Larian Studios Forums Baldur's Gate III Baldur's Gate III - General Advantage, yet still rolling 1s

How, and why? The chance of rolling 1s with two 20-sided die is literally .25%, and yet, it happens regularly in the game.

Just theorising here but Niara did some analysis and found the RNG to be busted. If you roll low, there’s a higher chance your next roll will be low too. For that reason I’d say there’s a higher chance of rolling snake eyes than there should be.

it's beyond busted, if you are still consistently getting 1s on advantage rolls considering the low probability. Not to mention that it's listing 1s being rolled on attacks with advantage as critical misses, which is not the case unless you are really rolling two 1s on the dice rolls.

If it's any consolation, I've had my halfling character roll a natural one on an ability check she had with advantage, and I've had it happen at least once on at attack roll in combat as well. Four 1s in a row is pretty darn poor luck, but it's slightly more likely with Larian's RNR than it should be. The inverse is true as well - high rolls have a higher chance of being followed by other high rolls as well.

I will say... I've been at a table where a player using psychical (tested as fair) dice has literally been playing a halfling and rolled double natural 1s on at attack. I've seen it happen once, exactly, in my table-top history, but I have seen it... It can happen. But you will see it happen more here in this game than you will outside of it, and that is a problem that their "Ha-ha, You're a cheater" dice check box doesn't actually fix.

I wonder... maybe the "1 is always a miss" code is overriding the "take the highest of two rolls" portion of the code. If so then you would expect MORE critical misses when you have advantage rather than fewer. Also, clearly a bug if this is the case.

It wouldn't surprise me if both were true given that they aren't even applying their own DOS-rules as they have them stated.

I will add that even with weighted dice, there was an instance of rolling 4 2's in a row. The longest streak on record I had with weighted dice.

I have the sneaking suspicion that Larian is treating 'advantage/disadvantage' as a +/- to the dice roll, rather than two RNG computations where the higher or lower is selected.

I did make a histogram of advantage and disadvantage Over here.

The good news is that the shape for Advantage/Disadvantage does not match a -/+ 5. to the attack rolls.

However when tracking Larian's true RNG I had quite a lot of streaks. To the extent that I had 3 numbers streaking over the same time period. I'm in the belief that weighted dice is added logic on Larian's base for RNG. So streaks were reduced BUT there is still something off with the RNG.


I'm theorizing this may be true as well. However with weighted dice, I did not see 1's on advantage more than expected.

If you have advantage, the chance of rolling a 1 should be reduced from 20% to 0.25% (5% of 5%).
Thats quite low but it can happen.

There was the same discussion in the owlcat forums. One player complained that the game is cheating because he misses all the time. Nope, he failed because he had the wrong feats and stats to get a good hit chance.
We also explained that anything that can happen will happen eventually, no matter how small the chance is, if you roll infinite times and playing a big RPG is a good approximation of infinite rolls.
I think somebody tested it in BG3 by doing some things 200 times in a row and he confirmed the % numbers.

by the way: What exactly does loaded dice do?

I have about... 2000 rolls recorded, about 1000 which is just near straight attack rolls after each other from running around the Ruined Village and bashing everything bashable only interrupted by a single saving throw or two from running over elemental ground effects.

The other 1000 was collected during regular play and is a mix of attack rolls and saving throws and skill rolls, but as I recorded that I felt more and more like I was missing out on so many rolls (there's still so many rolls that doesn't show up in the log and/or go past so quickly I have no chance of jot them down) that it wouldn't make for a very good comparison. But when you just run around bashing furniture for half an hour you can record every single roll right after the other.

Here is a sample of what weighted dice do in the game.

For the most part the weighted dice reduce streaks for dice rolls. Anything else about it is non-conclusive unless we get more samples from Larian or other players.

I may be wrong as I'm not familiar with 5e mechanics, but I also suspect it's the "roll a 1 is always a miss" issue at play. I notice that aboard the nautiloid, my chance to hit the imps with Eldritch Blast displays 95%, and after I cast Hex: Dexterity on them it stays at 95%. My understanding is that the Hex should force them to take Disadvantage on their Ranged Save, and that that is a Dexterity Save, so the 95% should become 99%. Is that what is *supposed* to happen or am I misinterpreting the rules somewhere?

To my extent of (limited) knowledge the chance to hit shouldn't change. Eldritch blast is a ranged attack, so it is defended by the imp's armor class, not ability checks.

Also, Hex gives disadvantage on ability checks, not saves. So even if you cast something that required a Dex Save (instead of a ranged attack as @ash elemental mentioned), hex would have no effect on your chances to hit.

Great, thanks for the info.

Thanks for the explanation, didn't know that either.

Considering the amount of attack rolls that this game must make a 1 in 400 chance will happen, and happen often; you were just on the wrong end of the RNG goods favorites list

The rolling system seems whacked sometimes. I was using Protection From Evil/ Good, and Imps hit me 4 times in a row. They should of had disadvantage, but it didn't seem like it.

It's not just 1/400 attack rolls, it's 1/400 attack rolls with advantage. That's when the game should be computing two 'd20 rolls' and taking the higher, and yet getting critical misses with advantage is still much more common than should be possible if their RNG system was using anything similar to actual probability as a basis.