Is Fate using only 1 RN?

[Ayra]

I've played a lot of Fire Emblem Fate so far and it really feels like the game is using only one RN instead of the two we've had since the GBA games regarding hit rates. "Unexpected" misses or hits seems to be far more common than what I'm used to while playing a Fire Emblem game.

I don't have formal data to present a formal proof and it might be only a personal bias, but... It's constantly happening all the time, both for the enemy and my advantage. I was playing Radiant Dawn right prior to playing Fate and a FE7 ROM hack before that, and I was not feeling this way.

Does anyone have any information regarding the RN in this game?

[XeKr]

Personally I don’t think it is. I feel like I’ve hit way too many 80-90ish percents for that to be the true hit.

Would be interesting to see something definitive though.

[Book of Ereshkigal]

I feel like I've been missing more 80s than normal, but I've been chalking it up to personal bias.

[thundershock125]

what's RN?

[Ryo]

^ Random number

[nordopolica]

I don't know about how many RNs they use but I have missed too much 93% and the enemy has hit way too many 30-40% for my liking.

[MCProductions]

It might very well be 2 RNs, I haven't played Fates for myself yet, but BB supposedly uses 2 RNs, and I miss 80s and 90s all the time. Until recently, I thought 2 RNs was an American only thing, especially after playing BB. This games RNG may just be similarly anal about hitrates.

[Minischew]

Naga is not on your side.

[sforzur]

I don't know about how many RNs they use but I have missed too much 93% and the enemy has hit way too many 30-40% for my liking.

Likewise here. Once I started approaching endgame, I just assumed any and all attacks would hit me. One of my units had a double attack with a 90% rate and missed both times; I was completely baffled. The best luck I've had was when a unit landed five crits in a row with a 27% rate immediately after their recruit.

[Airship Canon]

It might be confirmation bias leaning you to believe they've reverted from True Hit.
There was a lot of questioning it in Awakening too.

And FE6 has True Hit/2RN, and it likes throwing 90%? Miss. At you.

[nocturnal YL]

I asked the same question in Awakening. I do see an unusual amount of 90%+ misses, but that alone doesn't really say anything.

I'm not surprised if that's the case for both this game and Awakening. Unlike the previous hit rates, the newer games actually say "___%" in the battle screen.

[XeKr]

Generally we as people are bad about expected outcomes.

Saw we have 90% hit and send Kamui to get 5 enemies he doubles and one-rounds. I think most of us would (initially) think 90%, that's pretty good. But we shouldn't feel that comfortable that Kamui will kill them all, as the chance of 90% hitting 10 times in a row is around 35%. So it's around 65% likely that at least one will miss and enemies will survive, but we don't think of it like that. We think, "90% is really high, I can't believe he missed."

That's just 5 enemies. Imagine how many you fight in FE throughout the game.

Even at true hit, of 98.1%, it's around 82% of hitting all 10 hits, meaning there's still a reasonable chance of at least 1 miss (think about ~20% criticals).

this doesn't preclude some of kind of "reverse" confirmation bias on my account

[Vitezen]

Guy above me is correct, multiplicative probabilities for all instances together are totally different from the base probability for a given event occurring once.

Assume you've got a stream of enemies coming at Kamui, with a 90% accuracy for each of his attacks on them.

90% chance to hit, 1 hit = 90%

90% chance to hit, 2 hit = 81%

90% chance to hit, 3 hit = 72.9%

90% chance to hit, 4 hit = 65.61%

Even things that are likely to occur will fluctuate over time due to variance.

Edited July 16, 2015 by Vitezen

[Airship Canon]

As a note, it'd be pretty simple but tedious to gather enough data to estimate.

Find a My Castle Address- and get a handful of characters who pull sub-100 hit rates [preferably you want ~80%s-- 99 has a truly MASSIVE amount of hit rate between 1RN and 2RN, 1:100 vs 1:10,000 misses, but 99:100 is really damn accurate as-is, and would require around a good million samples to confirm].
Fight repeatedly making the same moves.

Record results until you've launched a few thousand attacks, check average hit rate against displayed. If you've got that large a sample size, the real pattern should appear evident. If it's really close to displayed, it's probably 1 RN, if it's significantly off (+/- ~12%) it's 2 RN.

I asked the same question in Awakening. I do see an unusual amount of 90%+ misses, but that alone doesn't really say anything.

I'm not surprised if that's the case for both this game and Awakening. Unlike the previous hit rates, the newer games actually say "___%" in the battle screen.

After thousands and thousands of attacks, you can generally get the feel of it-- Awakening had 2 RN.

[nocturnal YL]

After thousands and thousands of attacks, you can generally get the feel of it-- Awakening had 2 RN.

I said that just days after getting the game. Eventually, it felt like 2RN.

[sylvanllewelyn]

Stats nerd here with data, see below. Don't even play this game. Registered this forum just to post this.

It is possible that even 2RN underestimates hit rate.

Over a hundred data points, and the running average of the second column is still much higher than the running average of the third column. Someone please hack, deriving the actual function will take many thousands of data points because the feedback is just 0's and 1's.

htttps://www.youtube.com/watch?v=m6IG1EjQBT0

Displayed Hit / miss 2RN 72 0 84.6 86 1 96.22 90 1 98.1 85 1 95.65 75 1 87.75 68 1 79.84 77 1 89.65 68 1 79.84 77 1 89.65 81 0 92.97 81 1 92.97 84 1 95.04 56 1 61.72 55 1 59.95 83 1 94.39 63 1 72.99 63 1 72.99 93 1 99.09 86 1 96.22 69 1 81.09 69 1 81.09 83 1 94.39 67 1 78.55 71 1 83.47 55 1 59.95 55 0 59.95 57 1 63.45 89 0 97.69 85 1 95.65 89 1 97.69 99 1 99.99 67 1 78.55 87 1 96.75 74 1 86.74 62 1 71.5 77 1 89.65 93 1 99.09 84 1 95.04 65 1 75.85 75 1 87.75 69 1 81.09 72 1 84.6 72 1 84.6 73 1 85.69 90 1 98.1 85 1 95.65 90 1 98.1 91 1 98.47 50 1 50.5 50 1 50.5 92 1 98.8 65 1 75.85 73 0 85.69 55 0 59.95 73 1 85.69 60 0 68.4 63 1 72.99 63 1 72.99 83 1 94.39 43 0 37.41 43 1 37.41 23 1 10.81 75 1 87.75 75 1 87.75 96 1 99.72

htttps://www.youtube.com/watch?v=EYOBdPKD4Do

Displayed Hit / miss 2RN 72 0 84.6 86 1 96.22 90 1 98.1 85 1 95.65 75 1 87.75 68 1 79.84 77 1 89.65 68 1 79.84 77 1 89.65 81 0 92.97 81 1 92.97 84 1 95.04 56 1 61.72 55 1 59.95 83 1 94.39 63 1 72.99 63 1 72.99 93 1 99.09 86 1 96.22 69 1 81.09 69 1 81.09 83 1 94.39 67 1 78.55 71 1 83.47 55 1 59.95 55 0 59.95 57 1 63.45 89 0 97.69 85 1 95.65 89 1 97.69 99 1 99.99 67 1 78.55 87 1 96.75 74 1 86.74 62 1 71.5 77 1 89.65 93 1 99.09 84 1 95.04 65 1 75.85 75 1 87.75 69 1 81.09 72 1 84.6 72 1 84.6 73 1 85.69 90 1 98.1 85 1 95.65 90 1 98.1 91 1 98.47 50 1 50.5 50 1 50.5 92 1 98.8 65 1 75.85 73 0 85.69 55 0 59.95 73 1 85.69 60 0 68.4 63 1 72.99 63 1 72.99 83 1 94.39 43 0 37.41 43 1 37.41 23 1 10.81 75 1 87.75 75 1 87.75 96 1 99.72

Edited July 20, 2015 by sylvanllewelyn

[Book of Ereshkigal]

The best luck I've had was when a unit landed five crits in a row with a 27% rate immediately after their recruit.

To my knowledge, crits still use one Rn regardless of what hit is doing. (Can't actually confirm this for Fates yet, but that's how it's been historically)

[Bovinian]

To my knowledge, crits still use one Rn regardless of what hit is doing. (Can't actually confirm this for Fates yet, but that's how it's been historically)

Can confirm, multiple consecutive 4% crits and failed 96% crits are all too real

[Crimean Archivist]

It's going to take a lot of data points to "prove" 1 RN or 2 RNs, which I'm working on right now. I have 1000+ data points from all values of Displayed Hit, but primarily from 75 to 95, and I'm recording every single value as I go, ignoring Dual Guards. I estimate that it will take several thousand data points, probably in the range of 30,000, to determine for certain which it is using. Preliminary results are inconclusive; the variances are too high (100 points, even highly concentrated, is too few). However, if a few community members want to help me out by submitting your data points, we could figure this out in maybe a week (I've collected all my points across just two days).

I have an Excel on my computer doing linear vs. logistic fits on the data and charting as it goes, so it's all very streamlined. All one would have to do is send their data to me, preferably in this format:

Column A Column B

Displayed Hit (1-99) Battle Result (H/M)

If you do that then I can import it rather quickly and not worry about formatting and whatnot.

I'm new to SerenesForums but if you need to check my credibility, I'm Silas from the fire-emblem-strategy tumblr, and I do stuff like this all the time.