This game has seriously messed up RNG

[MN_14]

Crafters have tested the percentages before and yes, reclaim is actually 90% but you have to have a large enough sample size to confirm that. Everything really does average out so the % you see is accurate. Yes, you can have strings of misses, but I've also seen consecutive crafts where every last HT landed.

The bottom line is that while RNG is present, everyone is subject to the same effects. We all play by the same rules. Some do better than others and that simply boils down to a crafter's own handling of the mechanics.

Does it feel unfair when you 100% a whole bunch of crafts and then suddenly NQ the only two crafts that fell below to 91%? Yes, but it's part of the system. Most of the time, you'll get the HQ and whoever can consistently get 91% will HQ a lot more crafts than the person who only can manage 70%. So your ability as a crafter makes a difference.

The designers have given the player a lot of tools use. Even when crafting 4* from all NQ, there aren't too many situations where I'm at major risk of an NQ. Due to the balance and overpowered gear/abilities in HW, building 11 stacks of IQ is practically a given. I don't fall to 10 IQ too often and rarely fall into the single digits (maybe 1/30+). From all NQ, the only time I really get concerned is when I hit the 6+ miss mark. Missing 4 times does not pose any real problem and 5 misses can go either way (but it's still mostly HQ). The more robust your strategy is, the better you'll be able to mitigate misses.

Start with a moderate amount of quality and you should be fine almost 100% of the time.

[XiXiQ]

yes, you will have "strings of misses" the minute you start. I gave up on reclaim before HW, and even the small amount of crafting I do now still gives me "strings of misses" on every third craft. Seems my "string" started 3 years ago and hasn't finished yet. Last week - RS failed 5 times in a row under steady hand, and I barely craft at all anymore. (that's not what I'd call rare either, I probably see it weekly). If all these "if you do it a zillion times" claims are true, I should practically never see failure at all at my modest level - yet it keeps happening, the same pattern, over and over and over.

The RNG is either fine - in which case, there is no need for mitigation (because the numbers will bear it out). If you need to mitigate, the RNG simply isn't throwing out the expected numbers.

[MN_14]

I gave up on reclaim before HW, and even the small amount of crafting I do now still gives me "strings of misses" on every third craft. Seems my "string" started 3 years ago and hasn't finished yet. Last week - RS failed 5 times in a row under steady hand, and I barely craft at all anymore. (that's not what I'd call rare either, I probably see it weekly). If all these "if you do it a zillion times" claims are true, I should practically never see failure at all at my modest level - yet it keeps happening, the same pattern, over and over and over.

The RNG is either fine - in which case, there is no need for mitigation (because the numbers will bear it out). If you need to mitigate, the RNG simply isn't throwing out the expected numbers.

That's not how RNG works. Even with 11 HTs at 80% success, it can be calculated that that you will have 4 or more misses 16% of the time. Although, it will average out to 2-3 misses over a large sample size, the spread is known to be mostly 0-5 misses. The more RS or HT you use, the larger your spread. That's why you have to be able to mitigate misses (in terms of averages, the times you have few to no misses will offset your large numbers of misses and bring you to that average 2-3) and use other skills to manipulate the probabilities. It's part of the design and the strategy that the designers want you to employ.

If you're using a smaller sample size, you're much more likely to see discrepancies form the average than if you were to use a large sample size (ie. doing it a zillion times).

4-6 misses in a craft is normal and yes, I see it in every crafting session I go through. It is expected and you're supposed to be able to respond to it.

[Niwashi]

For the theory, it is correct to bet on red if it lands on black. You know as well as anyone that it will not always land on black, otherwise the perceived percentage would be almost 90-100%. People are not stupid. While it is not reliable to land on red, it is definitely not false.

After black, landing on red next is equally likely to landing on black again. Looking at both rolls together, there are four possibilities: black-red, black-black, red-red, red-black. Out of those, there is exactly one that has black followed by red, and one that has black followed by another black. 1 to 1 even odds. (Though perhaps coin flips would be a better example, since this ignores the green space that gives a slight (but still equal) disadvantage to both, enabling casinos to make money off of it.)

(And with longer sequences, the same thing happens. If you recorded 99 results and charted all the possibilities of 100 or 200 spins, you'd find exactly the same number of possibilities where the sequence you've observed is followed by a red as there are where that sequence is followed by a black.)

People only think red is more likely next because they're equating black-red as being the same thing as red-black. But to a roulette wheel or a computer, those are completely different results, as different as black-black is to red-red.

[XiXiQ]

That's not how RNG works. Even with 11 HTs at 80% success, it can be calculated that that you will have 4 or more misses 16% of the time.

If you're using a smaller sample size, you're much more likely to see discrepancies form the average than if you were to use a large sample size (ie. doing it a zillion times).

Then why do I see that pattern far more than 16%? I see it more like 30-40% of the time. It's tiresome, sitting there eye-rolling every time I'm crafting 10-20 items for something. and my gear is sufficient it doesn't really affect the outcome anymore. It's simply not working as it should, and it never has for the last 3 years I've been doing it. Just because you seem to be OK with the RNG, don't simply discount stories that bring it into question.

[Elamys]

I think we all are more inclined to believe actual numbers as opposed to biased anecdotes about how it doesn't work. If it helps at all, I'll make a list of Hasty Touches taken vs. succeeded and we can go from there, but that will take me a bit of time.

[XiXiQ]

You claim your RNG is working, I believe you and don't need logs. Many, many people have expressed their experience is the opposite - using a ps4 I cannot get any logs. (best I could do was take notes just prior to HW, which was what made me decide it was a mug's game) Not knowing how the seed works, there might be a difference between processors/OS - yours might always be perfectly as expected, someone else's might always come out worse. Sadly, there's no way to get data out of a PS4 and compare to a PC. Any test you do won't necessarilly equate to my particular experience, or anyone else's.

[Frizze]

Not knowing how the seed works, there might be a difference between processors/OS

The determination is made on the server itself, so which system you play on makes no difference.

[MN_14]

I've tested a small sample of HT under SH2 on a 80 DUR recipe using mm2 once and then reclaiming and following with 5 HT for a total of 14 HT per synth. 224 HT attempts in total.

Synth 1 - 4/14 misses
Synth 2 - 2/14 misses
Synth 3 - 1/14 misses
Synth 4 - 3/14 misses
Synth 5 - 1/14 misses
Synth 6 - 4/14 misses
Synth 7 - 0/14 misses
Synth 8 - 2/14 misses
Synth 9 - 2/14 misses
Synth 10 - 2/14 misses
Synth 11 - 6/14 misses
Synth 12 - 5/14 misses
Synth 13 - 2/14 misses
Synth 14 - 5/14 misses
Synth 15 - 3/14 misses
Synth 16 - 2/14 misses

I'll do some more tests later but these stats (smallish sample size) indicate an average miss rate of 19.6% and 31% of the crafts had 4 or more misses. With 14 touches, the expected % of synths with 4 or misses is around 30% so the results are in line with expectations. Only the craft that has 6 misses is of real concern when it comes to NQing a 4* from all NQ mats as the others will typically give you mostly 100% with the others usually falling to 75-95.

[MN_14]

I’ve tracked the results of over 1100 HT over the last while and got the following:

Set 1 from before (14 HT per set):

0 misses – 1 occurrence
1 miss – 2 occurrences
2 misses – 6 occurrences
3 misses – 2 occurrences
4 misses – 2 occurrences
5 misses – 2 occurrences
6 misses – 1 occurrence

%miss = 19.6

Set 2

0 misses – 0
1 miss – 2
2 misses – 5
3 misses – 3
4 misses – 3
5 misses – 1
6 misses – 2

%miss = 22.32%

Set 3

0 misses – 3
1 miss – 2
2 misses – 3
3 misses – 3
4 misses – 2
5 misses – 1
6 misses – 2

%miss = 18.75%

Set 4

0 misses – 1
1 miss – 4
2 misses – 2
3 misses – 4
4 misses – 1
5 misses – 3
6 misses – 1

%miss = 20.09%

0 misses – 1
1 miss – 3
2 misses – 7
3 misses - 4
4 misses - 1

%miss = 14.73%

This is the only slightly odd run but if you notice, the cause of the deviations is an increase in the number of synths with 2 misses, which is simply a slightly above average result. This is only a small subset so everything will even out over a larger set as there will undoubtedly be sets where the results are slightly below average. Believe it or not, a small batch of synths with 4 misses (say 7/16) is only slightly below average and craft balance has been designed for you to be able to cope with them. You might not 100% as many crafts as you usually do, but you should not be at any major risk of an NQ.

Overall miss rate (1120 HTs) = 19.11%

From this set of samples, I don’t see any evidence of horribly broken RNG. Everything is within expectations and if there are deviations, its due to small normal random streaks. You do get small clumps of 5 or 6 misses, but those don’t happen often enough to highly influence the averages. You also get small clumps of 0-1 misses at around the same rate but unlike the large number of misses, you don’t think as much about them.

I do have to say that examining the expected results have been interesting. Statistically, 70% of the time, your results will be either average or above average, nearly 20% of the time, they will be slightly below average, and a bit over 10% of the time, your results will be worse than that.

I find this fascinating as it falls completely in line with what I’ve observed. ~70% of the time, I can craft a 4* item from 0-100%, ~20% of the time I get 84-94%, and ~10% of the time, I get 84% or worse. Everything is completely predictable, falling in line with calculated probabilities. The implication here is that if they were to smooth out RNG so that you get what seems to be an average result all the time (in terms of both misses and procs), you’d have no trouble crafting from 0-100% all the time since that’s how crafts have been tuned: an average result is enough for a 0-100% run.