So I did more work on the Falco chaingrab:
At 54% (47 prior to throw) you have to start the dash grabbing to cover DI away. Prior to this, you do not need to move to cover all DI.
You have two DI options to cover. I found that on frame 7 after Falco has been hit, you are given a noticeable visual cue. Basically, his body is in line with your back for DI behind, and his feet stick out past your head for no DI and DI away. Prior to this, all DI looks very similar. Obviously you can notice differences, but unless you are really perceptive you might not see them in real time. I think I need to do some research on what goes into a visual cue to determine some criteria for this before I start identifying visual cue frames for other percents.
Moving on, I found that at 40%, Falco hits the ground on frame 28, meaning you have to grab him by 27. DI behind is the hardest option to cover, as it requires at least an additional frame. Therefore, you need to react to Falco's DI by frame 19 at the latest. That's a 12 frame window starting from the cue.
1000ms / 60 frames = 16.666... ms/frame
12 frames * 16.666 ms/frame = 200 ms
Considering the world record is about 100 ms and the average is 225 ms, this is well within human capabilities. However, 200 is an elite number - only about a quarter of the population has this. But it's definitely attainable for most people if they practice improving their reaction time.
Anyway, dash turn around grab starts at 83% (76%). So from 54% to 83%, you have three DI options to cover: DI in, slight DI and no DI away, and full DI away. This is the next DI range and what I'm going to study next. I think the visual cue will become apparent more quickly, but there are also more possibilities of slight DI to confuse you, so it might be necessary to wait a bit longer to confirm the DI. There's also the possibility of using option selects via dashing certain lengths to cover no DI away, slight DI away, and There's a chance that it might require faster reaction times at this range.
The last range is 83% - x%, where x is the percent turn around down-b works (haven't tested it yet). At this range, you're gonna wanna down-b for no DI and DI away to set up for the kill, ftilt DI behind only if they are close to the ledge (the kb is pretty weak compared to down-b), or regrab. It's pretty much that you might be able to kill early if their DI is bad. Once you hit x%, dthrow should kill, or at least set up for an easy edgeguard. Again, this one also could be harder in some ways.
I'm going to find all the reaction times required at all the percents possible prior to 54% using this method. I'll also be checking to see if the visual cue frame changes. Then I'll do that for the other ranges, starting with the 54 to 83 one. For this one and the last one, I'll be looking for option selects to cover no DI and DI that requires a slight dash. Not to mention, I'll be keeping track of the changing visual cue frames. Once I've done everything for Falco, I'll do it for Fox.
This approach will allow us to set reaction time goals and optimize how we chaingrab spacies. Any ideas, thoughts, criticisms on this approach?
EDIT: I was totally wrong on the reaction time part. I did some research on reaction time and learned about choice reaction time (CRT) , the reaction time for responding to stimuli when at least two options are present. The more options, the slower the CRT is. The average is about 350 ms for four options. I couldn't find the world record anywhere, but I read that it would be "hard to go under 250 ms." That puts a huge damp in the chaingrab in my opinion. Still going to do the tests, but I'm less hopeful than I was at first.