I can vouch for what zUkUu said. The close version of HK is indeed 7f start-up (at least in PC 1.06)
I believe Lars frame data is already done by someone. I have a problem finding the Lars section so I cannot give you a link, but I remember it being pretty much accurate.
I’ll try to confirm the 4 characters currently done, but I don’t have a recording device (I use a different method) so I’ll do it only if you’re okay.
The more the merrier! 
You and your insane knockdown dash method x).
I think after I’ve done all the DLC characters, I’ll also put up a list of their walk speed, dash frames, CADC and vitallity - or is that somewhere around?
I’ll double check any error you find to make sure I can find the error as well, so we have really a 100000% correct data down the line. I’d appreciate in particular if you do of all Charge moves (as in CADC moves) and super \ cross arts, Zonk\Ex Zonk (lvl 1), Ibuki’s Super Art and Dhalsim’s Yoga Blast\Fire. ^^
Also Paul’s Cross Art really 10F?
I also use 1hit-jump method 
There’s no data related to walk speed, dash, CADC, or health afaik. It’ll have to be tested separately. I do know that Lei has 17f forward dash, 1000 health, 4+34f jump diagonal and vertical (1f faster than Ryu), with the fastest CADC being 5+17f.
Sim’s Yoga Blast would be a difficult one, but I’ll try to do those after I’m done confirming what you already did. It’s gonna take a while since reading week doesn’t start til next week.
.
.
.
Paul’s cross art is indeed 10f
So far, from what I found in Cody:
**I have a problem with Cody’s close st.LK. If I use 1hit-jump method, it is +5 on hit and +1 on block, but if I try a normal -> normal link frame perfect, it acts like it’s +6 on hit and +2 on block. (eg. frame perfect cl.st.LK -> far st.HK on hit catches jumping opponents while they’re grounded, suggesting that it hit them during their 4f pre-jump frames. If it had 5f gap like it should’ve, they would get hit by far st.HK while airborne. In addition, this sequence trades with 4f moves without invincibility after they get hit by cl.st.LK). Same goes with Knife cr.HP; it is +6 for sure but it links into Knife cr.MP, cr.MK and sweep.
EDIT: WHOA wth. I wanna delete duplicate posts but I don’t see an option anywhere.
Thanks for the confirmation for Paul’s CA.
-Far s.hk is indeed 10F! I even reproduce why it was amiss. When Paul lands he ducks, and since s.hk hits rather high, Paul was ducking under it and raised into it. So I hit meaty with it. X
-far s.lp is +2. Converting error from my local spreadsheet. (aka typo).
-Knife s.lp is -1. 
-f.mp is 0.
-cl.lk is +5 on hit.
-s.lk is -1 on block. X
-c.hk is -7. X
I also checked a few other normals and they were all correct. Cody was the first character I’ve done and wasn’t quite comfy at that point. Also, I’ll stop using auto-jump for testing since it can seem to have an impact on testing higher hitting moves. Pause => Jump it is! takes more time, but isn’t worth the risk for errors that it seems to bring with it to use auto jump.
So thanks for pointing the errors out.
Full body Strike invulnerability, or just upper?
Knife c.lp has been fixed. Dunno why that was there. I remember thinking it was a low when setting up the raw data. After testing it and seeing that it was a high I thought I had changed it
Charge move are a hell to check since they always seem to be 1-2F slower then they actually are due to the whole charging crap. I found a new method for that tho. I record Paul doing Raze, block that and input with reversal timing. That way I have no 2F long button press or anything, at least I hope it doesn’t use a 2F long button press to charge it later when the move actually comes out. Would seem weird tho, so I don’t think that is the case. I also use that method now for CA \ Super Arts since he is airborne, you can see him moving as soon as the cinematic freeze ends.
That said, I just tried that for Ruffians and still come up with the same results, but this time MK Ruffian is also 1F slower. WTF IS GOING ON WITH THE GAME? Why does it add frames where ever it wants? When using Ex Ruffian for instance, there is a frame where cody stands, but doesn’t glow yellow. I can only fathom that there is 1f of animation where he stands up before the start up of the move starts. When I compare the pose Paul is in during the cinematic freeze, and that 1f where cody doesn’t glow, it’s exactly the same. So maybe for reversal timing, you have that additional frame of animation?
I mean that makes sense for Super Arts, since the cinematic freeze would otherwise consume 1F of start up, but why is it there for special moves? AND WHY ISN’T IT CONSISTENT FOR ALL SPECIALS?
I swear, this game…
Want to point out: Cody’s Zonk invincibility varies based on charge time just like in SF4. Level 3 has strike and projectile upper body invincibility until 2F before active while level 1 has significantly less.
TYVM for double checking. It’s strange tho, out of everything, f+MP being +1 on block was the one I was the most confident with.
I was wondering if you could test 2 of my testing methods with your recording device. I use:
If you could test f+MP with the above 2 methods that would be great. If you still get 0 on f+MP block then it means I would have to come up with a new method of measuring frame advantage.
Like I said before (I think in the Paul section), there seems to be 2 reversal timings, and the 2nd one is actually 1f slower than the 1st one. I swear I can punish a certain move with a reversal and sometimes I cannot punish the same move with the same reversal within the same range.
To test the special start-up I do:
Non-EX Zonk is upper body invincibility only.
Also, it seems like you can “cancel” the last recovery frame of normals such as Knife st.HP and Knife cr.HP with another move other than jump/dash. This is the only possibility I can think of that would explain Knife cr.HP being +6 on hit and being possible to link into cr.MK/sweep/Knife cr.MP. I know some of Lei’s stance normals do this, so it’s nothing quite new.
TY for pointing that out Eternal. I don’t know Cody much and didn’t know there were different Zonk levels. For the one I tested I charged for about 80 frames so I think that is level 1?
EDIT: Cody’s close st.LP, far st.LP, close st.MP, cr.MP, close st.HP, cr.HP, close st.LK, far st.LK, cr.LK, cr.HK, Knife st.LP, Knife cr.LP all have juggle potential of 2. Cody’s far st.MP, close st.MK, far st.MK, cr.MK, Knife st.MP, Knife cr.MP, b+MP, f+MP all have juggle potential of 3. Cody’s far st.HP, close st.HK, far st.HK, Knife st.HP, Knife cr.HP all have juggle potential of 4. Not sure about f+HK, but it has at least 4. The moves mentioned above all add 1 juggle count per hit.
Hwoarang:
Crouch HK: Cancel occurs on frame 14?
Spinning Trip Kick (misspelled): Block adv +1?
MK > MK: Cancelable into BC/LA
High Kick Combo: Cancelable into LA
Low & High Kick Combo: Cancelable into LA
Low & High Smash Kick Combo: Cancelable into BC/LA
Quick Kick Combo: Cancelable into LA
Flamingo Kick Combo: Cancelable into LA
Fire Storm: Cancelable into BC (heavy only)/LA
Special Step: Cancelable into SA
Jin:
Power Stance: not cancelable into Power Stance
Law:
Dragon Rush Combo: Cancelable into SM/CA, EX/SA (occurs on abnormal frame)
In general, it seems like recovery frames haven’t been updated in cases where frames were changed in moves. For example, Julia’s st.MP should have 10 recovery frames, and cr.MP should have 11.
Yeah I’ll do the recovery for moves that were mentioned in that way after everything is done. Not a high priority since the recovery isn’t very important to begin with, and is almost off for all chars by 1 frame anyway.
That you can’t cancel it into itself should be clear. Neither can you cancel sway into sway ^^
Added SM,EX,SA,CA to Dragon Rush.
Fixed Typo for move name. Will test Trip Block adv later and see when the cancel occurs for c.hk.
What is “BC”?
@Street11
I do it via jumping as well, but I rather see when the first jump frame occurs instead of when they land. I’ll do that later as well to give you feedback.
-I’ll also add a note about Zonk charge levels. I dunno if it’s worth to have 2 additional entries for it tho. Is the block advantage different?
-Blanka f.mp will be adjusted since it has a chargeable version that acts differently.
-Lastly, Lars is on the agenda.
BC is boost combo. LA is cross rush.
LA = Launcher
Okay I added all the Hwo cancels, and also Stance cancel note (same for Sway).
EX Sky Rocket has 5f start up (-1)
c.hk can be canceled after the 25th frame (26th is first start up of the special\super). You don’t seem to can cancel it during its active frame. Looks odd as hell too.
b+LK is 0 on block.
Added Blanka Rock Crusher Charge Note. The move is exactly the same frame wise, it just hits M for a 25F start up.
Added Zonk note. will probably test Zonk later tonight and see how the charge levels are different frame wise.
I tested Ruffians again and was able to confirm the start ups. (turbo function ftw).
@Street11
Very bad news. F.mp being 0 on block is true and wrong at the same time. It IS 0 on block, but ONLY against Paul (+others?). Against Cody, Zangief and Cammy, it’s +1 tho! (only tested it against those). I don’t know what causes this, if Paul is special, or if he is the only one, or if the 3 I tested are special in some way. If Paul is special, that means I can basically check EVERY DLC frame data AGAIN, since I always used Paul as dummy. I spot checked a few other normals, and they do seem right, so it might be a unique case with f.mp… or maybe not. I dunno. Please reaffirm this, I’m doubting my sanity… GOD I HATE THIS GAME >_>
Are you counting the hitstop frames (12 frames I believe)? Since it makes a blockstring with LK Hunting Hawk, I figured the cancel must be around 7 frames after the hit connects (frame 14).
I counted from button press to cancel. So subtract from that the start up etc. He does pull his leg back almost all the way before the cancel occurs. so i think you cancel part of the recovery.
Never tried with Paul (my dummy is Cody/Ryu/Lei), so I tried.
I got mindfucked even more. Against STANDING Paul, it indeed is 0 on block like you said. Well, on crouching Paul, it turns out that it is +1 on block. I was about to call GG, but after looking at the crouch blocking Paul, I think standing Paul is indeed the special case. I have zero idea why.
To make you even more confused: The thing is, the jump height difference between Paul and Cody after stand blocking f+MP (even) is actually different from Paul and Cody jumping at the same time from neutral (even). The jump height difference after crouch blocking f+MP(+1), is the same as Paul and Cody jumping at the same time from neutral (even).
At least for Cody frame data, the ones that I didn’t point out as errors are what I confirmed with Ryu/Lei/Cody (didn’t do specials, but for normals). I think it should mostly be fine.
THANK GOD IM NOT INSANE!
Was this in 1.06 also the case? That is the second bug already with paul. what did they do?
We also need to know if f.mp is the only move with that property AND if paul is the only character where that occurs.
Does that means I really have to redo all the frame data? ._. I really don’t want to… If you check the frames anyway, I think I’ll just redo the errors you find. Will take a look at knife s.lp again too. Gonna switch my dummy to cody for the future, since he has very iconic 2nd and 3rd jump frames, or ryu? I dunno :< /crisis?
Not sure about 1.06 tbh.
f+MP is not the only one. Cody’s Knife cr.HP is +7 on hit against Paul, while it is +6 on hit against Lei and Cody according to 1hit-jump method. However, Knife cr.HP -> Sweep combos on all three, suggesting it is +7 on hit, period. But it’s not, and sweep indeed has 7f start-up.
I’ll check the frames with more standard characters (Ryu, Ken and the testing character itself). We can talk about Paul after we’re done.
And don’t forget cr.LP!
EDIT:
LP Bad Stone is +5 on block (LP Bad Stone -> cr.MP is a true blockstring on Cody and Ryu)
MP Bad Stone is +3 on block
HP Bad Stone is +1 on block
Note: MP Bad Stone is +6 on hit yet can combo into sweep
HP Bad Stone is +4 on hit yet can combo into cr.MP
I am starting to think that the 1hit-jump method might not be the way to measure frame advantage in this game. Something is wrong here. It seems okay when measuring normals, but when it comes to unique attacks and specials, it is off by 1f in some occasions.