Salvation math stuff

By SomeKittens, in Star Wars: Armada

I did some simulations, figuring out the best turbolaser upgrade for Salvation. With a concentrate fire, Slaved Turrets wins the day, otherwise TRCs provide reliable results:

E4B9iZ5.png

Results void where prohibited, may cause cancer in the state of California, etc, etc.

- Assuming Salvation is a Support Refit, and you only care about how much Damage you can put out the front :D


I mean, I love the way its done, and the results shown - but all variables need to be accounted for. I for one, get a huge amount of Mileage out of being able to Anti-Squadron Fire out the wide sides as well... But YMMV ;D

+raymus? (re: cf token + cf dial)

Edited by ficklegreendice

- Assuming Salvation is a Support Refit, and you only care about how much Damage you can put out the front :D

I mean, I love the way its done, and the results shown - but all variables need to be accounted for. I for one, get a huge amount of Mileage out of being able to Anti-Squadron Fire out the wide sides as well... But YMMV ;D

The biggest variable I didn't account for was "SomeKittens' diceluck". At the tournament yesterday I never did more than four damage with Salvation (including an opening salvo roll).

I think if you're taking Salvation, "Damage out of the front" is the only thing you really care about.

+raymus? (re: cf token + cf dial)

Might be a side discussion but I think Intel Officer works better for Salvation than Raymus - buh-bye brace.

I don't tie to your results...

Can you state your assumptions on how you are using accuracy vs. defense tokens?

I don't tie to your results...

Can you state your assumptions on how you are using accuracy vs. defense tokens?

Yes, absolutely:

With TRCs, after rolling, they flip a die in this order:

- blank

- single

- acc

slaved turrets has been the only way I've ever run salvation

I wouldn't rule out XI7s for their ability to get damage thru a specific arc on the target.

The problem with XI7s is that you want to spend as many turns on target as possible, which is nice for an MC80, but for a Nebulon it is a sure fire way to drop speed and get some rather sever side shots coming in.

To me TRC seems like the best option.

Sure, lower max damage, but way more options for the ship. Being able to use another arc, use it for anti squadron duty, not forced to use cf every turn (because only with cf slaved is constantly better)

Salvation should be called Feast or Famine.

I either roll 8 hits with a crit or one hit. Least that's what it feels like.

So as a starting point, if we are ignoring defense tokens:

  1. Slaved without concentrate is exactly 4 expected damage. You roll four dice, each die is an independent trial, and each die has an expected damage value of 1 (8 outcomes: 1, 1, 2, 2*, 2*, 0^, 0, 0; the * indicates a crit that becomes 2 damage because of salvation, and the ^ indicates an accuracy).
  2. Slaved with concentrate is exactly 5 damage. Same logic as above.
  3. IF we are ignoring defense tokens, TRC yields an extra 2 damage on the worst die for the roll (e.g. randomize three, discard worst, replace with a crit which also does double damage for salvation). When I simulate this with 3 dice, I get approx. 4.7 mean damage over about 100k trials, and I get approx. 5.8 damage over 100k trials. For the curious, those numbers are 4 and 4.85 respectively without salvation, so Salv is gaining you about .75 to 1 damage per roll in this case.

Thus, I think your TRC numbers are understated (probably because you added flat damage instead of replacing the worst die, which is part of the real value of TRC - as in, if I roll hit, crit, double hit, blank with TRC, I'm flipping the blank to a crit or another double hit if I didn't have Salvation).

Edited by Reinholt

So as a starting point, if we are ignoring defense tokens:

  • Slaved without concentrate is exactly 4 expected damage. You roll four dice, each die is an independent trial, and each die has an expected damage value of 1 (8 outcomes: 1, 1, 2, 2*, 2*, 0^, 0, 0; the * indicates a crit that becomes 2 damage because of salvation, and the ^ indicates an accuracy).
  • Slaved with concentrate is exactly 5 damage. Same logic as above.
  • IF we are ignoring defense tokens, TRC yields an extra 2 damage on the worst die for the roll (e.g. randomize three, discard worst, replace with a crit which also does double damage for salvation). When I simulate this with 3 dice, I get approx. 4.7 mean damage over about 100k trials, and I get approx. 5.8 damage over 100k trials. For the curious, those numbers are 4 and 4.85 respectively without salvation, so Salv is gaining you about .75 to 1 damage per roll in this case.
Thus, I think your TRC numbers are understated (probably because you added flat damage instead of replacing the worst die, which is part of the real value of TRC - as in, if I roll hit, crit, double hit, blank with TRC, I'm flipping the blank to a crit or another double hit if I didn't have Salvation).

Someone did the math a while back, but it was somewhere in the neighborhood of 1.3-1.6 expected damage increase for 3-4 dice thrown.

Edited by Ardaedhel

So as a starting point, if we are ignoring defense tokens:

  1. Slaved without concentrate is exactly 4 expected damage. You roll four dice, each die is an independent trial, and each die has an expected damage value of 1 (8 outcomes: 1, 1, 2, 2*, 2*, 0^, 0, 0; the * indicates a crit that becomes 2 damage because of salvation, and the ^ indicates an accuracy).
  2. Slaved with concentrate is exactly 5 damage. Same logic as above.
  3. IF we are ignoring defense tokens, TRC yields an extra 2 damage on the worst die for the roll (e.g. randomize three, discard worst, replace with a crit which also does double damage for salvation). When I simulate this with 3 dice, I get approx. 4.7 mean damage over about 100k trials, and I get approx. 5.8 damage over 100k trials. For the curious, those numbers are 4 and 4.85 respectively without salvation, so Salv is gaining you about .75 to 1 damage per roll in this case.

Thus, I think your TRC numbers are understated (probably because you added flat damage instead of replacing the worst die, which is part of the real value of TRC - as in, if I roll hit, crit, double hit, blank with TRC, I'm flipping the blank to a crit or another double hit if I didn't have Salvation).

There is actual dice replacement there, I explained the order there in my last reply to you.

If that is true, you have some kind of error in your formulae, then...

What initially caught my eye was the lack of difference between the TRC base case and the TRC concentrate case. Adding a whole die with a 1 point damage expectation shouldn't produce a .04 increase in damage. As a floor, it should add 1.

If that is true, you have some kind of error in your formulae, then...

What initially caught my eye was the lack of difference between the TRC base case and the TRC concentrate case. Adding a whole die with a 1 point damage expectation shouldn't produce a .04 increase in damage. As a floor, it should add 1.

eh, I should probably post the code for review. It's entirely possible some number snuck in where it wasn't.

EDIT: here y'are, go wild: https://gist.github.com/SomeKittens/b87e031df0aa8a9b8bbc

Edited by SomeKittens

Welp, I'm an idiot. Typo in my code caused concentrate fire for TRS to be undervalued. Here's some better results:

9JalzOE.png

(Thanks to Reinholt for catching this). Might just be swapping out my turbolasers!

Ok, last one, I promise. Here's the things with better bars and vanilla (or XI7s, if you prefer) data added:

TR1oBZE.png

TRS?

TRS?

I assume he means "Turbolaser Reroute Circuits" but accidentally used an S instead of a C.