What is up with Conquest RNG?

[Seti]

Been playing Conquest Hard/Classic today until chapter 18.... I start the chapter, position my units and end the turn. As i expected the general with the hammer walks up to my great knight corn cob and attacks her with a 53% chance to hit. He hits and kills her. Ok fair enough bad luck so i reset and start the map again. I position my units the same way and let the turn end. Once again corny gets nailed by the 53% hammer. Bad luck again so i reset and start the chapter again and position my units the same way and corn cob gets nailed by the 53% hammer again. So i reset....and reset...and reset (way too many resets later). 

Alright you get the problem. Ive been nailed by a 53% hammer 20 times now and this is smelling fishy. Just to clarify this is NOT a battle save im reloading from. I am reloading from the prep screen and keep getting this BS. Even tried going back to castle a couple of times to "reset" the RNG but still not working. Is there some hidden value where effective weapons get some stupid hawkeye on the first turn??

EDIT: holy shit the hammer finally missed!!!!!! I guess this topic is a bit moot now however i still think there is something up with how fates calculates rng. For the 2 chapters i have had xander he has nailed more 10-15% criticals than i can count. 

Edited February 26, 2017 by Mr.Pikachu

[SpearOfLies]

Change strategy.

[DodgeDusk]

I haven't looked into how the RNG works in Fates, but I do remember an instance in Birthright when my Hana would constantly get hit by a 23% chance by a certain enemy (it would outright kill her) in Chapter... 13, if I'm recalling correctly. I thought Lady Luck wasn't on my side initially--and maybe that was the case--but after a number of resets too large to remember, I started to believe maybe the RNG string for battles is fixed; resetting the game doesn't reset the RNG like in some previous FE games. I wouldn't mind knowing how the RNG works in Fates.

[DragonFlames]

I also don't know how the RNG works, but from my experience, it does what it can to make you reset or lose a unit. For example, if there is one enemy that can kill one of your units and you have no way to move that unit out of the way or block the enemy off, that enemy will dodge anything that comes its way, no matter what it is. I've had this problem several times now. I don't remember the exact circumstances or units involved, since it was so many times, but I recall and example from Conquest Chapter 10.

Selena was low on HP and I had already used Elise's and Felicia's turns to heal up other units elsewhere on the map, who would otherwise have died, namely Silas and Niles. I had her kill an enemy Ninja that would otherwise have offed Odin, but this put her in a position where another unit - a Spear Fighter I believe - could come in and kill her (before you ask, Effie's and Arthurs' hit rates were too low and Odin didn't do enough damage to kill the Ninja reliably). I had Effie, Arthur and Odin nearby and still able to move, but all of them conveniently missed their 80% to 95% chances to hit JUST so the Spear Fighter could then go and kill Selena. And no, I couldn't have positioned them in a way that would have prevented the Spear Fighter from reaching and killing Selena, because I still had to be wary of both Hinata and another Spear Fighter.

Or this one where an enemy Berserker had a 1% chance to crit and needed that crit to kill Ryoma and guess what happened? Yes, he got the crit. Off a 5% chance to hit, no less. And Ryoma missed his initial attack at a 96% hit rate.

Other examples just like this have occurred so often that by now, I don't believe in the 'random' in RNG. The game is just straight up screwing me over at this point.

Edited February 26, 2017 by DragonFlames

[joshcja]

Rng is semi-seeded and runs off one Rn.

It's not practical to exploit this but if you get utter shit luck 10+ times in a row reload the chapter.

Edited February 26, 2017 by joshcja

[JJ48]

The RNG definitely seems far more brutal in Fates than in previous games in the series.

I was just playing Royal Royale and had gotten down to ONE opponent left.  Two of my units couldn't really hurt him, but one had a 93% chance of one-shotting him, while he only had a 17% chance of one-shotting my unit if I missed.  Of course, my unit missed and then got killed, leaving me to just stare blankly at the two useless units left and think about how I'd now have to play through that entire scenario all over again.

[Synjid]

8 minutes ago, JJ48 said:

The RNG definitely seems far more brutal in Fates than in previous games in the series.

Well if I understood correctly what @joshcja wrote, hit rolls are run with only one random number in fates, instead of two like in all the others FE games since binding blade. This indeed results in a more cruel rng, and I'm actually curious as to why this decision was made. Does anyone have information about that?

[Interdimensional Observer]

Practically every game in the series has been accused of wonky RNG at some point. But thinking the RNG is consciously looking out to get you just when you need everything to go right doesn't make sense if you think logically. How in the world would you program a game to look at the situation and see your strategy? Computers can't think, they can only crunch numbers/data and react according to their results. So unless there as a way of calculating how good the player must be feeling/progressing along on a given map, and then modifying just the proper RNs. But that'd be difficult to tell if I had two units doing independent actions on different sides of a map- Left Unit needs the to dodge at 19% or it'll die, but Right Unit is in no such situation. How could the game tell which I plan to move first? Add more units and the "proper RNs" become impossible to predict. I guess it could "tag" a separate set of screwy RNs to the proper character(s), but what if I decide "this is too risky, let me pull back"?

If it were possible for the game to predict exactly when I would feel the most frustration at being screwed over by the RNG, then that kind of AI programming would be better served elsewhere in terms of applications of technology if it wasn't already there.

[JJ48]

55 minutes ago, Interdimensional Observer said:

Practically every game in the series has been accused of wonky RNG at some point. But thinking the RNG is consciously looking out to get you just when you need everything to go right doesn't make sense if you think logically. How in the world would you program a game to look at the situation and see your strategy? Computers can't think, they can only crunch numbers/data and react according to their results. So unless there as a way of calculating how good the player must be feeling/progressing along on a given map, and then modifying just the proper RNs. But that'd be difficult to tell if I had two units doing independent actions on different sides of a map- Left Unit needs the to dodge at 19% or it'll die, but Right Unit is in no such situation. How could the game tell which I plan to move first? Add more units and the "proper RNs" become impossible to predict. I guess it could "tag" a separate set of screwy RNs to the proper character(s), but what if I decide "this is too risky, let me pull back"?

If it were possible for the game to predict exactly when I would feel the most frustration at being screwed over by the RNG, then that kind of AI programming would be better served elsewhere in terms of applications of technology if it wasn't already there.

This is the Internet!  Your facts and reason have no place here!

 

Seriously, though, I think it's less "the RNG is fifty-seven moves ahead of us!" and more "something is different about the RNG, and it's not as reliable as it used to be".  In previous titles, if I used a unit 10 times, and each time it had an accuracy of 90, I'd expect that it would miss approximately one time (0 if I'm lucky, two or even three if I'm particularly unlucky).  In Fates, it sometimes seems as if it's not even worth attacking with anything less than 100, as an accuracy of 90 seems to indicate that the unit will miss 7-8 times out of 10.  At least, that seems to be my experience (for my units, at least.  For enemies, an accuracy of 5 seems to be a guaranteed crit.)

[Interdimensional Observer]

Confirmation bias of one's mind and memories is a thing. You generally don't remember every 5% crit you do or 90% hit being your units dodge, because you never counted on them happening and they're just a nice bonus. Whereas you expect to dodge a 5% hit/crit as part of your strategy, and thus one actually hitting throws a wrench in things and turns out to be more memorable. Furthermore, you generally don't talk of your own low chance hits hitting, because players generally don't bother even attacking when hit rates are too low. Whereas you rely on 90% hits hitting as part of your strategy, and for one to miss is frustrating and thus more memorable.

It is possible that the RNG produces "valleys" of low RN values and "peaks" of high RN values. A string of low RNs means things will generally hit/crit, and a string of high RNs means things will generally miss. I heard the RNG in at least one Dragon Quest game features such highs and lows (which can be frustrating with you're trying to Zing somebody back to life, but you constantly miss because the RNG is going through a high). That would not be too difficult to check for, all we'd have to do is crack open the RNG, look at the numbers being produced, and see if we can find such trends.

[JJ48]

53 minutes ago, Interdimensional Observer said:

Confirmation bias of one's mind and memories is a thing. You generally don't remember every 5% crit you do or 90% hit being your units dodge, because you never counted on them happening and they're just a nice bonus. Whereas you expect to dodge a 5% hit/crit as part of your strategy, and thus one actually hitting throws a wrench in things and turns out to be more memorable. Furthermore, you generally don't talk of your own low chance hits hitting, because players generally don't bother even attacking when hit rates are too low. Whereas you rely on 90% hits hitting as part of your strategy, and for one to miss is frustrating and thus more memorable.

It is possible that the RNG produces "valleys" of low RN values and "peaks" of high RN values. A string of low RNs means things will generally hit/crit, and a string of high RNs means things will generally miss. I heard the RNG in at least one Dragon Quest game features such highs and lows (which can be frustrating with you're trying to Zing somebody back to life, but you constantly miss because the RNG is going through a high). That would not be too difficult to check for, all we'd have to do is crack open the RNG, look at the numbers being produced, and see if we can find such trends.

Yes, we understand basic statistics concepts.  Thank you.  The topic being discussed isn't, "something statistically unlikely happened," but rather, "there seems to be an issue with the RNG in Fates which was not present in previous FE titles, which we have played and experienced".

 

8 hours ago, Synjid said:

Well if I understood correctly what @joshcja wrote, hit rolls are run with only one random number in fates, instead of two like in all the others FE games since binding blade. This indeed results in a more cruel rng, and I'm actually curious as to why this decision was made. Does anyone have information about that?

This does sound interesting.  What do you mean by the other FE games using two rolls, and why does it affect the game so?  If it's too long to post here, can you link to an article that explains this?

[DoesntKnowHowToPlay]

FE6 through Awakening all use a 2 RN system for hitrates, sometimes called true hit. Basically instead of rolling one number and checking it against hit rates, the game would roll 2 numbers and average them. This means that the values it's comparing to the hit rate trend closer to 50, so displayed hit rates are not actually accurate unless they are 0%, 50%, or 100%. Hit rates above 50% are more likely to hit than displayed, and hit rates below 50% are less likely to hit than displayed. This usually favors the player, as they won't be making attempts to hit unlikely blows often, but will make a lot of attacks with high displayed hitrates. In short: unless you've played one of the (super) famicom games, ever Fire Emblem you've played has lied to you. You are trained on bullshit statistics, which is why Fates seems weird sometimes.

Fates changes its system somewhat. For attacks with 50% hitrate and below, the game uses a 1 RN system, meaning the chance to hit is actually what is displayed; the game isn't lying to you. For attacks with more than 50% hitrate, the game uses an unknown different system to nudge the odds upward. Source here: https://fire-emblem-strategy.tumblr.com/post/143452625727/how-fates-handles-hit-rates

Note that all of this only goes for hitrates, not crits or skill procs or other things. Another noteworthy detail is that Conquest is quite a bit harder than most other Fire Emblem games, so it often only takes a little bit of bad luck to derail your plans, making it more visible when things go awry.

e: read the OP again and if you have a 53% chance to die maybe figure out another way to go about the first turn?

Edited March 7, 2017 by DoesntKnowHowToPlay