Just wondering if the displayed "to hit" percentage is accurate. It seems I very often miss with my bow and a supposed 70%-75% chance to hit. Granted there's still a decent chance to miss, but when this happens three or four times in a row, it feels like more than bad luck (though it certainly could be).

Just wondering if there are factors the game sometimes doesn't figure into the "to hit" chance (elevation differences or terrain, enemy animations and a moving "hit box," more extreme distance to target, using a bow higher than your level, etc.).

Thanks,

Geezer

Not really. All those factors, if applicable, are already calculated into the chance to hit.

The thing is that chance to hit of about 70% does tend to produce some unusually high number of misses.
Ive noticed that way back and reported it, asked about it.

It doesnt seem like its the problem of the game, but rather some sort of weird probability freak result. Then again, if we use the Occam razor logic it would point out in the direction of some weird fault of the game mechanics calculations and math.

I really couldnt say anything more specific about it. We would need a dev to dive into the code and check it in detail... and some probability scientists to explore its murky depths.

I have this half formed theory about one third of chances to hit being chance to miss being somehow significant in the probability mathematics itself... more then usual, but that would take us into more esoteric parts of that science for which i am not knowledgeable enough to make any specific calculations about.

The only thing i can theorize about is that 70% chance to hit is two thirds of the whole probability field, which makes that part of 30 - 33 (35)% chances to miss roughly one third of the whole.... and then...

the math turns into two thirds versus one third?
and that one third somehow is prone to win a bit more?

Weird i tell you.

But then again, probability mathematics and science have always been weirder then anyone expected.

(could be just some small mistake in the equations and their execution in the code though)

I don't think the moving hitbox affects it. Could be wrong, but I believe if you're targeting them when you click, it should hit.

Elevation and corners definitely cause problems with it, as you'll see when enemy archers shoot a hillside or a wall because they don't realise that the arrow won't reach you.

30% to miss is quite a large probability to be honest. Most of the time people consider to-hit chances as "at X tries you will miss Y times". That will be truer(not true mind you) when your sampling universe is quite big. Such as at 10 million hits you will get closer to proposed hit chance but that might not be that visible or accurate over 100 or 1000 hits.

Also the question here is whether the hits are calculated independently or there is a mechanic which increases your hit chances at the background after consecutive misses or not. To make the hit chance more accurate even at lower number of hits some kind of modifications could be done. If they are independent...well you could miss 5 times in a row and get mad which is a very low probability but still within reason. (Vice versa is also possible)

In the end, to-hit chance indicators more like "decision-helpers" for players instead "absolute truths" (unless it is 100%) during the short time you spend on a battle.

70-75% hit chance means that 25-30% of the time you'll miss; you will miss more than 1/4 of your shots; the missing shots should be noticeable.

No, it means that every time you try to attack, your chance of success will be 70-75% regardless of your other attacks.

My mathematics is rusty.

Given a 75% chance to hit, am I right in thinking that four misses in a row comes out at a 0.4% chance? 1 in 250?

Chance to miss = 0.25
0.25^4
=0.004
==0.4%

I think it's a 31.64% chance to occur.

0.75 ^ 4

Edit: Ignore this. I forgot to think.

That's calculating four hits in a row though right, not four misses?

Actually, yeah. Sorry. I calculated it backwards.

So, to the OP, when you say you miss three or four times in a row, how often does this actually happen?

Again, taking the high end of your given to-hit chances, 75%, there's a 1.6% chance that you miss three times in a row and a 0.4% chance that you miss four times in a row.

Are these instances of consecutive misses just occasional happenings that seem more frequent because of how much they stand out, or is it something that's happening a whole bunch?

Very vague numbers coming up but, if you're missing three/four times in a row more than, let's say, once every two encounters, then I'd class that as pretty odd, enough to shed doubt on the advertised 75% to-hit chance.

EDIT: just going through the thread and I don't see it mentioned --

We know the game has a further to-hit penalty based on range from target, right? We're told as much in one of the loading screen tips. What I've not taken any time to research is whether the advertised to-hit percentage takes this extra modifier into consideration. If it does, then everything stated above still stands.

What if, however, the to-hit pecentage is simply a calculation between bow-user and target, worked out before variables like range are taken into consideration? And the game then applies its 'far range' negative modifier on top of that, unadvertised?

A way to test might be save a game at a convenient point in a battle and then start moving the bow-user faaaar back, and up close, see if the advertised to-hit changes. Or, alternatively, checking the to-hit percentage from way far back, popping a perception potion (decreases far range negative modifier) and seeing if the to-hit% is still the same.

Waaaaait a second. Hold up.

Which to-hit value are you reading? The one from your character sheet, given under Offence Rating, or the one from targeting an enemy with a high-Loremaster character?

EDIT: sorry, I'm an idiot who should go get more coffee. You are, of course, using the Big Obvious To Hit%,



which does take range between attacker and target into consideration, as well as the offence and defence ratings of attacker and target. I'm a dumbo.

If i'm not just unlucky hit chances seem to lower than what it says.

For me it's like:

90% = 70%

70% = 50%

50% = 25%

First, thanks for all the thoughtful, well-reasoned and helpful replies. I appreciate the analysis.

I'm a bit amazed nobody has brought up Tom Stoppard's wonderful "absurdist, existentialist tragicomic" play Rosencrantz and Guildenstern are Dead. It's arguably most memorable scene has them quite humorously discussing probability and the law of averages (start @55 seconds on youtube Rosencrantz and Guildenstern "Heads"). In short, Gary Oldman's Rosencrantz (there's an excellent movie version) keeps getting "heads" over and over in his flips of the coin while they (Tim Roth as Guildenstern) effectively muse about the seeming conflict between every individual toss having a 50-50 chance vs. the aggregate unlikelihood of getting one constant result over time (Oldman says he gets "heads" 78 times in a row).

Priceless, and entirely on topic...



Whether the game takes distance into account, as one response postulated, is precisely the kind of "hidden factor" I'm wondering about. It would be absolutely silly for the game not to reveal such, but it would help explain why, as Zerkain listed, 90% feels like 70%, 70% like 50%, 50% like 25% and so on.

I haven't been playing my ranger long enough to claim anything near 1,000 shots, or even 100. But I noticed this "serial missing" enough during last night's session to prompt me to ask. Noaloha's notion of replaying my most recent encounter (where I did miss four times in a row, as I had previously -- instead of "heads" picture me saying "no way" over and over) is compelling (since I saved just before), so I'll have to try that to see if the results are different.

Any further comments are most welcome.

And, yes, the whole Rosencrantz and Guildenstern are Dead is very much worth watching, if you are so inclined.

Thanks,

Geezer

In my case - constantly. From beta, over every patch, through playthrough after playthrough.

And im not talking about long range weapons or attacks. Im talking about my rogue standing right next to an enemy and trying to strike and missing all of the attacks in the turn at 73% (most often) throughout the game, repeatedly over and over and over.

Only sometimes, one of the hits connects.

As soon as i see that to hit chance i already know all the strikes will be misses.

I didnt count, but generally speaking, out of 10 turns - in each having three, four or more attacks (high dex, high speed character builds) i miss all attacks in about 7 or 8, while the rest give some kind of middle results. Or just one hit.

It actually feels as if my chance to hit is 33%, not my chance to miss.

Strangely, of course, i hit more times in a single turn when i have 50-ish or 60-ish chances to hit.
And maybe i hit better with high strength then with high dex characters.... but i cannot be certain of that. Madora built by default skills tends to miss her hits at 73% too. A lot.


- I am quite use to RPGs, RNGs and missing even with high chances. Its not that and if it was only me i would have chalked it down to just me having extreme bad luck -




I have actually not known that existed.
Much appreciated and thank you.

From what you say Hiver, I suspect something is wrong. That seems to be just too many misses to put in the bad luck category. The math is probably correct, though, since formulas are quite easy to verify. The only thing I can think of in that area is if the random number generator is being initiated again and again with the same seed, but that is a rather amateurish mistake.

Instead, I suspect it has something to do with graphic animations and mouse clicking. Maybe the game somehow considers the mouse click a miss even if it looks like a hit. If so it may be useful to notice when you miss and when you hit. Ranged vs. melee, front vs. back or side, etc. There is an issue with the animation as enemies may move away just as the clicking is done. Maybe there is a delay or something so that the mouse click is not in synch with the animation. Does it matter where on the enemy you click, etc. A systematic approach trying to answer those questions may prove useful to find any errors in the game.

Personally, I have not often encountered the too many misses error. But I have had moments where I felt like something was wrong. I try to click in the middle of the target and sometimes I wait to see what the animation does in order to find a safe spot to click no matter what the animation is doing. It seems to help since I have not felt like being treated unfairly by the random generator for quite a while now.

I haven't really noticed any strange behaviour there, but that doesn't say much.
Probabilities are devious things and distinguishing a streak of bad luck from a real error in the RNG (or the displayed numbers) is almost impossible without some very methodical approach.

So , if someone has a convenient save and lots of time at hands, how about doing the same fight (or better yet, attack) a couple hundred times?

I don't know if everyone has already read up on what I'll call the "hit box funkiness" of D:OS. I'm not even sure whether this issue was resolved or not (last post I read on it was from mid-July).

So, while I've no idea if this is helpful, here is what I gathered:

- At least up until fairly recently, enemy animations would also cause their "hit box" to move -- even while "standing still" -- and so folks could have their cursor on the enemy at one moment only to have it slip off them the next.
- From what I read, this resulted mostly in the character moving to an adjacent spot next to the enemy rather than attacking it (because when the mouse click activated, the cursor wasn't on the enemy, but rather on the ground...enemy hit box had shifted). This, of course, explains "mis-clicks" but not missed attacks.
- One particularly frustrating outcome of this "hit box funkiness" specifically pertained to backstabbing. Because D:OS is finicky when it comes to positioning the dagger-wielding backstabber "just right" behind the intended victim (at which point a special backstab cursor appears...a fist clenching a dagger), the hit box could shift enough that the attacker executes a regular attack but not a backstab. So no super-excellent backstab bonus damage.

Several counters to the above "hit box funkiness" have been recommended:
- Instead of mousing over the foe, mouse over their portrait in the initiative display at the top of the screen. This guarantees you attack the intended enemy and makes hit box positioning irrelevant. I did read that this may not be possible with regular/auto attacks...which I think leaves backstabbers out of luck.
- Try turning on enemy markers and using the top-down view in combat. Apparently, the former keeps the hit box steady and the latter makes precisely targeting the foe's "back" (in some cases on foes that don't actually have backs, like plants) less of a guessing game.
- Target low, closer to the enemy's feet (if they have feet...again, plants). Apparently, this base area of the hit box is less likely to move.

I think that covers "D:OS, hit boxes and you." Again, this may be useful for folks who end up clicking on the ground and thus moving rather than attacking as intended, and backstabbers who understandably want to execute a backstab rather than a regular attack, but it seemingly has nothing to do with attacks with an advertised high percent chance to hit missing way too often. That's not a "mouse thing" but a game mechanics/math thing.

- Geezer

It has nothing to do with the issue of hit boxes and animations and miss-clicks caused by all that.
I have no issue with that because i simply take my time in the TB combat instead of wildly clicking around like im in shooter.

Besides - if you missclick and your character starts moving, wasting APs - you can stop the movement of the character and save most APs by just right clicking.


I have to say that i also didnt encounter too many consecutive misses - except when the chance to hit is around 73%. Its some kind of a sweet spot.

My advice:
Any time you see 73% chance to hit, save the game.
Then attack. Record the result.
Reload your save, and attack again.
Repeat at least 20 times.

If clinical observation confirms your suspicion, then post a bug report.

20 times is not a great sample size, I recommend atleast 1000+ preferably 10.000.

With kind regards,

Rashar.

I wanted to suggest something which would be physically possible.

Well 1000-10.000 times is physically possible aswell. I do understand what you meant, though 20 times is just not enough to be able to say that something is wrong/right.

With kind regards,

Rashar.

This is my first post and I specifically registered here for this issue as it was driving me insane in my first playthrough.

I played COOP, Main Singlehanded with shield pure Man-at-Arms, Madorra pure Man-ar-Arms with two-handed, my buddy went lone-wolf glass-cannon mage (not that it matters for this issue, just to give you an overview).

What I noticed is that even when my main-character or Madorra stood surrounded by 4 enemies with a displayed chance to hit of 100% (was higher, if not internally capped thanks to precision stance and blessing), Flurry would miss 1 out of the 4 hits almost always. Also Whirlwind and Dust-Devil where almost certain to miss at least one of the 4 enemies (remember: all of them had a displayed hit-chance of 100%).

Now, granted, I'm only an IT-guy and therefore not extremly high skilled in math, but so far I took a 100% chance to do anything for granted my whole life ^^

tldr: Not even a displayed chance to hit of 100% is really 100% chance to hit.

While I have no reason to assume that you are not telling the truth, I find it hard to believe.

Mostly because of the following 2 reasons

1: In 3+ playthroughs I did notice anything like that at all. And don't get me wrong I had some misses with auto-attacks aswell sometimes missing 2 times in a row while having 82% to hit, but that happens. I'm an old school d&d player and during my years I have seen a lot of lucky or unlucky stuff happen. In every case something weird happens I would put my money on variance. (which does not mean that you are wrong, I'm just not willing to give you the benefit of the doubt)

2: Skills like flurry, whirlwind, and dust devil don't require a to hit roll and when executing these skills you don't see the to hit %.

I am not saying that I am right/you are wrong, however if this really and I will quote you: "almost always" happens then it should be easy to reproduce and get it on video.

With kind regards,

Rashar.

I can absolutely understand your doubts as I would probably say the same if I had never encountered this (and it really doesn't make any sense as I do believe the responsible devs are kind of skilled at what they do).

All I can tell you is that it happened for me and it did happen often enough to be noticeable and drive me nuts.

Regarding your 2. point (Skils and Hit rolls): I have no idea how exactly their hit-change is calculated. While you're right about not needing to roll for the skill itself, the skills do have a have a chance to hit / miss the target. So far I was under the impression that this hit-chance was the same as the shown chance when hovering over a target with the mouse, but maybe I'm gravely mistaken here and skills have a different hit-chance calculation than auto attacks. That actually would explain a lot.

Anyway, in my next playthrough I'll try to catch it on Video.

A logical diagnostic approach. I'll do it. Until results are in, avoid the number 73 in all forms (lottery tickets, sports uniforms, house numbers, the 73rd floor of extremely tall buildings, etc.) -- this phenomenon could be universal . Unless of course...you want to miss.

- Geezer

Question. What stats can a baddie have that could counter our % to hit? And if that exists is that factored into the displayed To-Hit %? Or is this part of the overall question to Larian, "Does the % to hit display factor in all your reasons to hit and all the enimies reason to evade?"

Side note: As we know Magic is very strong in this game, crowd control probably being the main strength. I ultimately edited skillsdata.txt to add in some real %'s to miss and for CC affects to miss more often, to me it was just too easy to make statues of the enemies and just pick them off. Also added reduction in rounds of how long CC would last as well. To me a better game this way.

Shields, the shield specialist skill and Defence rating (affected by Dexterity) factor into it, I believe.

Horrorscope: Try playing a game without magic. It's a lot of fun and gets rid of that whole "destroy everything effortlessly" thing. At least until you leave the Cyseal area.

Repeating it 10,000 times would take approximately one entire day - 24 hours - of constant testing.
That is 3 work-days straight of doing nothing but pressing two buttons and then recording the results.

Maybe someone on earth would voluntarily do that, but I cannot fathom that person.

This might be a flawed approach, depending on how the game generates its random numbers. If they actually do a random number generation at the time of the 'dice' roll, then fine. But many games use a pregenerated random string of numbers so reloading won't necessarily affect a 'dice' roll.

This game does not employ a fixed random seed. So his method would work. It'll be extremely time-consuming and boring, but it'll work.

The method is already applied. Its called playing the game.

btw i have noticed similar thing happening in other games. When it comes to repeated consecutive misses in one turn, 71 to 73 % seems to be some universal sweet spot. Or a freak spot.

Its not so often as in OS but it is noticeable.
Currently being observed in Age of Decadence and wasteland 2.


Have we discovered a flaring fault in the Matrix?
And there are experiments trying to do the same wasting millions currently going on. Hey guise, you can do it for 39,99 you know?

INTERESTING FIND -- LOREMASTER REVEALS DIFFERENT CHANCE TO HIT!

Last night I was playing after my main character's Loremaster score had been boosted more than a few levels by gear (perhaps now at L4, maybe even L5). And when I "moused over" the foe while holding down CTRL I noticed something intriguing in the stat box displayed via Loremaster. While the attack cursor reported a 98% chance to hit, the enemy's stat box said "Chance To Hit: 61".

There was no "%" sign after the 61, but I think it's safe to assume that's 61%. Why such a difference between the two numbers (98% and 61%)? I have no idea. But it'd be interesting to find out.

- Geezer.

I'm not sure of other factors but I can confirm that the chance-to-hit shown in combat does not take block chance into account. One of my characters has a 70% chance to block, but the chance to hit shown is 78-84%.

Loremaster reveals their to hit chance...

With kind regards,

Rashar.

a blocked shot isn't a miss; it did hit, but being blocked. very much like damage absorbed by armor; it's just that a successful block avoids all damage.

I know that probably is a b.... and that with 80% CTH you can still miss 10 times in a row and it doesn't have to be a bug.

But - from experience - this means that anything combat related with a chance of success below 90% is unreliable to the point of being mostly useless. 70% = completely useless. Not just in D:OS. In any game.

So here is a question I'm having trouble answering. My main character has 16 str, 2 pts in 2h weapons, and is using a flaming 2h sword. Sebille has 17 finesse, 2 pts in ranged, and is using a crossbow. The Red Prince is using a dagger and shield, has 11 str and 10 finesse, 1 pt in 1h weapons. I was having no trouble yesterday, was fighting source hounds, magisters, undead, and other random creatures. Today, Sebille and I can't hit anything, worst streak was 16 misses on targets listed at 100% advertised chance to hit. Red Prince hasn't missed. Ever. We are 1-2 levels above the targets, weapons are all level 4, and we are fighting largely the same targets. I'm barely past the fort playing on classic.


Edit: Reloading a save seems to have fixed it for me. Trying to slog through battles with half the party unable to hit the broad side of Rivellon makes for an interesting challenge.

This is a D:OS topic, not D:OS2 btw. Blind seems to have lasting after effects, solvable by reload.