Does Awakening use Single RNG or Double RNG for hit?

[shadowofchaos]

I got an interesting comment thread on one of my Youtube videos with the Chapter 3-4 Lunatic+ "guide" I made from Interceptor's thread.

I did assume that FE13 had a double RNG simply considering that 1% hits happened to me a lot less often than 1% crits.

But I did see their point after reading a bit and viewing the video...

Video in question:

https://www.youtube.com/watch?v=mGihfURFRsA

Quote:

Well, sure, but way to miss most of the point there. That point being that if the RNG worked like you're claiming it does in this game, 90%s would miss and 10%s would hit a LOT less. It's been a while since I watched this video, but I'm pretty sure 90%+s miss 2 or 3 times in this map alone (and this isn't an outlier, either)-- and with with 2 RNG that's actually close to 1% probability. Meaning you'd only need to play through the game like once to see it happen 50 times at that rate, forget 2 years.

Should also be noted that 1% crit is actually much more common to see than 1% hit because of how the stats work (crit rates are generally going to be low, whereas hit rates are generally going to be high), so even if you're going to say you've seen more 1% crits than 1% hits in FE, that's because you've got more samples to work with. There's also a bias in that those crits are just that more memorable (they usually spell disaster for someone whereas the fluke hit/miss is often not relevant). If you were to compare with 99% chances to hit that missed to 1% crits, I bet your totals would be much closer.

Seriously if you're skeptical just march a swordmaster into a bunch of troops without vantage a few times and see how often you don't get hit ever in this game and then do it in say FE7 and compare results. No offense, but this honestly seems like an argument a theorycrafter would have rather than someone who actually plays the games a lot, cause it's kinda really obvious this game doesn't use the same hit formula as others in the series if you try to evade tank much ever.

So what about your guys' experiences?

Which RNG do you think it uses?

Or am I crazy in throwing it out there that maybe that it switches to single RNG on Lunatic?

And there's really no way to confirm this other than decrypting the ROM and looking at the assembly portion.

[Czar_Yoshi]

Well, I know I've never seen a 99% miss (under a single RNG system, that's just as likely as a 1% hit), and those are much more common hitrates. I've seen quite a few 1% criticals, though (including one stacked with Lethality... preceded by a 9% Lethality crit). The highest I've ever seen was a 97% miss, and I've only gotten one of those.

If anyone wants to try to get a data sample, it should be done using 75 listed Hit- according to the site's True Hit chart, that's 12.75% off of the actual rate, which is the farthest away True Hit ever gets (so you'll have the highest chance of noticing a difference).

It's also worth remembering that even a tiny increase in percentage when the odds are that slim makes a major difference. In a map of 25 enemies, if all your hitrates were 90 listed and it took two hits to KO all foes, you'd be seeing an average of one miss per map (98.1%). But if you were following Int's old guide, I've got a feeling you'd be seeing a lot more hitrate variation to account for misses.

In fact, I've got a chunk of free time right now, so I'm going to watch that through carefully and count up all the hits/misses and their hitrates, and we'll see how unusual they look. Obviously I don't care about 100% rates.

Turn 1:

Kellam hits with 87 listed -> 96.75% true

Knight hits with 76 listed -> 88.72% true

Archer hits with 69 listed -> 81.09% true

Archer hits with 98 listed -> 99.94% true

Robin hits with 86 listed -> 96.22% true

Kellam misses with 77 listed -> 89.65% true

Turn 2:

Miriel misses with 90 listed -> 98.10% true

Kellam hits with 87 listed -> 96.75% true

Sumia hits with 98 listed -> 99.94% true

Chrom hits with 78 listed -> 90.54% true

Sumia hits with 98 listed -> 99.94% true

Virion hits with 92 listed -> 98.80% true

Soldier hits with 76 listed -> 88.72% true

Vaike hits with 82 listed -> 93.70% true

Turn 3: You've started loading bookmarks now so you're going to have some instances of deliberately avoiding bad RNG spots. Numbers from here on are skewed in you favor.

Archer hits with 74 listed -> 86.74% true

Fred hits with 83 listed -> 94.39% true

Fred hits with 93 listed -> 99.09% true

Chrom hits with 82 listed -> 93.70% true

Sumia hits with 79 listed -> 91.39% true

Turn 4:

Virion hits with 95 listed -> 99.55% true

Miriel hits with 77 listed -> 89.65% true

Kellam hits with 98 listed -> 99.94% true

Bottom side down. I could do more, but I think it's pretty obvious at this point- there were two misses, at 77 and 90 (89.65% and 98.10% respectively) out of 22 attacks, which is pretty normal. But more numbers just because.

With True Hit, you've got an average Hit rate of 94.2414% over those specific attacks. With single RN Hit, you've got an average Hit rate of 85.2273%.

With True Hit, that means you'd expect to see a miss roughly every 17.3 attacks- which is extremely close to what you got. With Single RN, you'd expect to see a miss every 6.8 attacks. That's a little farther off.

With True Hit, you'd have a 23.3927% chance of getting exactly two misses over the average hitrate. With Single RN, you'd have a 20.6113% chance of getting exactly two, which is also a little worse.

So while those differences are fairly small and could just be the product of chance, they are on the side of True Hit- and the size of the differences is less important than the sample size. Of course, if there's a very prolonged string of bad luck later on that I'm missing, that could affect things.

Edited January 29, 2015 by Czar_Yoshi

[Ownagepuffs]

Considering how many sub 35% hit axes Chrom liked to introduce to his face in my early game I wouldn't even be surprised if it went to the 1 RN system for Lunatic, lol.

[dondon151]

i've been running FE10 recently and i've had numerous runs fail due to missing 88+ displayed hit. these things happen.

once tauroneo missed double 94s which is ARGH

[Irysa]

Although my thread from a while ago was ultimately inconclusive in determining specifics, I can debunk the 1 RN per hit theory because it makes calculated RN burn simply not add up.

Refer to the thread if you want specifics but basically I was manipulating and counting the number of RNs I was theoretically using up and was able to accurately predict and produce the exact same level ups when I counted the RNs correctly according to the laid out presets. Assuming 1 RN doesn't work beacuse the third example given becomes 9, wheras the first two are 8, yet the results are the same. The same would be able to be replicated in detail if I still had the bookmark to show for other numbers but you'll have to trust me on this.

I ran into a wall because the amount of instances of combat appeared to advance the RNG or at least burn an indeterminately large amount of RNs. I suspect the RNG may advance based off another RN since the gap appears to be so large, but I have no way of proving this.

Edited January 29, 2015 by Irysa