Greninja beats Robin IMO. His jab is not some insurmountable wall. Greninja has the second best dashgrab in the game, with which he can punish almost anything badly spaced. Robin can't just press jab to stop an approach.
Greninja can camp Robin fairly well, is short enough to dash under Arcfire during startup, and can generally choose to disengage if he's worried about grab or jab. Both of those moves hurt, so it's not wise for Greninja to stay up close for long periods of time. Arcfire is actually the worst thing about the MU IMO, because without it it's a simple case of Greninja pelting a very immobile target with fast shurikens. Arcfire blocking them is very good for Robin. I recall a video by Dath and Raziek where both said they don't find Arcfire useful against Greninja, so I don't know how their experience differs.
When Greninja does want to go in, his Nair is -4 on shield, which is too good for Robin to punish with his f7 grab. Even if the Greninja messes up the timing a bit, he can spotdodge. Approaching Robin is only made difficult by Arcfire.
Robin is too immobile to deal with continued, smart pressure from Greninja IMO. He has some very powerful up close options, but Greninja can choose to rarely/never be in the situations where he can get hit by them (similar to what Emblem Lord said about Greninja vs Ryu; Greninja can choose to disengage). His recovery being linear Hydro Pump fodder doesn't help either.
I feel that both players are coming at this MU from a pre-patch angle, where I might have agreed it wasn't in Greninja's favour. The changes to shieldstun and Greninja's buffs have turned it around.
Dathx
Hey, nice of you to comment.
I think your chart is quite accurate in general.
Others can't be blamed for using theory/not being top players to discuss this MU when you yourself have said you don't have enough data on it. I know that you recognise this, though.
I don't think discussing (not complaining - why does this word get thrown around so much?) a +/- 5 change is dumb in the context of a MU chart as that can be the difference between an even and advantaged MU. For what it's worth I think it's an error of +/- 10 anyway.