Stacking Deflection Bonuses - How Much is Too Much?

[Strill]

So based on the wiki, you can get...

 

Fighter Base: 25 Deflection

Large Shield: +35 Deflection

Cautious Attack: +15 Deflection

Weapon & Shield Style: +10 Deflection

Superior Deflection: +5 Deflection

Defender: +5 Deflection

Guardian: +10 Deflection

 

This brings you up to 105 Deflection. Just how much accuracy are enemies going to have? A character with this much deflection is completely immune to enemies with 10 or less accuracy, and that's before we even consider equipment. Doesn't that kind of break the game?

 

With this many deflection bonuses, the discrepency between specialized tanks and non-specialized characters becomes extremely wide.  Either enemy accuracy scales so high that non-specialized defenders are woefully inadequate, or you can create a tank character that's invulnerable to conventional attacks. Either way it seems like these numbers break the game, provided that they're correct.

Edited November 16, 2014 by Strill

[Valorian]

Everything is fine except (large) shields; they give too much deflection.

 

Also, W&S style is a passive net benefit to deflection and is more powerful than the other listed abilities, which are modal, that grant deflection at the expense of something else.

[Sensuki]

There's a buff suppression system so this is never the case. I'll make a level 12 Fighter now and tell you how high I can get it.

Edited November 16, 2014 by Sensuki

[Strill]

Everything is fine except (large) shields; they give too much deflection.

 

Also, W&S style is a passive net benefit to deflection and is more powerful than the other listed abilities, which are modal, that grant deflection at the expense of something else.

Most of that time, the "something else" is just damage. Is it really such a loss to sacrifice the damage on your tank in exchange for invulnerability?

 

There's a buff suppression system so this is never the case. I'll make a level 12 Fighter now and tell you how high I can get it.

What, some traits/modal abilities don't stack?

[Valorian]

 

Everything is fine except (large) shields; they give too much deflection.

 

Also, W&S style is a passive net benefit to deflection and is more powerful than the other listed abilities, which are modal, that grant deflection at the expense of something else.

Most of that time, the "something else" is just damage. Is it really such a loss to sacrifice the damage on your tank in exchange for invulnerability?

 

 

 

Depends on whether the AI is good enough to not exclusively target the closest character.

[Sensuki]

Okay I managed to get 138 Deflection with

 

Moon Godlike Fighter

20 Intellect

Defender Mode on [Modal, can only have one modal at a time]

Wary Defender Talent

Superior Deflection Talent

Sword and Shield Style Talent

Exceptional Large Shield equipped

 

That's with no buffs, it could be higher with an item that granted bonus Intellect and an item that granted bonus Deflection.

[Strill]

Okay I managed to get 138 Deflection with

 

Moon Godlike Fighter

20 Intellect

Defender Mode on [Modal, can only have one modal at a time]

Wary Defender Talent

Superior Deflection Talent

Sword and Shield Style Talent

Exceptional Large Shield equipped

 

That's with no buffs, it could be higher with an item that granted bonus Intellect and an item that granted bonus Deflection.

So then there is indeed a serious issue with stacking deflection bonuses?

[Sensuki]

No there isn't because enemies will have probably the same or higher Accuracy as that. The Dragon in the trailer had over 100 accuracy for instance. That content is probably balanced at level 7-8 or so. There will be side content more challenging than that.

 

The Moon Godlike character I created was level 12, that's the max level.

Edited November 16, 2014 by Sensuki

[Sensuki]

I saw some suppression rules on Josh's whiteboard in the twitch stream so it's possible that they intend to change it

[Strill]

No there isn't because enemies will have probably the same or higher Accuracy as that. The Dragon in the trailer had over 100 accuracy for instance. That content is probably balanced at level 7-8 or so. There will be side content more challenging than that.

The Moon Godlike character I created was level 12, that's the max level.

 

 

That's exactly one of the issues I mentioned. If anyone but a dedicated tank gets hit they're screwed.  IMO that's even worse.  I hate it when games try to create "difficulty" through increasing enemy damage rather than puzzle-based encounters.

Edited November 16, 2014 by Strill

[Sensuki]

You're probably right there, actually. Other party members will get critted to the high heavens. That's one of the things I don't like about the Attack Resolution system.

[rjshae]

Mathematically, it looks like a problem with trying to add bonuses when really the probabilities need to be multiplied. Possibly they could simulate it a little better by computing a weighted sum, with the higher weights being applied to the larger bonuses. For example: 100% * D1 + 85% * D2 + 60% * D3 + 50% * D4 + ... .

[Lephys]

I bet they haven't tuned those numbers yet.

 

But, passive/permanent Deflection bonuses (armor, talents, etc.) should probably not be able to get you very high, while active/temporary bonuses can much more easily get you to "ludicrous" levels of Deflection without really being too ridiculous (especially with modals and the like detrimenting your offensive capabilities as a trade-off).

 

I'd really prefer to see this stuff fine-tuned to where the biggest passive gap between a naked Wizard's Deflection and a decked-out Fighter's Deflection is maybe 30 points or so, with each point of Accuracy making a pretty big difference. You don't really want to get to where one character has 20 Accuracy and/or 20 Deflection, and another, without any buffs or modals, just waltzes around with 70+ Deflection and/or Accuracy. Because, as has been pointed out by others, you end up with those Max-guy-vs-Min-guy scenarios, in which the numbers are just ridiculous. "50% chance to crit?! REALLY?!"

 

Now, if an enemy creature, especially something like a dragon, has ludicrous numbers, that's different. But, you shouldn't run into a party of 5 guys, 3 of whom have 80 Accuracy and/or 80 Deflection , and you have 2 people with 70 Accuracy. It needs to all revolve around an average, with a 10-point difference mattering a good deal.

 

Maybe for the purposes of PoE's design, D&D's 5%-per-point to-hit system was a little too much. But, at least when you got +1 to that or AC, it felt significant.

[Strill]

I think they would've been better served taking a hint from D&D 5e. They made it a point to be very stingy with d20 bonuses so that +15 is the general limit to permanent d20 bonuses, and no DC goes beyond 30. Permanent AC bonuses are also extremely limited, and the vast majority of AC sources are mutually exclusive.  They also made it so that the bulk of those bonuses are unconditional. Even a very poorly optimized character will probably end up with a passive +11 to their attack rolls at max level.

 

That means that you never get enemies who are completely unhittable, nor do you get enemies who you can always hit.  That also means that even weak enemies can still be a threat in large numbers, even at very high levels, which improves gameplay variety.

 

With the more limited scaling of +Hit and +AC values, characters in D&D 5e instead scale primarily through health bonuses, damage bonuses, and extra attacks.  That makes the scaling curve closer to linear than exponential.

Edited November 18, 2014 by Strill

[Sensuki]

Now, if an enemy creature, especially something like a dragon, has ludicrous numbers, that's different.

 

Not really, it's even worse - if a dragon has way higher accuracy than average party defense, then yeah their breath attacks are going to be one shot critting party members just like in the pre-order trailer, two characters dropped in a single hit. Gee, so fun.

 

That's one problem with this (or these) system(s), unfortunately.

Edited November 18, 2014 by Sensuki

[Lephys]

True... I shouldn't have said "ludicrous" for that, because I just meant that if something like a dragon has 10 more accuracy/defense than anything else in the game, that's one thing. If it pushes the bounds just a bit, it just makes it a huge challenge, without making it just-plain-stupid.

[general_azure]

If I understand you guys correctly, the actual problem here is not the deflection value but how hit chance is calculated from that. Slap some gaussian pdf on it and most of your problems should vanish (hit chance always greater than zero, rather flat in the region of characters who didn't invest in this stat, noticeable slope for those who did, flattens off near the end for diminishing returns)...

 

Out of curiosity, what calculation is used right now?

[Lephys]

It's attacker's Accuracy minus defender's Deflection (in the case of Deflection being the defense that's attacked). When that number is 0, the Attack Resolution is as follows (on a d100 roll):

 

5% Miss(1-5), 45% Graze(6-50), 45% Hit(51-95), 5% Crit(96-100).

 

The ACC-DEF is a modifier which shifts the scale, so to speak. So, if your Accuracy is 10 greater than your opponent's defense, it shifts in your favor, to:

 

0% Miss, 35% Graze(1-35), 50% Hit(36-85), 15% Crit(86-100).

 

And if your Accuracy were 10 points lower, the opposite would occur (no chance to crit, higher chance to miss and graze, etc.).

 

I apologize if I've gotten a detail incorrect, but that should be the general idea.

[general_azure]

If I've gotten that right, this means every 5 points advantage in att vs def lets you shift a d20 roll by 1. So the 35 def listed for the large shield above would be the equivalent of 7ac in good old BG (actually even better, since it also reduces the enemy crit range by a whopping 7 points which was impossible back then).

The idea that you can completely shut down some of the outcomes (because there is no guaranteed minimum 5% chance) is just icing on the cake. And yeah, same thing goes the other way around for offensive stacking as well.

 

Somehow, I don't think this scaling will stay

[Lephys]

Oh, well, sorry... I did forget to mention that there's a minimum 5% chance to Graze (on the up-shift) and Hit (on the down-shift). Basically, you can never ONLY hit and crit, and you can never ONLY miss and Graze.

 

Of course, that may need to be adjusted, too. But, yeah, having like 150 Deflection is a bit ridiculous, when one person on your team has 140 Accuracy and everyone else has less than 100.

 

The differences shouldn't be that huge, I don't think. I mean, if crazy Deflection is one enemy's shtick, then sure, I guess. "Oh no, this giant turtle thing... it's just really hard to hit!". It has other defenses, so there are ways to target not-Deflection if you use the right abilities and plan accordingly. But, the biggest issue comes with crazy accuracy numbers. Because, as others here already pointed out, if something with 100 Accuracy attacks people with less than 60 or 70 Deflection, they're gonna be real real dead, real real fast.

[general_azure]

Ah, there's the good old "1 in 20 no matter your stats" I was missing. Sounds better this way ^^

 

I'll crunch some numbers and see how much raw damage each point of accuracy is worth in various ranges, should make the comparison more solid.

[Jarmo]

Yep. Looks a bit bad when a shield+sword guy has easily 50 more deflection than someone with a 2-hander.

No doubt enemies are designed so they'll still hit and damage a typical tank, meaning something like 50 more accuracy

than a 2-hander has deflection.

 

..making my plan of going melee with a pollaxe something like "it's a valid tactic, only you get kilt real quick-like".

[general_azure]

There you go, fancy colorful lines (the x-axis is accuracy-deflection, in case you're wondering)... but I think we're still something missing about that calculation. Those plateaus where hit chance becomes constant (because miss chance can't go any lower and the probabilites from grace already go to crit) look pretty weird to me.

This would mean that your chance for a regular hit can never be above 50% and critting becomes the most likely outcome from accuracy-deflection=50 onwards.

 

Other than that, the offensive part of the system actually looks rather good to me. Sure, you might end up getting a lot more crits than in other games, but the rather small multiplier should make up for that.

On the very defensive end the effective HP rise pretty sharply until the cutoff, so stacking bonuses is way more effective there (result of the 1/dam in the calculation, happens in almost all games).

[Sheikh]

Wouldnt this need to be in the backer beta forum?

[waveman]

on PoTD deflection is everything, front line combatants need all the deflection they can get, otherwise they just don't last.