
enthusiast

enthusiast
Joined: Jun 2020

^ Pretty awesome! That's certainly a more professional way to do it, I like. Was there anything else? I think some had some suspected that the probabilities displayed on spells/saving throws were wrong. From the numbers as such I just checked, they seem alright  I just tried to cast Bane and Charm Person with Shadowheart on a bunch of goblins with CHA 8 each (1). The DC of her spells is 13, so for the necessary CHA saving throw they needed to roll a 14 each. When casting Bane, one goblin actually succeeded in rolling exactly a 14, the others not. This would be a 35% chance for the Goblins to succeed, ergo a 65% chance of success for Shadowheart. edit: Unsuccesful throws should arguably been shown in the log too. This currently also applies some for concentration checks, only if concentration is broken, is the saving throw shown (or is it actually vice versa?). I'd also strongly urge to make nodamage rolls not look like misses in some way, because misses they aren't.
Last edited by Sven_; 22/10/20 12:23 AM.




apprentice

apprentice
Joined: Oct 2020

Was there anything else? I think some had some suspected that the probabilities displayed on spells/saving throws were wrong.
I check from time to time whether the displayed hit percentage match the actual role difficulty. And so far, including advantages, they seem to display correctly. Not so sure on (spell) saving throws for monsters. I can't (or don't know how) examine the enemy's saving throw proficiencies. But when taking educated guesses, they seem to be alright. I also advanced in the game some more and updated the google sheet. Had a couple of bad rolls for high percentage chances, which "tainted" the results (I am sure in the long run this will disappear again). But currently the 9095 range only succeeds with a rate of 79%... The other high percentage ranges seem fine though. The low percent ranges are kinda fishy due to a low number of rolls. I might try to take some unfavorable attacks to get more rolls in the low percentage range in.




journeyman

journeyman
Joined: Oct 2020

If someone wants to check this, I do suspect that critical fails are broken. I've seen a bunch with advantage, leading me to think that it might be taking a 1 from either dice to determine crit fails, which is a plausible programming mistake. Otherwise the dice are working, and people are just so bad at math that they think they are missing too much when the chance to hit as been artificially inflated.




enthusiast

enthusiast
Joined: Jun 2020

I also advanced in the game some more and updated the google sheet. Had a couple of bad rolls for high percentage chances, which "tainted" the results (I am sure in the long run this will disappear again). But currently the 9095 range only succeeds with a rate of 79%... The other high percentage ranges seem fine though. The low percent ranges are kinda fishy due to a low number of rolls. I might try to take some unfavorable attacks to get more rolls in the low percentage range in.
Possibly worth checking. That said, with some added luck, even over 50 rolls, rather than ending at the "expected" 81%, we had a 90% to hit ratio with a 81% chance. If someone wants to check this, I do suspect that critical fails are broken. I've seen a bunch with advantage, leading me to think that it might be taking a 1 from either dice to determine crit fails, which is a plausible programming mistake. Otherwise the dice are working, and people are just so bad at math that they think they are missing too much when the chance to hit as been artificially inflated. I did one test with advantage, but couldn't see anything out of the ordinary.. Advantage tends to produce critical hits as opposed to misses (as you're always taking the higher of two rolls), disadvantage vice versa. Could do another advantage test again though, perhaps over a couple more rolls.




enthusiast

enthusiast
Joined: Jun 2020

Ok, here's 75 rolls on advantage (on a 94% hit chance each). There's quite a lot of question marks in there, which is because I took my Tiefling Wizard again and had her attack from behindn with her staff (D61 damage). That means every time she rolled a 2 or less on the D6, no damage. They should REALLY change this either in the animations or otherwise, because I'm sure most confuse these nodamage rolls for misses. They are not.
As expected advantage favors Critical Hits, so nothing unusual again. For there to be a critical miss you'd need to roll TWO 1s with advantage. No big data again, but coupled with my other tests on advantage, this appears to work correctly too.
? hit 20 critical hit ? hit ? hit 16 hit ? hit 5 miss 14 hit 14 hit 17 hit ? hit 5 miss ? hit ? hit 13 hit ? hit 12 hit 18 hit 8 hit 13 hit 8 hit 13 hit 15 hit ? hit 19 hit 19 hit ? hit 12 hit 20 critical hit 6 hit ? hit ? hit ? hit 10 hit 16 hit 17 hit 6 hit 15 hit 19 hit 12 hit 7 hit 2 miss ? hit 20 critical hit 4 miss 2 miss 12 hit 18 hit ? hit 8 hit 15 hit ? hit 19 hit 16 hit 8 hit ? hit ? hit 14 hit 14 hit 15 hit 20 critical hit 8 hit 18 hit 6 hit 15 hit ? hit 10 hit 16 hit ? hit 17 hits 20 critical hit ? hit 13 hit 10 hit ? hit
TOTAL ROLLS: 75 TOTAL HITS: 70 EXPECTED TO HIT RATIO: 94% TO HIT RATIO: 93,33%
Last edited by Sven_; 22/10/20 10:32 AM.




member

member
Joined: Oct 2020

I agree op there is something really not working at this time more than just RNG. And skills / spells / weapons don't seem to be doing their full effects of damage..




enthusiast

enthusiast
Joined: Jun 2020

skills / spells / weapons don't seem to be doing their full effects of damage.. Ah, another thing to verify. Who's in?




enthusiast

enthusiast
Joined: Jun 2020

So, is there anything obviously wrong with the damage roll? DanteYoda brought it up.
Picked a bang average Warlock and let him cast his Eldritch Blast. Eldritch Blast deals 1D10 (110) force damage, for an average of 5.50. Opposition doesn't have resistance against that. Here we roll.
4 10 3 8 3 10 10 10 3 2 10 10 5 2 4 4 1 8 4 7 4 3 7 6 7 1 4 4 1 9 7 10 4 9 1 4 1 10 7 9 8 5 7 8 6 8 10 4 3 6
TOTAL ROLLS: 50 TOTAL DAMAGE: 291 EXPECTED AVERAGE: 5.50 AVERAGE DAMAGE: 5.82
Last edited by Sven_; 22/10/20 07:08 PM.




enthusiast

enthusiast
Joined: Jun 2020

Is there something obvious wrong with the damage rolls? Part Deux.
This time I picked a rogue (+3 from dexterity), and let her attack with her short bow. That's 1D6+3 (49) for an average of 6.5. However, the opposition has resistance against piercing damage, which slashes it all in half. This time I stopped early, because it was obvious where this was going again.
3 2 2 4 4 4 2 2 2 4 4 3 2 4 3 2 3 4 3 4 3 3 4 2 2 2 5 (CRITICAL HIT) 4 3 3 TOTAL ROLLS: 30 TOTAL DAMAGE: 92 EXPECTED AVERAGE: 3.25 AVERAGE: 3.07




journeyman

journeyman
Joined: Oct 2020

I also advanced in the game some more and updated the google sheet. Had a couple of bad rolls for high percentage chances, which "tainted" the results (I am sure in the long run this will disappear again). But currently the 9095 range only succeeds with a rate of 79%... The other high percentage ranges seem fine though. The low percent ranges are kinda fishy due to a low number of rolls. I might try to take some unfavorable attacks to get more rolls in the low percentage range in.
Possibly worth checking. That said, with some added luck, even over 50 rolls, rather than ending at the "expected" 81%, we had a 90% to hit ratio with a 81% chance. If someone wants to check this, I do suspect that critical fails are broken. I've seen a bunch with advantage, leading me to think that it might be taking a 1 from either dice to determine crit fails, which is a plausible programming mistake. Otherwise the dice are working, and people are just so bad at math that they think they are missing too much when the chance to hit as been artificially inflated. I did one test with advantage, but couldn't see anything out of the ordinary.. Advantage tends to produce critical hits as opposed to misses (as you're always taking the higher of two rolls), disadvantage vice versa. Could do another advantage test again though, perhaps over a couple more rolls. Thanks! I guess I was just that one in 16000, I've rolled around 50 advantage rolls and gotten 4 crit fails, which made me suspect something was broken. I could see it being a real mistake in terms of the engine.




apprentice

apprentice
Joined: Oct 2009

I got a critical miss on 98% chance to hit, but that still does mean that theres a 2% chance to miss and that's what happened.
RNG is RNG. Unless it says 100%, don't expect not to fail. Rolling a 1 is ALWAYS a miss (hence it's called critical miss), even if your total hit modifier is 9 and their AC is only 7, therefore 100% doesn't exist, 1/20 = 5%. And therefore on a normal roll the maximum is 95% assuming your "base hit" > "their AC". This also means that critical hit (20) is also always at least 5% chance no matter what, unless you got a disadvantage roll, then it's 2.5% (2 dice, taking the worst result). 98% I assume has to be a "your base hit" > "their AC" + "advantage roll", because then you will roll 2 dice and take the best result, which is technically 97,5%, but I suppose they rounded it off and having a miss with that means you rolled 2 x 1, really bad luck. Lol
Last edited by Genky; 22/10/20 09:47 PM.




enthusiast

enthusiast
Joined: Jun 2020

But currently the 9095 range only succeeds with a rate of 79%...
Not sure if you've updated in the meantime, but do you think they may be worth investigating again for bigger sample sizes? I've also seen another claim: That the percentages on dualwielding would be bugged. I've yet to dualwield anything tbh on this.
Last edited by Sven_; 23/10/20 07:05 AM.




enthusiast

enthusiast
Joined: Jun 2020

I was outside. On the second level if temple of Selune.
With my party + 1 arch druid we killed everyone in the courtyard. My party wiped when 2 drunk bugbea s missed/dodged every attack in four rounds.
Do you mean these guys? I attacked those two for a while now with my party (main fighter two handed axe, Gale cantrips, Lae'zel her sword, Shadowheart mace and didn't see anything out of the ordinary, all blessed but Lae'zel). I mean I'm sure I would have gotten unlucky too at some point, but... Here are all rolls against them over multiple reloads. 56% attack hit 88% critical hit 65% hit 60% hit 65% hit 60% miss 75% hit 88% hit 60% hit 65% hit 60% hit 56% miss 65% hit 84% critical hit 88% hit 50% miss 65% hit 60% miss 75% miss 65% hit 65% miss 60% miss 75% critical hit 65% miss 65% hit 65% hit 65% hit 65% hit 60% hit 65% hit 84% hit 65% miss 60% hit 65% miss 55% hit Total rolls:35 Total hits: 25 Average hit chance: 66.54% Tohit ratio: 71.42%
Last edited by Sven_; 23/10/20 09:21 AM.




stranger

stranger
Joined: Oct 2020

I got a critical miss on 98% chance to hit, but that still does mean that theres a 2% chance to miss and that's what happened.
RNG is RNG. Unless it says 100%, don't expect not to fail. Well I missed with 100% hit chance as a rogue against a TURRENT soo.. you are just plain out wrong. And the guy above who said that he missed 14 times in a row with 50+ hit chance might have been overexaggerating but not by a lot I had plenty of cases where I simply reload the game because out of 14 hits (And yes I did count them) with 70 hit chance being the fucking lowest i hit twice sooo you know although it is possible to happen it hardly happend once. On top of it all at one point i got so fucking fed up with this BULLSHIT i decided to save and load HIT FOR HIT 51% hit chance it took me 11 loads to hit the target so either I am being unlucky all day every day OR the hit and miss system is BS. Doesn't matter now because I stopped playign the game for the time being until it is fixed stil the problem is there and is real.




stranger

stranger
Joined: Oct 2020

I got a critical miss on 98% chance to hit, but that still does mean that theres a 2% chance to miss and that's what happened.
RNG is RNG. Unless it says 100%, don't expect not to fail. Rolling a 1 is ALWAYS a miss (hence it's called critical miss), even if your total hit modifier is 9 and their AC is only 7, therefore 100% doesn't exist, 1/20 = 5%. And therefore on a normal roll the maximum is 95% assuming your "base hit" > "their AC". This also means that critical hit (20) is also always at least 5% chance no matter what, unless you got a disadvantage roll, then it's 2.5% (2 dice, taking the worst result). 98% I assume has to be a "your base hit" > "their AC" + "advantage roll", because then you will roll 2 dice and take the best result, which is technically 97,5%, but I suppose they rounded it off and having a miss with that means you rolled 2 x 1, really bad luck. Lol How do you miss with a 100% hit chance? Not CRITICAL miss just a miss? My rogue did so obviously the numbers we are shown are not to be trusted. And that 98% hit chance and a miss is nice and all and I could buy it IF it was really bad luck because honestly the amount of misses i get with 98 hit chance is un fucking real! This is a 1 in 50 hits that is supposed to be a miss well it is not ... not even by a long shot.




apprentice

apprentice
Joined: Oct 2009

How do you miss with a 100% hit chance? Not CRITICAL miss just a miss? My rogue did so obviously the numbers we are shown are not to be trusted. And that 98% hit chance and a miss is nice and all and I could buy it IF it was really bad luck because honestly the amount of misses i get with 98 hit chance is un fucking real! This is a 1 in 50 hits that is supposed to be a miss well it is not ... not even by a long shot. What is your rogue's base attack + bonus attack for the action you're using and what is the AC of the target? unless your total attack is 12 and their AC 10, you never have a 100% hit (not taking critical miss into account), because even if you have perfect dexterity and you're level 4, your total attack is never more than 6 or 7 (maybe 8 or 9 fully buffed with attack bonuses) unless you used scripts or developer commands to alter them and there aren't a lot of creatures in the game with 10 AC. Also, as I said, 100% hit chance doesn't exist because you can't remove the 1 from your 20 dice and even if your total attack is 12 with buffs and everything and their AC is only 10, rolling a 1 makes that 13 but it still misses, that's why it's called a critical miss ... a critical miss basically just means you rolled a 1 on your attack dice.
Last edited by Genky; 23/10/20 02:40 PM.




enthusiast

enthusiast
Joined: Jun 2020

Unsurprisingly from my own experience so far, I see others who actually sit down to take notes find much the same as me (and TyPinOwly). My guess is that people are confused by the rolls that deal no damage, still. They LOOK like misses, but aren't. https://www.reddit.com/r/BaldursGate3/comments/j8xode/rng_testing_how_r_is_the_ng/There's also a guy on Steam who actually recorded all individual dice over 110 throws on a 65% chance (needed an 8 for a hit). 1 : 4 times 2 : 5 times 3 : 6 times 4 : 3 times 5 : 7 times 6 : 7 times 7 : 7 times 8 : 8 times 9 : 5 times 10 : 8 times 11 : 7 times 12 : 5 times 13 : 3 times 14 : 6 times 15 : 6 times 16 : 2 times 17 : 3 times 18 : 7 times 19 : 4 times 20 : 7 times
To sum it up:
71 hits (~64,5% chance to hit) 7 crits (~6,4% chance to crit) 29 misses (~35,5% chance to miss) [ Also, as I said, 100% hit chance doesn't exist because you can't remove the 1 from your 20 dice . In the UI, it does. Attack bonus of 5 vs an AC of 7 (needs to roll a 2) on ADVANTAGE. That said, according to this graph, this exists  even if you rolled two 1s?? http://media.zerohitpoints.com/images/higherthan.pngGoing to test it now.
Last edited by Sven_; 23/10/20 03:43 PM.




addict

addict
Joined: Sep 2020

I got a critical miss on 98% chance to hit, but that still does mean that theres a 2% chance to miss and that's what happened.
RNG is RNG. Unless it says 100%, don't expect not to fail. Rolling a 1 is ALWAYS a miss (hence it's called critical miss), even if your total hit modifier is 9 and their AC is only 7, therefore 100% doesn't exist, 1/20 = 5%. And therefore on a normal roll the maximum is 95% assuming your "base hit" > "their AC". This also means that critical hit (20) is also always at least 5% chance no matter what, unless you got a disadvantage roll, then it's 2.5% (2 dice, taking the worst result). 98% I assume has to be a "your base hit" > "their AC" + "advantage roll", because then you will roll 2 dice and take the best result, which is technically 97,5%, but I suppose they rounded it off and having a miss with that means you rolled 2 x 1, really bad luck. Lol You can get 100% vs prone or unsuspecting enemies.
Last edited by DumbleDorf; 23/10/20 03:50 PM.




enthusiast

enthusiast
Joined: Jun 2020

You can get 100% vs prone or unsuspecting enemies.
How does it work though? Technically you could still roll 2 1s. Anyway, here's 50 100% throws. As usual, advantage favors critical hits over critical misses. Even over these comparably small samples, you can always see where this is headed. I think I have recorded well over 5600 rolls now in this thread. TyPinOwly contributed some atop of that.If there are bugs right with the RNG, they sure aren't obvious any. 18 hit 15 hit 18 hit 13 hit 19 hit 15 hit 11 hit 14 hit 20 critical hit 14 hit 15 hit 8 hit 14 hit 17 hit 19 hit 15 hit 20 critical hit 20 critical hit 17 hit 15 hit 18 hit 19 hit 17 hit 18 hit 19 hit 12 hit 5 hit 11 hit 19 hit 5 hit 6 hit 16 hit 18 hit 16 hit 16 hit 11 hit 15 hit 14 hit 12 hit 13 hit 16 hit 20 critical hit 16 hit 14 hit 14 hit 20 critical hit 17 hit 18 hit 17 hit 20 critical hit TOTAL ROLLS: 50 TOTAL HITS: 50 TOHIT RATIO 100% (as expected)
Last edited by Sven_; 23/10/20 04:25 PM.




enthusiast

enthusiast
Joined: Jun 2020

There's one segment in TyPinOwly's hit charts at the moment that look suspicious, the 9095% range. So I've increased my sample size for this one and went with a 90% chance. Made a Gyth fighter with an overall attack bonus of +4, and let her melee attack the AC 7 opposition. That's a 90% chance precisely, as she needed to roll at least a 3. This is no cherry picking, I was fortunate enough to be able to explain some of the human psychology at work again. The expectation would be 1 miss in every 10 hits, after all. 19 hit 13 hit 16 hit 18 hit 13 hit 13 hit 1 critical miss* 13 hit 13 hit 15 hit 12 hit 15 hit 17 hit 10 hit 18 hit 18 hit 20 critical hit 17 hit 18 hit 9 hit 8 hit 10 hit 18 hit 1 critical miss* 5 hit 18 hit 9 hit 4 hit 18 hit 9 hit 6 hit 16 hit 3 hit 2 miss* 19 hit 5 hit 7 hit 14 hit 2 miss* 15 hit 5 hit 16 hit 14 hit 11 hit 3 hit 5 hit  1 critical miss* 1 critical miss* 2 miss* 9 hit 6 hit 2 miss* 1 critical miss* WHAT PEOPLE WILL REMEMBER 2 miss* 12 hit 13 hit 13 hit 1 critical miss* 12 hit 1 critical miss*  16 hit 9 hit 18 hit 15 hit 16 hit 9 hit 11 hit 11 hit 11 hit 10 hit 9 hit 8 hit 12 hit 13 hit 15 hit 19 hit 3 hit 15 hit 40 HITS IN A ROW, WHAT PEOPLE WON'T REMEMBER 17 hit 6 hit 6 hit 11 hit 9 hit 7 hit 18 hit 14 hit 14 hit 14 hit 13 hit 14 hit 11 hit 17 hit 16 hit 10 hit 13 hit 10 hit 5 hit 8 hit 18 hit 7 hit  OVERALL ROLLS: 100 AVERAGE ROLL: 10.85 (EXPECTED 10.5) TOTAL HITS: 88 TOTAL HITS RATIO: 88% (EXPECTED 90)
Last edited by Sven_; 23/10/20 06:58 PM.




