Problem I see is the move is fairly telegraphed, and has a large amount of hitstop, as well as a large amount of recovery it looked like. I can try and figure out exactly how fast it is (current build) to give a rough idea for the final version (probably different I'm sure but its still nice to have a bookmark). Hopefully its fast enough to catch a bad roll and such though.
EDIT:
Really only one great example to analyze.
Raw Data
4644-Begin
4650-Dash starts
4656-12 frames in move hits
4664-Hitstop phase 1 ends
4667-Hitstop phase 2 ends
4686-Weapon disappears, bow reappears, likely point of move end
Frame Count
0-Move Activation
6-When the dash begins and likely when the "hitbox" starts
12-The moves dash goes on for at least 6 frames, unknown the final amount
20-From analyzing the frames, there were two basic frames Pit was stuck in from the hitstop from the move. The first lasted from 12 frames in to 20 frames in
23-Second part of a frozen follow through, lasting only 3 frames, making the total hitstop value 11 frames
42-42 frames later, Pit is almost back in a neutral stance and has his Bow reequipped. Note that this is about the maximum period it'd be out, minimum could be before the fist is unequipped.
This is of course on hit. Subtract the hitstop value and you have 33 (those 2 frozen frames counting but not the repeats) frames. Obviously it's unknown how it reacts to Shields and how much shield stun it creates, as well as how long the move is actually out before it either auto swings or falls flat. I'd say it can be a great punish tool if you make a good read, but it does look to have quite a big recovery window.