Fixing Evasion without touching Evasion? Revisiting Acrobatics/Phase Acrobatics

Summary:
-------------
Dodge is now an additive bonus, rather than multiplicative. This adjustment means that by the time you complete the dodge circle, evasion begins to provide fairly linear returns at moderately high evasion rating. Actually, slightly increasing returns. Completing the dodge circle eliminates diminishing returns from evasion rating.

Scale of benefits for average players:
-----------------------------------------
At 50% evasion from rating, this system is just as powerful as the old system.
At 55% evasion from rating, the player gains 1.2% more total avoidance than under the old system.
At 60% evasion from rating, a 2.4% gain.
At 65% evasion from rating, a 3.6% gain.

This is the range I'd expect nearly everyone to fall into.

Below 50% evasion from rating, characters take a similar loss under this new system. That's half the reason why our proposal includes a new life notable inside the A/PA circle. The other reason for that node is just to provide another strong benefit to weigh against giving up armour and ES.


Breakdown:
--------------

Acrobatics: 8% additional chance to evade attacks
Removes all armour and energy shield
Discipline and Training: 12% increased life. +30 life
Acrobatics Improvement: 1% additional chance to evade attacks
Phase Acrobatics: 20% chance to dodge spell damage

The layout is somewhat important; specifically, having disc. and training near the beginning is kind of important. That's because the actual acrobatics keystone loses a little power. Having a fat life bonus early in the circle offsets it. Also, these values are chosen carefully against the cluster costing 7 points. If the values change, the point cost should probably change too.

---------
Thanks for everyone's input.

I only say this because of the amount of work put into this... and I think it can be presented better:

Can you summarize what that would provide characters at low, middle, and high level? Although perhaps intended, it wasn't clear to me in that summary, nor the previous graphs which might have been better with labels and formulas.

From what I gathered based on what you stated, at extremely high nearly perfect gear you can get up to 5% more total avoidance, with diminishing returns for lesser gear...

For this, early game takes a hit on defense (it doesn't look like a small hit from the graphs.)

I am fairly sure that explanation doesn't really encapsulate the purpose of this overhaul. Making evasion scale better at levels of evasion rating that are so astronomically high that only a few players would ever reach it... is good mathematically, but the practical application and consequence to pure evasion players seems negligible.

The overarching question is: what does this change for the average evasion player at low/mid/high mean, and by how much?

Even if it's only applicable to extreme end game, I do feel it's still worthwhile, but more than likely 2 formulas would need to be used... one for the normal game, and yours after the cross-over point.

edit: I see, you were suggesting I change my summary. Got it.


Sure, zeto.

Yes, poorly geared players wind up with less avoidance in the new system than in the old system; this is one of the reasons why it needed a flat life modifier. Discipline and training is more than enough to make up the difference. The graphs don't even account for that flat 30 life (though they do include the 4% incr life from trading those points), so the new EH curve should actually always be higher than the old EH curve.

I'm assuming most players would gladly give up an 8% incr life node for a discipline and training node. Some others will buy it without dropping an 8% node from their build.

Hm, the best way to do this is with a table. It breaks even at 50% evade from rating and gains ground from there:

%EV (from rating) - total avoidance(old system) - total %EV (new system)
35.00%....................50.60%.............................47.00%
40.00%....................54.40%.............................52.00%
45.00%....................58.20%.............................57.00%
50.00%....................62.00%.............................62.00%
55.00%....................65.80%.............................67.00%
60.00%....................69.60%.............................72.00%
65.00%....................73.40%.............................77.00%
70.00%....................77.20%.............................82.00%

I think 70% chance to evade from gear alone is pretty much impossible. It's just like, total max case for OCD evaders under a defensive flask. That case needs to be included in the analysis, though. Let's say 99% of all players picking up acrobatics will fall between 50% and 65%. So between breaking even and gaining around 4% avoidance, that's where most people land. Plus the extra life.

If you want to move the break-even point a little lower, 1 or 2 more acrobatics improvements would do nicely. But each one has to cost a skillpoint.

Is this a good answer, or do I need to do a better explanation?

Speaking from experience, you would never take acrobatics with less than 55% chance to evade.

The numbers can be tweaked. My concern is, is this layout okay for you to work with? If so, I step back and let you smart folk do your thing. :)

No, that's perfect, thanks a ton!

The only reason I asked for your help a second time, it's a pretty major difference to put the life at the beginning or at the end. It would have been irritating for me to leave that as it was.

I know the numbers would probably change.

Edited the summary so it's a little more organized and less confusing.

So overall, you take away some early avoidance but give them life to compensate. You then give end game avoidance if they specialize in evasion via acrobatics and what's locked behind it at a rate of 1-4% over the old system.

I assume total avoidance over EV incorporates block as well as acrobatics?

The ultimate question is perhaps: are these points now worth it? Is the modest improvement in overall total avoidance at end game worth spending all those skill points as an opportunity cost against alternatives... namely armor/life/damage? (In other words, would people WANT to go this route due specifically to these changes or are more changes required?)

No doubt there's no direct answer to that... perhaps with the other tree changes acrobatics is less of a death trap in either system ;)

Now best get a short summary, a few lines, + a decent photoshop, and put it up in the first post so people coming into the thread can see the end result.

That aside, even at 40% evade the node would provide more, and easy access to a juicy health node.
That's two points for a noticeable jump in survivability, at the cost of barely anything at that level.
Only then the advantages of the new system begin to approach the old one is it really noticeable to have only evasion as a defense.

"

zeto wrote:

I assume total avoidance over EV incorporates block as well as acrobatics?

It's really hard to do the analysis with a shield, so no. But I can tell you, qualitatively: the difference between the new system and the old system shrinks when you have a shield equipped.

That's because blocking is still a multiplicative whatever percent of attacks that get through evasion. If you gain avoidance from another source, avoidance contributed from block goes down. If you lose avoidance from another source, avoidance contributed from block goes up.

So if the new system grants me 3% higher chance to evade, my shield probably grants me 1.5% less avoidance.

Not sure if I'm being very clear there. I'm trying to get better with these explanations.

"

The ultimate question is perhaps: are these points now worth it?

I would suggest yes, though I think you and I are looking at the whole thing from different angles. Tell me if I'm misinterpreting your point of view.

First, we're not just talking about amazing gearing here. I focused too much on that in the analysis, but that's necessary. I wanted to make sure the system doesn't break anything, that's why. This system provides a nominal benefit (between 0 and 4% avoidance) to almost everyone that will buy it.

So you're asking, is that enough on its own?

But here's the thing: it completely changes evasion scaling. After you've bought the dodge circle, gear upgrades suddenly mean more. Before, another 100 evasion rating or 20 dex on gear didn't mean much because you were already way off in massive diminishing returns land. Same thing for passives: very few people keep buying 10% evasion passives.

This system changes all that. After you buy this circle there are no more diminishing returns. Upgrades and additional %incr evasion rating passives become worth far more than before.

Now with that context, I can answer your question. Yes, it suddenly becomes worth it to buy A/PA. That was my goal, anyway, and I tried to come up with a system that wasn't blatantly overpowered. In doing that I may have under-tuned it.

Now, the scaling might need tweaking to make that a more compelling argument. If the circle provided 14% additional chance to evade, in total, for instance. There'd be a profound increasing return on evasion rating suddenly, and my argument would hold more weight.

Example with block.

A character has 40% block, and 60% evasion from rating.

Under the old system, this character has 81.76% total avoidance.
Under the new system, this character has 83.20% total avoidance.

This character gained 1.44% total avoidance with the change, but more of that avoidance is from evasion, and you'll get considerably fewer blocks. That's good for being stunned less often, and de-emphasizes "block recovery" nodes.

Same character (same %EV), but no block:

Under the old system, same character has 69.6% evade.
Under the new system, same character has 72% evade.

This character gained 2.4% total avoidance with the change.

So with block, you gain less avoidance under this new system (relative to how much block you have) - both characters also get stunned less frequently. But you're already at higher avoidance, with a shield, so that smaller gain is worth the same amount. Relatively, both characters get the same overall benefit.

Does that illustrate it reasonably well?