Frame Data Discrepancies, Reading Correctly?

[Kobayashi]

So I was dicking around in training mode to try various moves to see if they would beat FB Hyappo Shinshou (Jam's bullshit). The frame data says that the move is invulnerable to strikes from frames 3 through 13, with startup 14 and 3 active. This would imply that on frame 14, Jam is vulnerable and can be hit out of her FB. Tests to prove this using moves with long active windows like Ky Stun Raise (FB j.d) showed otherwise. While Jam did get hit by the defender's move, the Hyappo actually connected. This evidence would imply that Jam is indeed invulnerable to strikes on the 14th frame, losing invulnerability on the first active frame, allowing the move to trade with, in this case, Ky's FB. Obviously, this conflicts with the data given in the frame data charts. Similarly, when looking through the frame data on Axl's S Bentengari (his S DP), data showed that this move grants him above ankle invulnerability on frames 1 through 4, yet his startup window is listed as 5 frames. To my knowledge, Axl's S DP trades with many mid and high moves, again implying that the move actually has above ankle invulnerability through frame 5, allowing the move to trade with others when it would otherwise be beat. So I have found two instances where the listed frame data is not accurate, based on experiment and match experience. Is the frame data really not accurate, or am I not accounting for some factor that accounts for these, and quite possibly other, one frame discrepancies?

[Teyah]

I think the problem here is that you aren't reading the data correctly. For Jam, she is invincible from frame 3 -> 13. Then the move goes active on frame 14, where she becomes hittable - but since the move is already active, it is guaranteed to come out for at least that first frame. Same for Axl - he's partially invincible all the way up to the first active frame (5).

[rtl42]

iirc this was brought up in the frame data thread (and i don't know why you didn't just post in there, btw >:o ). start-up # = frame # that the move becomes active. active # = frame interval for which the move is active. recovery # = frame interval for which the move recovers. so, if that's correct, then maybe that clears things up for you? and yeah, it's kind of a silly convention, but i heard it had something to do with writing frame data the way they used to do for older games, or something. (i'm sure POscrub knows better.)