Bullying the Weak (X95 Equipment & Skill Analysis)
While SR-3MP may be the most popular direct-fire offtank in GFL, X95 makes for an interesting alternative.
First obtainable as a drop during Isomer, X95 is a 5* SMG who targets low-HP enemies during her skill and gains a damage boost proportional to their missing health percentage. Compared to Shrimp, she has 3 more base FP, 1 more accuracy, and the same EVA/ROF/HP. She trades Shrimp's crit rate tiles for accuracy, gives a 3% smaller ROF boost, and hits the whole backline with them while on 8. Her skill will almost always give a smaller boost than Shrimp's 260% multiplier.
As done before, let's start with an equipment chart for X95.
One new addition here - if you've oathed X95, instead refer to the boundaries set by the dashed lines. The colored zones correspond to the solid (unoathed) lines. It would appear that oathing her doesn't change much when deciding between EOTs/ITIs, but the small accuracy bonus it provides does give VFLs a slightly better chance of being the best choice.
Now, to keep this from being an entire copy-and-paste of the Shrimp thing, let's get some data and see if we can estimate the mean effective damage boost from X95's skill. I'll do the same for Shrimp as a way to estimate the base error.
Note: my X95 is oathed. To keep the comparison fair, I'll use equipment at slightly less than max enhancement to "de-oath" her without use of any unethical affection-modification techniques. She gains 2 FP, 1 acc, and 3 EVA (see this handy site) from oath. A +9 maximum calibrated ITI gives +28 accuracy, so X95 will use a properly maxed out sight and have a 1 acc advantage over her unoathed state. Before external buffs, this changes her hitrate against 15 evasion from 75% to 75.41%, so it's very minor.
In some fights against the large nightmare group, X95 gets knocked down to 4 links towards the end. She stayed at 5 links in the other battles.
Random targeting, overkill, corpse whipping, that 1 point of accuracy, and such make it hard to get any exact results. The situation isn't entirely doomed, as we can run several trials to lower variance. Comparing her results with some other reference options will also help.
| Enemy group | Backline options tested | Fairy used |
|---|---|---|
| Nightmare (x155) | AN-94/M4A1/G41 and G36/AUG/G41 | Artillery (5*, Damage II) |
| Nightmare (x72) | AN-94/M4A1/G41 | None |
| Aegis (x12), Nemeum (x24) | AN-94/M4A1/G41 | None |
Luckily, X95's total damage dealt is linear with respect to average percent of enemy HP missing. This lets me compare her damage dealt in the fight, the fight length, and simulator numbers at two different percents to find her effective damage multiplier and the average percent health missing of the stuff she shot. The runs with Shrimp gave a rough idea of the error between real combat and the simulator. Shrimp's multiplier by this method was between 222-246%, taking a 5-20% hit from the expected value depending on the encounter. It got less accurate the lower the effective HP/link of the enemies were, as expected.
Conclusion
Here's the summary of my estimated values for X95:
| Enemy | Damage multiplier | Mean Missing HP% |
|---|---|---|
| Nightmares x155 | 189-214% | 63-71% |
| Nightmares x72 | 179% | 60% |
| Aegis x12, Nemeum x24 | 113% | 38% |
Target practice enemies are limited, making it difficult to generalize these results. Against low armor, her performance was somewhere in the middle between Shrimp's and weaker direct-fire offtanks like Honey Badger and C-MS (damage mode). X95's skill provided an effective damage buff around 160-200%. Versus high armor, she suffers more and contributed very little against the Aegis/Nemeum node tested. Armor in general is an enemy of ARSMG teams, so this is a fight you should bring RFHG/MGSG for anyways.
While Shrimp has a clear advantage in straight damage, X95 does fairly well herself. She offers a unique targeting method and different tiles. For typical content, if you're simulating some teams, I'd put in 60-70% HP missing for X95 when comparing options.
As usual, further tests would be useful to see how all this varies against different enemies. These results were based on a total of 58 battles, rather arbitrarily distributed between nodes.
More like this
Experiments vs Molotovs