DPS Balancing between classes.

Level 7
Joined
Apr 16, 2014
Messages
379
Edit: THESE TESTS WERE DONE in 1.2A(12)

So I was doing a little bit of DPS testing on the practice dummies in Mytargas. I was trying to ensure that all the DPS classes came out with their pluses and minuses. Its important to me that each class fills a role, and that some classes don't become too much greater than others.

With that being said I'll post the results of my approximated DPS tests. These tests were done with what I thought would allow for the most versatile of the characters. So I wasn't exactly trying to just pick all the 'max' dps talents (though in a lot of cases I did end up picking ones that related more to DPS than support).

All characters were level 50 with what should be max gear. Changing out one or two pieces of gear for their alternative piece from the foothills region shouldn't have a huge disparity in DPS anyway

Sorcerer - 380 DPS

Necromancer - 325 DPS (Likely could of been higher with better play. Interrupted a few spells too early)

Hunter - 250-310 DPS (Skill resets played a big key here and some other things down below)

Bishop - 325 DPS

Berserker - 250-290 DPS. Crits played a large factor here, as I took the 50% chance of applying gaping wounds on crit. I also took the berserker rage damage malus removal every time, as I love this talent and think its the greatest thing ever for survival

Assassin - ~325 DPS (Its higher because of into the shades + bladefury from behind, but this was hard to actually calculate)

Druid - ~210 DPS. Efreet + Nymph with heal gear and flaming arrow/barbed arrow. No actual druid auto attacks were included

Monk - 230 DPS. This was more of a tanky build. Took the ignore armor with Crippling instead of 5% chance to debuff with crippling.

Bard - ~225 DPS.

Crusader - ~150-170 DPS.

So a couple things stand out here..

1. Sorcerer is still the highest raw damage output. This is fine, if hes forced to move his damage takes a much larger hit than any other class. Necro could of been higher, he was not played with great efficiency, but is still a very good overall DPS, and works much better than Sorcerer should they both run out of mana.

2. Hunter varies greatly. She relies almost 100% completely on her abilities to do damage. Shes better than a necromancer when oom'd, but still takes a large hit to DPS if she runs out of mana.

Also, I noticed there is a large difference in DPS depending on how you cast your abilities in relation to attack speed. Because of her not so great attack speed, it takes her a while to actually go through her animation of attacking. Its similar to casting a spell for a caster class. If you don't let her finish her attack animation (which is as soon as the arrow is launched, her attack has finished) before using an ability, you lose out on that potential auto attack. For other classes this isn't normally a big deal. Casters do very little damage with auto attacks, thieves attack much faster so if you miss one auto attack because of mistimed abilities its not going to have a big effect. However because hunter does a lot of damage through an auto attack, missing hers due to poorly timed abilities can have a big difference in her total DPS.

So basically, if you're lazy and don't feel like paying much attention, you may want to use bow of the legion. If you're try harding and paying good attention, shalehorn was clearly better.

3. Assassin and Bard are great now.

4. Bishop and Druid are fine as well. The Bishop test was done with pure DPS in mind. In reality a bishop is going to need to move and heal and other things and can't constantly soul strike in the test. Druids can be built to be more heal oriented and thus would reduce their DPS as well.

5. Berserker and Monk are perfect now as well. Monk is more tanky than a berserker, and it shows in the difference in their DPS. Both are still capable of being full tanks, or more DPS oriented depending on how you build, and ultimately I think thats the trade off that is desired here.

6. Crusaders are fine too. They could maybe use a bit of a bump in DPS, but honestly its not worth changing anything imo. They're still by far the most tanky class, and are nearly impossible to kill. By having a crusader in the party it allows for your healers to focus more on DPS than healing, so honestly the crusaders low DPS is made up for by allowing other members to be more oriented towards DPS.

7. One issue... I almost cried when I loaded my Bishop. He had 355 health at level 50. The problem lies within his itemization however. All of the items I had on him didn't give a single amount of health. So I took the cleric talent of +15% health and this boosted him to slightly over 400. I then took Mind Breaker and decided to use that. The new haste formula really devalues these small amounts of haste on items and thus the HP and increased SP from Mind Breaker over White Willow Staff seemed much more valuable. I did not have Hat of the Archmage to replace Blessed Opal Circlet though. If I did I would have gotten over 500 health and been fine.

The issue is.. I don't think I put very much health on the Bishop specific items for the next version. It seems silly to need to use a general hat over a Bishop specific one just because of the health disparity.

Druid somewhat has the same issue.

Every other class seemed to get enough health on things to have over 550 health by taking at least one +15% health talent. So I think just a friendly reminder to double check items in the next version to ensure that theres going to be enough health on items to be at values we'd like to be. When Ihaz and I created the items we were doing it based on the old attribute system so I think we figured we didn't need to stack as much health on certain items because we could stack health through adding constitution values. Bard was 675 health in the old version by just maxing out constitution for example, and now she is well well under that :p.

Has anyone else done some any testing and found any results? Do your numbers match up? Have any issue with my analysis? Feedback would be great :).
 
Last edited:
Level 7
Joined
Apr 16, 2014
Messages
379
Ah you know, I didn't have totem of the troll spirit or whatever it is. Either way with bugged efreet he Oom's so fast that its not a very good sample of a DPS test :p

I think my druid code is from the beta and thus didn't load all my items :(.. I think I have the old code laying around though.

Edit:
Ok loaded an actual geared Druid.. and with glitched efreet was over 400 DPS until efreet oom'd xD.

Edit2:
Respec'd druid into full self DPS + Treant + Non-bugged efreet (Didn't take the pet Spell haste talent). I achieved 270 DPS on a practice dummy. Realistically I don't think I'd ever use full DPS gear on a druid unless I was soloing. Remedy was ticking for a bit over 70 so this obviously isn't going to work for team bosses as a solo healer. I think druid is well balanced and that a realistic figure of ~200 DPS for Treant + Efreet is accurate.

Edit3:
One more test. A Realistic team healer build that still features Treant + Efreet. I Kept flaming arrow + barbed arrow and launched them as much as possible but I didn't do any auto attacks with my druid (as in a realistic situation I should be healing). The end result was about 230 DPS when Efreet oom'd. Pretty decent but in no way broken or OP.
 
Last edited:
Level 7
Joined
Aug 23, 2014
Messages
208
Feedback would be great :).
Awesome testing!

A few thoughts about the purity of the experiment:
- Dummy doesn't have any armor. Therefore, results may vary downwards in real fight, because Armor Penetration value different for classes (some talents even grants 100% armor ignore).
- Dummy doesn't have any evasion. Therefore, results in real fight will be lower for all classes without '+hit chance' talents.
- It would be very informative to have information about each pet's DPS and pure Hero's DPS.
- Pets are OP in this version due to stat gain boost and cast time glitch.

Personally, I think current DPS scatter is too high. In my opinion it would be better to see something like this (tunneling DPS in 1 target):
80% DPS - Tanks (Tank Crus, Tank Monk) & Heal Druid with Nymph. (lower than standart, but still noticeable)
100% DPS - Versatile (Buff Bard, DPS Monk, Tank Zerk, DPS Crus, DPS Druid with Nymph) & Heal Bishop
110% DPS - AoE Melee (DPS Zerk) & Ranged DPS (Hunter) & DPS Bard (because of AoE damage bonus/tunneling damage with bows)
120% DPS - Assassin & DPS Druid with Efreet & Versatile Casters (Buff/Debuff Sorc/Necr/Bishop)
130% DPS - Casters (DPS Bishop, DPS Sorc/Necr) & possible Druid with Servant&Efreet (because of limited mana pool)
 
Last edited:
Level 7
Joined
Apr 16, 2014
Messages
379
I agree the tests aren't perfect, but you do get a rough idea in how the classes compare to one another in terms of potential output. Casters being so much higher than physical can be fixed by giving enemies/bosses a bit higher standard resistances. Then if you'd like specific mechanics to work a certain way (such as the blazing flame being immune to fire spells) you'd further tweaks these resistances on monsters as you see fit.

And outside of the Efreet glitch the pets don't seem TOO op to me. Part of what makes hunter enjoyable is that the pet is actually so strong. Its not like its pushing hunter into something extreme here.

I like your breakdown :p. I don't think the numbers need to be exact, but it would be nice for them to at least break down in that order. Im not too concerned about MLG tier balancing here, Im more interested in ensuring that each class has its perks and drawbacks. If Zwieb feels like putting in the time/effort into trying to ensure each class/build fits into a certain scale like that is completely up to him. I think going through all the items and trying to redefine what does what would be too much work, but certain steps can be taken to make sure certain classes are actually much closer to where they need to be.

As it stands right now, Im not sure why anyone would pick a monk for example. They offer very little DPS, and they're nowhere near as tanky as a crusader. In just about every scenario you'd be better off taking the Crusaders little less DPS in favor of his massive tankiness edge.
 
Level 9
Joined
Jul 11, 2011
Messages
599
yup just did some testing again (i didnt kill the dummy, but I went till I oomed, which was after dealing like 65k damage) and I had 141 dps at the end of it with full int gear, and over 1k hp, so you cant even really say i was try-harding dpsing

Even did a round of just pure auto-attacking with crusader and I didnt drop below 90... after dealing around 10k damage and got bored, but it was hovering around there
 
Thanks for these tests; this is extremely valuable data for me (and btw thanks for the great idea of adding those training dummies ... nice return of investment here ;) ).

After checking out these numbers, I think I know how to fix the numbers into their intended range.

1) Pet stats will get a 20-30% nerf to their stat gain, to compensate the increase of base stats.
2) Thieves and the corresponding advanced classes get a slight boost to their base attack speed
3) Monks get a slight boost to their base attack speed
4) Auto-Stat gain nerfed a little (by 10-15%) for all classes
5) Re-evaluation of auto stat gain of monks, berserkers and hybrids; Rangers (not their advanced types) will get a little bit more int
6) Int-gain of sorcerer and necro nerfed a little (instead, they get a little bit more agi, as imho, they profit way too much from agi due to the low agi score)

The passive playstyle for monks will hopefully change with the new 1.2B skills. I think this will finally make monks a viable class.
 
Last edited:
Actually, after doing the maths again, I noticed that the primary stat of each class isn't really higher than it was in 1.2A(8). In fact, even non-hybrids mostly only gained 15-17 additional points in their main stat (compared to 1.2A(8) with 25 points max). It's just the additional gain in secondary stats that is affecting the stat gain of pets too much. So I'll go with just nerfing the pet stat gain instead of both auto-stat and pet stat gain.


Also, I will refer to this table when designing the new bosses. Very helpful in terms of selecting the HP of bosses as I usually apply this formula:

HP = [Expected battle length in minutes] * 60 * number of players * average DPS of properly equipped heroes * "busy factor"

So, a tank & spank encounter gets a busy factor of 1.0, whereas an encounter that requires a lot of movement gets a lower factor. Number of players is usually 4 for dungeons and 3 for world bosses.
 
Level 7
Joined
Apr 16, 2014
Messages
379
Alls good. I think the super-pet is the main reason for hunter being able to somewhat keep up with casters. With their incoming changes I think the game as a whole will be right where everything needs to be.

Casting haste actually isn't a huge deal at the moment. We're only barely getting to 50%, which isn't all that high when put into the current haste formula. Either way, a bit of a nerf to casters SP would be a good thing overall.
 
Casting haste actually isn't a huge deal at the moment. We're only barely getting to 50%, which isn't all that high when put into the current haste formula. Either way, a bit of a nerf to casters SP would be a good thing overall.
Yeah I like the new formula a lot. It's good to have a new item stat that I can stack farther than 100%. A huge problem that came with +crit and power creep of items.

It would probably make sense to introduce a counter stat to crit chance at some place ("resilience"?), to allow people to stack more than 100% crit and still only crit like 30% of the time against certain bosses. But it feels unintuitive. I guess it makes more sense to just stop putting more and more crit on items, as basicly no crit item comes with crit% alone, so people will replace their +4% crit items with other +4% crit items at a later point, if the additional SP or AP are worth the switch.

I made sure that the stat total of all classes is equal for balancing reasons. But the pure caster classes have the highest primary attribute scores of all classes, so it's not surprising they deal the highest DPS. They are kind of skewed into the glass-cannon direction with a highly focused main stat and barely anything else. Spreading the scores a little bit more over to AGI and STR might alleviate that issue already. After all, as 100% of their damage comes from skills, not from white damage, a 20% decrease in SP also means a 20% decrease in damage dealt.
 
Level 7
Joined
Apr 16, 2014
Messages
379
Yup very true. Just did a test in A(11) with a sorcerer and his DPS was largely unchanged. Which I think is really fine. Even something like 10% magic resists on bosses will have a large impact on a casters realistic DPS in a way that simple armor values don't quite affect white damage quite so harshly.

Also agree with the crit. Ihaz and I did a good job with the next versions items in that we made sure not to increase crit values. The items that were given some crit have deficiencies in other stat gains to make gear choices more thoughtful.

One thing on casters though. Their health is much higher than the leather classes end up with. Also their mana pool is very high. In that sorcerer test I dealt out about 38,000 damage before running out of mana... without using a potion. I didn't take any of the MP increasing talents either.

I think if you decreased their natural health/mana gain, and added level 5 talents for each class that is basically a choice between 15% health, 15% mana, or a minor increase to worn item stats you could solve a few problems. Leather classes benefit by being offered another way to increase health. And you also allow for a basic level 5 talent that enables every class to have something to choose a little bit quicker into the game.
 
Yup very true. Just did a test in A(11) with a sorcerer and his DPS was largely unchanged. Which I think is really fine. Even something like 10% magic resists on bosses will have a large impact on a casters realistic DPS in a way that simple armor values don't quite affect white damage quite so harshly.
Did you test this on a single training dummy or did you use AoE effects on the double training dummy?
Let's also factor in that Sorcerer/Necro damage is very dependent on the movement during battle. While the other DPSers can basicly deal damage as soon as they stop moving (or even while moving), a spellcaster would have to recast. If you play a very dynamic encounter that requires lots of movement, chances are your DPS drops way more dramatical as for leather classes. But I still think another nerf is neccesary. But not in 1.2A anymore...

One thing on casters though. Their health is much higher than the leather classes end up with. Also their mana pool is very high. In that sorcerer test I dealt out about 38,000 damage before running out of mana... without using a potion. I didn't take any of the MP increasing talents either.
Did you compare the health pools naked or with equip? I think part of the problem is the amount of HP on cloth items.

38.000 damage before running out of mana seems... okay? Your test character was a max geared level 50 I suppose? If we divide the 38k by a DPS of 250, then that equals roughly 2:30 min of pure DPS. A more realistic DPS of 300 means only 2:00. Which is a lot less than the expected duration of most future content (5-7 minute battles).
A mana potion on a 5 minute cooldown also fits into that scale perfectly. Which means you can "survive" on mana potions alone if you move like 50% of the time.
I don't want players to be required to use a bard or druid to beat endgame bosses. But it should still feel like a useful addition. And I think this is exactly the case here. Maybe a little nerf by 15-20% mana, but not that much more.
 
Level 7
Joined
Apr 16, 2014
Messages
379
Alright thats fine! All the tests were done on a single target dummy, so no AOE's were used.

38,000 damage just seemed like a lot for one pool of mana. Gargoyle currently has the most HP of any monster in the game and he only has like 56,000? I realize some of the damage would ultimately go towards the bugs but ~38,000 damage from a total mana pool still just seemed high based on our current encounters. As we move forward and fight stronger things then yeah I think our mana pools will be better suited.

I kinda figured during those longer encounters you could partially supplement mana issues by having certain mechanics that spawned some sort of mana regen over time, similar to the healing salve used in the squire class up quest. Timing when the properly use the power ups could be crucial for the overall encounter!

At any rate, after testing most of the rest of the classes, things seemed to turn out great!
 
Alright thats fine! All the tests were done on a single target dummy, so no AOE's were used.

38,000 damage just seemed like a lot for one pool of mana. Gargoyle currently has the most HP of any monster in the game and he only has like 56,000? I realize some of the damage would ultimately go towards the bugs but ~38,000 damage from a total mana pool still just seemed high based on our current encounters. As we move forward and fight stronger things then yeah I think our mana pools will be better suited.
Remember that we are talking about a boss that is balanced around the fact that melees can't hurt him 50% of the time and that you have to deal with ground spawns constantly. And we are also talking about an endgame geared level 50 character. All the current bosses are balanced for "green geared" level 40-45 characters. And for teams of only 3-4 people (as they are outdoor bosses). And for those, 56k HP can be a tough nut to crack.

The new 1.2B bosses, will be designed for foothills-geared level 50 characters by default. Which means they will be a huge step upwards in durability and damage output.
Most of the bosses will have at least 200-300k HP. This is balanced around the assumption (again, thanks for all this data, it's extremely helpful for future balancing!) that endgame geared characters deal roughly 250DPS under ideal conditions. This equals roughly 3-4 minutes of combat in a party of 4. Busy encounters should take longer.

But yes, I agree; a little nerf to max mana won't hurt (again, 15-20% less should be perfect; I don't think it's a good idea to cut it harder than that, with future content in mind).

I kinda figured during those longer encounters you could partially supplement mana issues by having certain mechanics that spawned some sort of mana regen over time, similar to the healing salve used in the squire class up quest. Timing when the properly use the power ups could be crucial for the overall encounter!
This is more part of the boss design then. But the general solution for mana issues involves new items and abilities.

At any rate, after testing most of the rest of the classes, things seemed to turn out great!
Nice to hear. It would be great if 1.2A(11) is the last iteration of 1.2A. I never really intended to take so many iterations on this update...
 
Level 7
Joined
Apr 16, 2014
Messages
379
Updated the first post with the changes that have been applied in 1.2A(12).

Overall everything is great. Just a few itemization issues in regards to health for a couple classes but thats easily fixable.
 
Yeah I'm pretty satisfied with the new results. I suppose the 380 DPS on sorcerer are with specced water elemental? I think Water elemental is a little bit overpowered currently, as it literally does 100 DPS on its own... then again, the new abilities that come with 1.2B are mostly utility spells for sorcerers, so I expect DPS to even out a little compared to the other classes that all get at least one new fast-CD DPS spell.

Itemization is a problem, yes. Though not really for the current content (there's always the option to take the +15% life and 400HP should be enough for the current meta, especially with high resistances and divine shield).
Would you guys be willing to check the new items again and add some HP across all the items? I really think we need to up the HP of all characters in general for upcoming content; compared to the average health of creeps, characters are extremely weak in terms of HP.
 
Level 7
Joined
Apr 16, 2014
Messages
379
Yeah that is likely something we can do over this week if necessary.

And yeah the sorc was spec'd with the talent for increased water elemental time. Either way with sorc being high, its not like hes massively better than anything else considering a realistic playstyle.
 
Yeah that is likely something we can do over this week if necessary.
Great; I can send you the map on thursday afternoon; I won't have much time on the weekend anyway, so we wouldn't even need to merge data afterwards.

And yeah the sorc was spec'd with the talent for increased water elemental time. Either way with sorc being high, its not like hes massively better than anything else considering a realistic playstyle.
True. And I see nothing wrong with Sorcerer being top DPS. It's the typical glass-cannon, after all and if you intentionally leave out all the survivability and utility abilities like spell mirror, dispel magic, fire shield (and the upcoming blink ability) just to burst the DPS meters, then that's totally cool.

I also like how the low attack speed of hunters require the player to time abilities perfectly to score better DPS. Adds a nice element of skill to a class that has a clear edge over other DPS classes due to having the highest range and sustainable damage that doesn't add threat to the character from the pet.
 
Top