Mood Changes Data 1.02

Ukujiku

[08] Mercenary
I couldn't find any raw data about mood changes based on Tira's (After 1.02) moves so I decided to do it myself. Here is the data I recorded:

A+B (1) on Whiff (JS → GS)
No Transform: 70 (75%)
Headache: 6 (6%)
Transform: 17 (18%)
Total Attempts: 93

A+B (1) on Hit (JS → GS)
No Transform: 33 (32%)
Headache: 0
Transform: 70 (68%)
Total Attempts: 103

A+B (1) on Block (JS → GS)
No Transform: 54 (48%)
Transform: 58 (52%)
Headache: 0
Total Attempts: 112

A+B Hold (3 hits) on Whiff (JS→GS)
No Transform: 30 (33%)
Headache: 19 (21%)
Transform: 42 (46%)
Total Attempts: 91

G+K (JS → GS)
No Transform: 15 (15%)
Transform: 85 (85%)
Total Attempts: 100

4K on Hit (JS → GS)
No Transform: 0
Headache: 75 (82%)
Transform: 17 (18%)
Total Attempts: 92

4K on Block (JS → GS)
No Transform: 128 (92.8%)
Headache: 0
Transform: 10 (7.2%)
Total Attempts: 138

4K on Whiff (JS → GS)
No Transform: X
Headache: 0
Transform: 0
Total Attempts: X (Lots of attempts with no transform or headache)

4K on Hit (GS → JS)
No Transform: 142 (94%)
Headache: 0
Transform: 9 (6%)
Total Attempts: 151

4K on Block (GS → JS)
No Transform: 105 (89%)
Transform: 13 (11%)
Total Attempts: 118

4K on Whiff (GS → JS)
No Transform: 65 (57.5%)
Transform: 48 (42.5%)
Total Attempts: 113

Running Kick (low) on hit (JS → GS)
No Transform: 54 (61%)
Transform: 35 (39%)
Total Attempts: 89

Running Kick (low) on Block (JS → GS)
No Transform: 60 (70.6%)
Transform: 25 (29.4%)
Total Attempts: 85

66B on Hit (GS → JS)
No Transform: 61 (86%)
Transform: 10 (14%)
Total Attempts: 71

66B . A+B+K on Hit (GS → JS)
No Transform: 68 (55%)
Transform: 55 (45%)
Total Attempts: 123

666B on Hit (GS → JS)
No Transform: 15 (24%)
Transform: 48 (76%)
Total Attempts: 63

666B . A+B+K on Hit (GS → JS)
No Transform: 44 (48.4%)
Transform: 47 (51.6%)
Total Attempts: 91

666B . A+B+K on Block (GS → JS)
No Transform: 40 (49%)
Transform: 42 (51%)
Total Attempts: 82

77A/11A on Hit (GS → JS)
No Transform: 47 (64%)
Transform: 26 (36%)
Total Attempts: 73

WR B+K . K . A on Hit (JS → GS)
No Transform: 57 (78%)
Transform: 16 (22%)
Total Attempts: 73

↑ B+K . A on Hit (JS → GS)
No Transform: 52 (59%)
Transform: 36 (41%)
Total Attempts: 88

236K . A+B+K on Hit [All] (JS → GS)
No Transform: 8 (21%)
Transform: 30 (79%)
Total Attempts: 38

236K . A+B+K on Hit [All] (GS → JS)
No Transform: 10 (27%)
Transform: 27 (73%)
Total Attempts: 37
 

Age_of_Truth

[12] Conqueror
Great stuff, thanks for this.

You mean 4K rather than 4B.

Also A+B+C is written A+B+K using the in-game notation.

If you want I'll try to organise this info into smaller, easily digestible nuggets.
 

Ukujiku

[08] Mercenary
This is mostly for people wanting to get into Tira and not knowing the rough chances of mood changes. Of course it's all random chance and you play with the odds. I don't like playing games of chance without knowing the odds. For example, what should you do at the end of a round? I've seen many Tira players do the A+B Hold when the taunt is clearly better.

Do what you like with the data. I think if someone knows statistics you can do confidence intervals or at least get something more out of the data, but I'm too rusty to remember how to do it. Thanks for the corrections Age_of_Truth.
 

Age_of_Truth

[12] Conqueror
I think this stuff is really useful. I will do some tests of my own to back yours up. I also want to check a few other things, e.g. when I am in training spamming 4K in Gloomy, I almost always get a JS transition on the 3rd whiff, but that might just be coincidence.
 

Wandrian

I don't know.
Wouldn't just taunting at the end of the round pretty much be all anyone could use though? The odds of persona shifts on anything else doesn't really seem applicable due to the unreliable nature of non-CE and non-health reasons.
 

Nightfall

[08] Mercenary
A bunch of unhelpful comments combined with misunderstanding of probability and standard deviation.
Pay no attention. This is fantastic data and the kind of thing I've been waiting for since the day the game came out. Even with only 100-150 tries, depending on the number of states for each move (almost all of which have exactly 2), it's more than enough to get within +/- 5% most of the time (100 coin flips will, on average, be off by 4%). I've already found this useful. In the past I've been reluctant to use 4k out of Gloomy due to fear of changing back to Jolly, but if the chance is only 10% (which is the most likely "true" number given the data), I'm much more comfortable using it. I for one will be looking over this list very carefully.

This also provides a jumping off point if anyone wants to refine these numbers. The results of any further experiments can be added directly to this one.
 

sillycatsteven

[09] Warrior
awesome job! most have taken you quite some time.

just one small question, is this tested in training mode or actual vs mode. i always had a sneaking suspicion that the chance for mood swing is different depending on whether you're in training or vs.

I wonder if these probabilities change if Tira reaches yellow health. If i have the time ill test it out
^ i was wondered about that too. i think if they are going to change the probability it'll be after she does the auto switch at lower health (i think it's 40% health remaining right?).


EDIT:
you guys are probably going to hate me for suggesting this, because it's very difficult to test. But pre-1.02, i suspect the change rate is dropped to low levels right after CE (not sure if this is also true for mood swing by other methods too). i usually can use 2~3 GS666B consecutively without triggering a change back to JS. i normally do JS CE ~GS 666B 666B 666B all as tech traps or if they are just lying there... i know you can normally safely use 2 without switching back to JS, the third is quite risky, but still not as often as normal. i say pre-1.02 because i didn't play against humans quite as much post-1.02, and this really only works on people. CPU normally block the first or second one.
of course this could is just be my imagination (or selective memory) because people generally only remember the times they succeeded.
 

Ukujiku

[08] Mercenary
All my data is under training mode at 100% health. Please test at other circumstances if you think anything else could be a variable. (i.e. low health or in vs) I don't think training mode vs versus mode would be a variable but feel free to prove me wrong.
 

Xyerith

SPHERECUCK
Figured that this would be the place to post this. A+B and 236K BE guarantee persona change on Just guard, the jg flash hides the persona flash often times, but you guys can test this out for yourselves if you don't believe me (provided this thread isn't dead while Im posting) Any who this could prove useful in the games future as our community does not know what JG could hold in the future (Viola can JG aGIs as an example, who knows what else could be discovered). Also Slade put the JG frames in the wiki for any of those interested in looking at how negative they are.
 

Zane

Spirit of the Wolf
How can Tira start a match in Gloomy? Is that random?
Ideally she would change when she's down by a round or two. This can be seen in Elysium. If the opponent is 1 or two rounds up and Elysium's next round weapon is Tira's, then she gets gloomy state
 

Kakaroth79

[08] Mercenary
Unlucky, you mean :p ; the only time i noticed that was when i played Decopon.

Rewatching the fight, i see it happened only the very first match (the one right after the button check) in the FT3, so i wondered if there were any trick, but yeah, a low probability random chance makes the most sense.