Board Thread:Wikia Discussion/@comment-26091666-20151103015609/@comment-26486243-20151129182143

API wise, the armorbreak takes only two values, 0 and 1. While the 'incomplete' sources  hypothesise that there's at least 3 debuff levels (none, M only, JM combined), if that was true, it'd imply the game server has a way of notifying the game client of the precise state of the armorbreak OUTSIDE the API, which from all we know so far, is impossible.

The theory according to pure API data would be that there are only two states: armorbroken and non-armorbroken. When the Submarine Princess displays the armorbroken graphic, she's armorbroken; when she does not, she is not. As far as we know, the activation conditions for the armor break is 'J node boss kill + M node boss kill', where 'kill' here is taken as 'boss only is sufficient' by some sources and 'S-rank is necessary' by others. Between how the graphic does not change in STF state as well as the name of the map ('sortie a CTF to...'), the only possible conclusion is that STF do not have access to the armorbreak mechanic while CTF does, and the intent of the developers regarding player behaviour on Hard is to:
 * 1) Drain boss HP with STF until last dance
 * 2) Defeat J node in STF
 * 3) Defeat M node in STF
 * 4) Swap to CTF to get final kill, calling node support expedition in order to bring escort fleet safely to boss

To override the API-based information requires statistical information of a precision and reliability greater than what we're using as a source at the moment. As Fuji has mentioned, if the armor values are derived from data, they should differ slightly due to statistical error - identical sets of values between STF and CTF implies that the armor values are not derived from raw data, but from some form of theory-based framework.

Remember also that the devs announced that this event will have mechanics significantly easier to understand than the critdmg% buff system in Summer. An armor break mechanic that triggers in spite of the graphic not triggering does not fit the description of easy-to-understand, but more 'deliberately confusing'. And if it was a bug, it should have been fixed at this point, since all the previous bugs have already been detected and solved.

Based on the above points, I would disregard the source and consider the armorbreak to be a binary value (on/off only). Remember that our armor break information from E5 also initially reflected 15 kills in HP bar from our frontliners' information, before someone replaced it with the (wrong) claim from Wikiwiki that it took 20. At this point, unverified sources claiming to be based on stats but without the raw data to back it up may be worse than no source at all.