FFXIV Ninja in 3.1: A Numerical Prediction
3.1 introduces a few changes to Ninja: there's a TP reduction in Shadow Fang and Mutilate and mudra fixes. I'm predicting how much extra sustain time a Ninja receives with these fixes, following this priority chain:AC > DE > SF > MU > AE .And this rotation:DE -> SF -> MU -> AC|AE -> DE -> SF -> AC|AE -> DE -> MU -> SF -> AC|AE -> DE -> SF -> AC|AE -> MU -> DE -> SF -> AC|AE -> Repeat
No real way to go in-game right now and test out the new reductions myself because 3.1 isn't released, so I have to make some reasonable assumptions for the model:
1.Huton is active throughout the entire test and the ninja has 2.1 GCD.
2.AE is automatically used in place of AC. They take the same amount of time and have the same number of actions in their combos. Normally, AC would be substituted in for AE per the 40 second rule but I elected to ignore that.
3.Only GCDs are used and there is virtually no GCD clipping.
1000 TP at Start; TP+60 every 3 seconds after start.
4.No Invigorate/Goad used in the base case, and no Goad used in the Invigorate case.
5.There are no breaks between attacks and the target is a dummy.
6.Virtually no latency/lag between attacks.
A few tables and figures were moved around and created: time to complete combos, actions involved in combos, altered base rotation, and GCD TP values. The data provided from them are used later when making the cases.
For the base case, the following rows were made: Time (s), Ability, TP Old, TP New, TP Regen, Old Total, and New Total. Columns were made with .1 second time intervals and the time value is in the Time (s) row. Ability row hosts an icon with the ability's name when used at a certain point in time. TP Old and TP New are the TP costs of abilities before and after 3.1. TP Regen is value 60 added to the totals every 3 seconds. Old Total and New Total are how much TP a player has based on the TP costs in TP Old and TP New.
After filling in the appropriate values, the table below was constructed and held the results.
Table 1: No Goad/Invigorate Case
Category Time Units
Est. Pre 3.1 113.4 seconds
Est. Post 3.1 128.1 seconds
% Increase 12.96 percent
7.5% Tol. Up 137.7 seconds
A 7.5% tolerance represents the quantifying assumptions in: GCD clipping, taking break between attacks, etc. Some mudra sequences, such as Huton, Raiton, and Suiton, have 2-3 mudras involved and can clip GCD. Additionally, there may be mechanics that force you to take breaks so you're not always hitting a target.
To see how well adding a TP regeneration skill works, we slotted in Invigorate (Lancer, lvl. 22 cross-class skill). For Ninjas, the TP boost is only 400 compared to a high level DRG's 500. Since it only restores 400 TP, the skill is only used when either Old Total or New Total are less than or equal to 600 at any point during the model. To simulate a GCD, the skill waited .5 seconds before activating after the total hit at or below 600. Additionally, If it did hit either one of those totals, it only increased the total that hit the limit switch and left the other one alone; the skill was disabled for 120s (its cooldown) after use.
With the Invigorate @600TP assumption included, the model was adjusted and produced the results below:
Table 2: Invigorate @600TP
Category Time Units
Est. Pre 3.1 161.7 seconds
Est. Post 3.1 233.1 seconds
% Increase 44.16 percent
7.5% Tol. Up 250.6 seconds
Like in Table 1, the 7.5% Tol. Up category multiples Est. Post 3.1 by 1.075 to represent quantified assumptions. There's a large difference between Est. Pre 3.1 and Est. Post 3.1, though. Invigorate was first used in Pre 3.1 at 44.6 seconds and in Post 3.1 at 46.7 seconds. Because the former ran out of TP for the next action before Invigorate could be used, their runtime stopped at 161.7 seconds and did not continue. The latter, Post 3.1, achieved a second Invigorate at 166.7 seconds and continued its run until it ran out of TP at 233.1 seconds. With the 7.5% tolerance multipled one, this value became 250.6 seconds.
There's a significant (or noticeable) improvement in how long a Ninja can maintain their rotation. The numbers and estimates generated, however, may not be completely representative in practice based on the assumptions used.