I feel this is Teh_Troll's fault and he should be made to do it.
Khyber: Aggrim (Completionist!)
All of my builds are grossly out of date. Just roll a human or drow mechanic / assassin rogue thing.
Blind insta-kills floating eye balls.
A couple of links:
The order that they are checked has absolutely no impact on the usefulness of AC. At all. No matter how many times you claim that it is important, it really isn't. 1.00*.90*.75*<miss from AC> is exactly the same as 1.00*<miss from AC>*.75*.90. The number of times that people see 'dodge' or 'miss' float across the screen does not really matter. People are not rejecting AC at high level and high difficulty quests because of some perception of how they are getting hit. They are rejecting it because it is much more costly in terms of giving up other build aspects than the other tiers of defense are, and because it scales down with your foes to-hit value, which Dodge, Displace/Blur, and Incorporeal do not. What you should be trying to test is precisely what to-hit value EE monsters at certain points have, which would be relevant to the usefulness of AC, rather than the order of operations that things go through, which is not actually important.
Certainty Levels are very important when you are talking about any kind of test like this. People generally strive for either a 95% certainty, or a 99% certainty in their results. And a sample size of 200 is not going to give you a very reliable test case.
I really do think that the "AC doesn't matter in EE" and "AC150 <insert ever increasing numbers here> gets hit on 2+ anyway" and other statements of that sort are based on people running content and taking hits and never even looking at what actually happens. I really do think that they see "Blurry" or "Dodge" but so seldom see "Miss" that they just assume that it is a miss on 1 event. I really do think that it is important to explain this in order to debunk all of the claims that are not actually based on any testing but are just "what I observe" types of posts.
At the moment I'm up to almost 350 events for at level testing. The trend is towards supporting the conclusions that the checked order is dodge, incorporeality, concealment, armor. I'm willing to move on to eElite testing with just dodge and armor (because I can't really do much about my character's dodge value) if that is the direction that we should be going.
This presumes general acceptance of your premise that the objections to AC have something other than a perceptual basis. I am aware that there are some who are now saying that AC is just too hard to get while the others are easy.
And, you make a good observation when you note that the other factors are constant values while AC has a sliding value based on monster difficulty.
But, I don't really think these are the reasons why people make the claims they do. If it were they would be more honest and say things like "AC would matter if I could get there."
In any case, thanks for your observations. Worth thinking about on my part with regards to where to go with the testing.
This is a really interesting thread/argument.
I get the point that mathematically, because the overall miss probability is the product of the constituent "DICA" probabilities, their order is commutative, and therefore mathematically irrelevant, but that the fact that AC, it seems, is checked last, makes it less perceptively less significant in game.
I'm intrigued by AC of 150 too...
Can someone point me at a thread that shows how to reach extreme levels of AC, and what levels of AC become statisitcally significant enough to give miss chances of say 10%, 20%, and 30%, in EN, EH, EE, or (accetping I've over-simplified here) similar such?
Coz that would be pretty useful!
I wonder if the monster to hit formula is the same as the formula used for player to hit.
For example, do monsters get the automatic +20-25%(whichever it is) to hit bonus for proficiency? I would assume that all mobs are considered proficient with whatever they attack with.
I've heard people say that most of the statistics of mobs are based on a formula which basically extrapolates them from their cr rating. How does that explain what must surely be absurdly high to hit values, if having an ac of 160 has no value whatsoever. If that's accurate, what must monster to hit be? +200? +300? More? Why aren't their ac and saves equally that absurd? I know people say that the saves are too high, but if they were proportional to their apparent to hit values, then no attainable dc would be high enough to work. You could level drain them a dozen times and still only land 5% of the time.
If mobs had an ac of 160, you would not be hitting them very often. At best maybe 50% of the time. Probably not even that much.
I'm just curious how this all works.
Last edited by Encair; 09-15-2013 at 02:00 PM.
Shahang Nezhat Bellezza Wipekin Farida of Ghallanda
For monsters it is AB+10.5/AC*2 where AB is the monster's attack bonus and AC is the player's armor class.
So a monster with an attack bonus of 100 would hit a player with an armor class of 80 110.5/160=.69 or 69% of the time and the armor would be giving a 31% chance to miss.
For players the formula is slightly different because it is adjusted to a d20 result.
Just look above for the post with the links in it for full details or use this quote from the original post:
Last edited by Therigar; 09-15-2013 at 01:48 PM.
Lets assume, you are correct. Optically, you would be right - if AC is checked last, players would optically assume, that AC sucks. Mathematically, since protection layers stacking is always the same, despite the order in which it is shown, this should be very easy to verify in dimenzions: Worthwile - totally useless.
Lets assume 100 Hits. Lets assume mob without any form of true seeing, incorp breaking etc. (I.E. majority of them). lets assume 20% dodge, 20 blurr, 30 incorp. Calculating from the top of my head: 100/20% is 80. 80/20% is 64. 64/30% is 44 or 45. Now, if we take in consideration the fact, that there should be 5% miss chance for any creature, it is 42. 42 hits that will pass your defences. Math in this is really easy and you can apply it on your real toon (with different numbers I assume) while you will be getting your butt kicked by CR 45 (or any epic elite in general) monster. Just calc the estimated number of hits you SHOULD receive after all defenses but AC are broken, add 5% to it after those were destroyed and if your 150 AC char will be getting hit about the right number of times (+-3 hits as that is the tolerable amount of miscalculation), then AC is not worthless only optically, but in reality as well. If you will be getting less hits then you estimeted number +-3, then AC is a factor.
That said - I dont really think you will get the result you are looking for.
Since you have the AC fixed you can plug this into a graphing ap to see what hit chances relate to what attack bonuses.Code:(Attack Bonus + 10.5) / (Armor Class * 2)
This means that in order for an AC of 150 to be hit 95% of the time, the monster's to-hit would have to be +275. I do not think that the EE Stormhorns trash mobs have that high of to-hit values. The last time that I saw someone post real to-hit value data was when eGH first came out, and it was estimated that the trash giants had around ~140 to-hit. That would give that 150 AC a roughly 50% chance of being missed, by EE GH Tor trash.
Advance that to the newer, higher level quests. I have not seen anyone actually do anything to try and calculate what the new quest's to-hit values are, but given the rate that stats were inflating let's use +230 as an example. That would give 150 AC just under a 20% miss chance, or equivalent to a blur spell/gear. All of the build and gear choices to get 150 AC come out to about the same as a level two spell, or very commonly slotted item property.
Again, I would point out that the 230 to-hit was a very loose guess based on how monster to-hit values ramped up from the low level epic quests to the higher level epic quests, and I do not know of any good attempts at estimating the EE Stormhorns to-hit values. It could be that the rampant stat inflation that occurred from level 20 to level 24 quests has been evened out as they move from level 24 to 27 quests. Without a good idea of what the new EE to-hit values are, a more meaningful discussion on the actual value of AC can't really be built on anything solid.
EDIT: As to the discussion branch about 5% auto-miss/hit, I really wish that I could effectively search the old pre-beta MotU stuff. I can not find any of the relevant posts, but I had thought that the developers said that Players were going to retain the effect of 'natural 1 always misses/natural 20 always hits' rolls, but NPCs were not because they were not doing any kind of d20 roll facade (the number rolled on the screen means less than most players realize) for monsters? My understanding was that they use straight percentages.
Last edited by RedHost; 09-15-2013 at 04:27 PM.
This form's session has expired. You need to reload the page.Reload