|
|
Post by October Raven on Jul 18, 2020 11:52:38 GMT
So, I've got this odd situation where edits who are designed to be jobbers end up dominating grapples. Oddly this phenomenon seems to have gotten worse over time, to the point where they almost never lose grapples.
Fiddling with the usual suspect AI parameters seems to have no effect, either. It's weird.
My current theories on this:
1) AI going for small grapples tend to win more than med/large (unlikely because this is also happening in small damage, but I can't discount the possibility) 2) Certain types of moves win grapples more (The joshi forearm seems to be a particular culprit here) 3) Something is making the game weigh grapples towards jobbers (charisma? overall? parameters? points total?)
I'll be testing these theories gradually here with some test edits, to try to isolate the issue and hopefully build a better jobber.
|
|
|
|
Post by OrochiGeese on Jul 18, 2020 12:20:54 GMT
Yeah, I continually see this too. I know that it's supposed to be based on random chance (the whole RNG/roulette wheel concept) but some edits just seem to win grapples more than others do.
I had to stop simming my giant-sized edit Wang Gom-Bu because it just got too frustrating to watch him completely dominate opponents. It wasn't even a question of not being able to be lifted or ukemi/reversals. It was simply that he kept winning 75% of grapples consistently from match to match.
I think there's some hidden variable that allows some edits to have a higher percentage chance to win grapples.
I also think it's weird when you get a match where one edit wins every grapple for a five minute span.
|
|
|
|
Post by October Raven on Jul 18, 2020 17:10:41 GMT
I also think it's weird when you get a match where one edit wins every grapple for a five minute span. It is, but I'm not even talking "a match". This is a consistent thing. Different jobbers, different opponents. I thought I had it fixed but when I streamed a match with The Beer Guy it went 27 minutes and he almost won every single grapple! |
|
|
|
Post by OrochiGeese on Jul 19, 2020 0:15:09 GMT
Yeah, like there's something in the edit that's causing this to consistently happen against all opponents.
It could be a glitch for all we know. I used to joke that an edit wearing red boots gives them a +10 RNG advantage and maybe something like that is actually happening.
FPW is still riddled with glitches and unintentional consequences. Maybe an edit with a certain Charisma and Ranking set can steamroll over others. The RNG is supposed to be random but that doesn't mean there isn't some hidden, unintentional factor that is skewing it.
|
|
|
|
Post by October Raven on Jul 19, 2020 0:25:40 GMT
So, first, the control: Two edits that are identical (except for their morphsuit color). All 1s in attributes, same moveset, same AI (does only small, then only med, then only big).
As expected, the match is even. Red got a few more reversals in, but nothing that seemed out of the norm.
|
|
|
|
Post by October Raven on Jul 19, 2020 0:47:39 GMT
First experiment time: This time Red has three small grapple chops he uses in small and med, and 100% leg trip in Big. I forgot to put percentages into a prio for a pin but the experiment isn't about which one wins, it's about if Red dominates in grapples or not. And the results are... interesting.
Note that, as the match got further into large damage, Red started winning more and more grapples. This doesn't entirely match with what I've been observing with my edits, but it also doesn't disprove the concept. So the grapple progression theory may be part of the cause, but I still doubt it to be the cause.
|
|
|
|
Post by October Raven on Jul 19, 2020 1:16:22 GMT
Last experiment for now: Blue has all strikes, Red has all throws and slams.
So Blue ended up winning before getting into large damage, but that's not what was being tested. What is being tested is winning grapples, and Red again wins a lot more grapples near the end. This is suggesting that grapple wins get weighed not by any parameters or AI or skills, but rather by amount of damage taken. Still, it's not the near 100% domination I've been seeing in sims outside of these experiments and also debunks the factor that was being tested (that is, if strikes win more than throws and slams)
|
|
|
|
Post by LankyLefty17 on Jul 19, 2020 2:59:43 GMT
Given that grapples are won based on RNG- I would run more than three matches before making any conclusions personally.
|
|
|
|
Post by LankyLefty17 on Jul 19, 2020 3:36:46 GMT
Just one more thing, because I feel like this is going to potentially spin people up. The move that's chosen, based on logic, is chosen after a grapple has been won. So it is impossible to assume that grapple wins are dictated by the type of move or the damage slot the move is in. Grapples are won the exact same way as Returns- through the RNG system. |
|
|
|
Post by ligerbear on Jul 19, 2020 16:13:55 GMT
am i remembering wrong. i thought at the match onset certain rng values were assigned which would dictate predominately how often a participant would win the grapples. as a result yeah since its random potentially someone can dominate a match or by chance just keep winning grapples for a huge stretch
|
|
|
|
Post by October Raven on Jul 19, 2020 23:21:12 GMT
Per Lanky's suggestion, I'll be doing the experiments in three match bursts to ensure consistency. Next I'm moving to my third test. Equal AI and moveset, but A (Blue) has all 1s in offense while B (Red) has all 10s.
So while this is inconclusive as, despite the larger data set, it's not as defined as my pre-experiment observations. However, in all three matches there's a pattern: After B hits his first Rolling Elbow, A wins every grapple until he too lands at least one Rolling Elbow. This suggests (but not confirms) that there is some parity at play here. The consistency in that sequence however, is worth exploring, so I ramped it up a bit: giving B all 10s in defense as well as offense.
Okay, now we're getting somewhere.
There's consistently two rallies; one that happens to bring parity to med damage, and a MASSIVE, dominating comeback at Near Death. And it's the latter comeback that is replicating what I've been seeing when I've added self-damaging moves.
I'm going to do some further testing, but I won't be capping the next round of testing because, if I'm right, I could possibly get these test dummies to go 60 minutes three times in a row.
|
|
|
|
Post by October Raven on Jul 20, 2020 0:43:21 GMT
Experiment 4- The one I can't in good conscience subject y'all to videos of.
So, A and B are the same as in 3B, but with one key difference: A has one of my self-damage moves for small grapple.
When A hits one.... A keeps hitting one. Constantly. For the entire match. Nothing else happens. There is no end. There are only forearms.
So, yeah. I've replicated my issue. And now, I'm pretty sure, I've got down why it's happening.
NEXT: The solution.
|
|
|
|
Post by OrochiGeese on Jul 20, 2020 5:19:35 GMT
Experiment 4- The one I can't in good conscience subject y'all to videos of. So, A and B are the same as in 3B, but with one key difference: A has one of my self-damage moves for small grapple. When A hits one.... A keeps hitting one. Constantly. For the entire match. Nothing else happens. There is no end. There are only forearms. So wait... could custom moves be somehow messing with the engine/RNG? Or is it possible that self-damaging moves somehow mess with things? Chorus of a song for FP:M volume 2? 😁 |
|
wepner
Steel Johnson
Posts: 144
|
Post by wepner on Jul 20, 2020 5:42:31 GMT
Experiment 4- The one I can't in good conscience subject y'all to videos of. So, A and B are the same as in 3B, but with one key difference: A has one of my self-damage moves for small grapple. When A hits one.... A keeps hitting one. Constantly. For the entire match. Nothing else happens. There is no end. There are only forearms. So wait... could custom moves be somehow messing with the engine/RNG? Or is it possible that self-damaging moves somehow mess with things? Chorus of a song for FP:M volume 2? 😁 The chanting of a cult that worships Popeye. |
|
|
|
Post by IamAres on Jul 20, 2020 6:48:21 GMT
It's entirely possible that there's a mechanic that provides a greater likelihood of winning under specific conditions, like being immensely more damaged than the opponent. I do feel like, around the time a guy gets his Ukemi boost, he seems to win the next several grapples at least, although that's fairly anecdotal on my part. A mechanic like that would explain some of these findings, though.
It shouldn't be able to overtly change the timing each edit can use - that's governed by one value that applies to the whole match; it's why you can't actually set guys to different difficulties, or set aspects like grapple timing or submission escaping to different levels, even with modding. But it might be able to enforce a slight delay on the opponent when grappling, which I think is how Carl's "jobber" feature (the part that makes them unlikely to win grapples, anyway) works now. Or there could be another undiscovered mechanic at play.
I don't know offhand how one would go about circumventing such a mechanic using in-game methods, though.
|
|
