What does a healthy talent tree look like?

General Discussion
1

TL;DR: What does a healthy talent tree (or tier) look like to you? What are the in-game examples?

There’s quite an uproar over the Whitemane rework, and to a lesser extent the Junkrat rework. A complaint regarding Junkrat is he only had one build. Chen was reworked and STILL only has one (competitive) build.

Rehgar’s tier 1 and 7 have meaningful, healthy choices (they aren’t perfect, but pretty good). They each provide a unique tactical benefit, taking one noticeably changes the play-style/impact over the other options, and the offer adaptable options depending on the enemy team comp.

Bad Talent trees

Signs of a bad talent tier/tree: majorly disproportionate pick rate of a talent, mathematical advantage of a talent, and/or lacks meaningful play-style shift. Ex: Chen’s tier 1, 4, and 16.

Overall, I think HotS talent trees do a poor job creating meaningful choices. In many cases, I think these issues stem from lack of tier coherency or failure to do basic math.
An example of tier coherency are Qhira’s and Varian’s lvl 7; all health regen talents, but the way the regen activates is different.
Those tiers fail when it comes to basic math, though, thus disproportionate pick rates. Another example of math fail is Chen’s rework initial dps and level 1 talents. These issues still persist.

I recognize creating meaningful choices might be the toughest thing to do in designing a hero, but there are examples showing it is possible. Tier coherency is the easiest way to create meaningful choices, but is formulaic and can stifle creativity/trying new things. Basic math is extremely easy to address, but it’s still not done. The same mistakes with Chen seem to be repeating with Whitemane.

Note: I’m using HeroesProfile to view stats.

1 Like
2

All the choices have a reasonable pickrate, OR if they don’t, it is because it is an ultra niche talent.

Valla’s talent tree is often brought up as a very good tree, and i agree.
Murky’s I would say is pretty good outside of the abomination of terrible that is Egg Shell.
Nazeebo has a pretty good talent tree I think.
Stitches talent tree is good

A majority disproportionate pick rate of a talent doesn’t always make it bad.
Stitches is a really good example of that, where his hook range is by far the most picked.

It isn’t because the other choices are bad, they’re viable, but because getting those hooks feels really good.

Varian’s level 7 on the other hand, is bad because Lionheart is terrible, and second Wind is picked a massive amount.
Victory Rush has a decent pickrate, and on some maps it does keep up with Second Wind. If not for that, I would say that Varian has a healthy talent tree.

His level 1 talents are skewed also, heavily towards HKQ. But that is fine, as the other choices are viable and have their places at times.

So I would say that a disproportionate pick rate isn’t a problem, if the other choices have a niche where they are the best choice.

The thing is, “the strongest choice” is not always the most picked choice.
Look at tracer, her level 16 talent tier, the developers straight up tell you how much extra damage each does, and the lowest damage increase is picked the most!

You might want to note what patch duration you’re using (They allow up to 4 major patches) and what modes and ranking of players (their ranking doesn’t take directly from the game, it is percentage based off of the MMR that your uploaded games gets/gives you).

i.e. When comparing strength of talents if a hero hasn’t had a change in the last few patches, I’ll select Storm League + Unranked Draft with all maps selected and Diamond+Master players.

2 Likes
3

Valla is a good example, with the exception of Caltrops. I still don’t think it’s good even after the buffs, simply because it doesn’t have synergies with anything, in particular at level 7 where taking Death Dealer is an anti-synergy while taking Repeating Arrow means you skipped Monster Hunter at one which is a pretty big no-no.

Why Valla’s tree is good: It has three clear build paths: Q build, Trait/AA build and W build. All of these have at least 3 talents tied to them with possible deviations at the other tiers that still synergise.

For example level 13 for Q build where Siphoning arrow competes with Gloom which is always a good choice.

Or level 16 where both Seething Hatred and Frost Shot are amazing additions to W build.

What is better than having 3 clear build options? Hanzo’s tree that provides three options that are all customisable.

Hanzo also builds Q, W or AA, just like Valla, but unlike Valla, ALL his level 7 and 13 talents boost two possible builds.

The Dragon Hungers (7) and Piercing Arrows (16) affect both Q and W.

Never Outmatched (7) and Giant Slayer (16) affect both AA and W. Note: With Never outmatched you can pick both Simple Geometry (more difficult but safer) and Redemption (easier but less value if you die).

Finally, Sharpened Arrowheads (7) and Flawless Technique (16) affect both AA and Q. In fact, Sharpened affects all 3 builds, but if you want W build, you probably want Never Outmatched.

Then his 4 and 13 tiers are waveclear and utility respectively, but in such a way that they are highly situational. (Except for Ignore all Distractions and Mounted Archery, they’re crap don’t pick them).

And finally, what is better than having 3 viable builds with deviations on multiple tiers? Sylvanas’ tree that does much of the same but literally all her talents are viable.

Sylvanas has a Q build, a W build and a trait/AA build as well.

At level one Might of the Banshee Queen boost all 3 with Unfurling Shadow affecting W build (recommended in longer games/on maps where teams get clumped a lot) or Overwhelming Affliction (paired with a Jaina/Arthas/Reghar/Blaze/etc especially).

Level 4 is waveclear/camp clear except this time all 3 options have their place.

Level 7 sees Barbed Shot as more of a Q build niche for maps such as BoE or maps with bosses. Festering Wounds is there for all builds, since all get bonuses from maximum trait stacks and Lost Soul is great for both AA and W.

All her 13 talents add value to all her builds. Period. Armor reduction, Increased mobility (with more Q’s AND the synergy with Festering Wounds), or more range and applying stacks to two (or more) heroes are all amazing. You literally cannot misclick on this tier (even if Windrunner pulls slightly ahead).

16 is survivability in the forms of sustain, speed or unstoppable. Again, you literally cannot misclick here.

Seeing how Sylvanas is also very much a generalist/hybrid hero with both Burst and Sustained damage, both Physical and Spell damage, I think she is certainly a strong contender for the best talent tree of the game.

EDIT: I chose these three Ranged Assassins because they largely do the same thing (hybrid damage) and can be easily compared to each other. Besides these three, here are other heroes that I think have decent to solid talent trees: Pretty much all tanks other than Garrosh (I’m also not sure about Muradin, Tyrael and Cho as I don’t play them), Cassia, Falstad, Zul’jin, Alexstrasza (with E build being on the weaker side), Rehgar, Uther.

I’m sure there’s more but I either don’t play them or I’d have to double check and look through their talents.

6 Likes
4

good talents tree means you have truly different viable options
Like jaina is a good example, if you want to go frost armor then u pick the sustain at 1, otherwise you can go arcane inte

Zarya is another very good example, very versatile regarding the comps. Same for Blaze

Then there are heroes who really shine with the same build all the time, if you go another build it’s because you want to have fun, enjoy it. Like Stukov

1 Like
5

I disagree on Jaina. Jaina has one build (Fingers>Q>Q>Icy Veins>Root) with SLIGHT deviations in case you need Q Range (vs a hammer maybe) that then forces you into Arcane Intellect or if you need Frost Armor. Both of these mean you give up Pierce so oftentimes you go the standard build anyway.

Zarya doesn’t strike me as a contender but that’s because she herself is very niche so mostly you pick her to Speed Barrier a Garrosh or Arthas.

I agree with Blaze, he has a decent tree. Stukov might too but I pretty much only ever do one build on him. The fact that it isn’t the standard build everyone else takes (and isn’t the E build either) might be a sign of a good tree though.

6

I´ll assume a standart three-talent tier if not specified otherwise. I am going by usefullness and not by pick rate as people have a habit to just stick to what they usually pick, not necessarily what would be a good/usefull pick in the situation.

A good tier should not have an talent that simply outperforms the other options in the vast majority of the cases. Even if it is generalist vs niche, the niche talents should at least be useful in 20% of cases. If all are generalist or all are niche the pick-rates should be roughly equal.

Cross synergy options are good in my opinion, where talents for other abilities are an alternative to just picking all talents for that specific ability. Or more general speaking, a tree that allows for mix and match between talents of different abilities to a good degree but also going mono-pick for one ability.

There should be a form of tension between picks for different abilities, just being able to pick all talents for one ability and then pick for ability two as the former are all 1-7 and the latter all 13-20 is boring. There needs to be some decision making what one wants to build. (Example Ana´s Grenade and Dart talents).

In general every ability and and the trait should have at least two talents on different tiers.

Themed talent tiers are fine by me if each of the options fosters a different playstyle. Let us for example assume Whitemane had this level 1 tier (modified version from my WM rework thread):

Level 1 (Mana Management Tier)

  • Unwavering Faith
    • Moved from Level 4
    • Added Functionality : Additional cost for Desparation stacks reduced to 50.
  • High Inquisitor
    • Moved from Level 4
  • Indulgence
    • Moved from Level 4
  • [New Talent] - Scarlet Wrath
    • Basic Attacks against enemy Heroes increase the duration of Zeal by .75 0.5 seconds [alternative: restore mana depending on stacks of Desparation]

This would be a themed tier for mana managemant, but each of the options plays differently:

  1. general decreased cost, but no refunds
  2. timing with stacks for refunds
  3. stack independent, but requires good skillshot hits for refunds
  4. duration increase to get more out of mana spent, requires hitting basic attacks constantly

I never pick any Q talent – apart from pierce rarely – nor Icy Veins and root only if they are elusive and got a 60%+ winrate in the mess that is QM anyways with her. Jaina has plenty of build options.

2 Likes
7

(For the record, you pulled this card first)

In Master rank Storm League (where I play) and any high level tournament, the build I mentioned is pretty much mandatory. Anything else is subpar and laughed at.

8

I cannot comment on the laughed part, but it doesn’t look subpar?
Based on: Masters (i believe top 1% of players is how HeroesProfile does it), ranked games, last 4 major patches.

ALL of her level 1 talents are competitive by winrate.
Arcane Intellect is competitive at level 4.
ALL of her level 7 talents are competitive by winrate.
ALL of her level 13 talents are competitive by winrate.
Everything at 10 and 16 and 20 looks competitive also.

Frost Armor being the only talent that looks like it sticks out as subpar.

So yeah, it’s popular but this seems to be a perception issue, not an issue of actual power level/performance.

AND if you add in Diamond players (I believe Heroes Profile tracks the next 9-10% of players in this bracket) frost armor rises up and everything else is still competitive.

So It looks a lot more like “that’s what Jaina has to build because that is what everyone things Jaina has to build.” AND NOT “Jaina has to build that way because her other options are not viable.”

4 Likes
9

“Healthy” is pretty much just a buzzword here that has no actual meaning so for the purposes of the thread, I’m gonna pretend what you’re actually saying is “balanced”.

A talent tier can be perfectly balanced despite one talent having a high pickrate. The other talents on that tier can simply be more situational, created for niche uses or be talents made more for beginners.

I assume you’re talking about some talents giving characters more damage or more healing? Those aren’t always the best talents on a given tier. Many times, armor buffs/debuffs and access to CC trump raw healing or damage. If you were talking about winrate, who cares? It’s rare for a single talent to be so good that it boosts up a character’s overall winrate or be so bad it tank’s a characters winrate. This winds up being a statistically insignificant metric.

You don’t have to need to shift a hero’s playstyle to be a balanced talent though.

Frankly, the easiest way to have “balanced” talent trees is for the devs to make sure that, in ACTUAL games, certain talents are actually GOOD. The problem is that the Devs don’t really communicate with Masters/Grandmasters/Pro players, therefore they have to really on their personal skill which tends to be pretty terrible. Ultimately you need communication with many of the best players or need to hire some of them to help create balance that makes sense.

1 Like
10

Frot Armor is amazing with the root against Tracer, Genji and such as you do not need to hit any ability before to get the root out. One needs to put it against the right targets though.

11

I don’t think pick-winrates tell you much in this case. Mostly it’s about the combo you gain when you have all these talents: W+E+Ring+Icy Veins then your Q becomes your basic attack and everyone is instantly dead. Every single one of these talents adds to that, take one out and you lose a lot of value. That’s why people pick Jaina at that level, not to Basic attack or to wait for the third Blizzard to hit cause it has 60% more damage.

12

Well, it is telling you that NOT getting that combo is just as viable from a “did we win” perspective.

It DOES NOT tell the story of, was it harder to win? Did it cause the player to be at a disadvantage? Did it confuse the enemy because they’re used to Q build Jaina?

But it clearly shows that purely in terms of “does not going this build win as much?” Well, the answer is a solid NO. Not going that build does not win less.

And as a matter of fact, it isn’t the most winning talent build on Jaina in Masters or in Masters+Diamond with the original criteria i stated above.

2 Likes
13

Literally what are you talking about? I just clicked the link with your settings and not only do these talents have 80%+ pick rate (except Numbing Blast cause Northern Exposure is also really good)

And all the builds shown below are exactly the build I mentioned, with variations at 16 (between Numbing and Exposure) and at 13 cause sometimes Ice Barrier is better, but mostly people don’t know how good Icy Veins is.

Watch/Ask any high level streamer/tournament and they’ll say the same.

14

by default it sorts by popularity.
Sorting by unique level 1 or level 4 and you get other builds.
Level 7 unique also has a different build.

Unique level 13/16 goes back to Q build, as there are common talents.

And it doesn’t change if you add Diamond players in, or go add in QM.

Probably has to do with things being “meta” as well as they probably having that as the most safe build. And if you assume you’re better than your enemy, all you have to do is ensure that you don’t lose, and eventually you’ll win.

15

I know I am biased, but I think Artanis has one of the best talent trees in the game at the moment. Other than Psionic Synergy at level 4, all of his talents are strong, and there are all sorts of neat interactions between a whole bunch of different talents. There really isn’t a single meta build for him, and even his situational talents are generally useful even outside of their niche.

2 Likes
16

D U B B L E-Y U E B U I L D

17

My favorite talent tree used to be Tyrande, the version where she had lunar/battle momentum. The different ways you could spec her and truly change up the way you played as well as be adaptable to teammates and enemies was amazing. That included a lot of generics too, which some people have this irrational fear of.

Even though her current tree is a step up from whatever the heck that stackathon monstrosity was in the previous build, I do miss her ability to be a more offensive oriented support in double healer comps.

There’s a sweet spot in Blizzard’s history where some of the heroes have talent trees where Blizzard finally started trying to make unviable talents viable, yet not constantly prune them or rework them into forced themed tiers with only three options.

I think Zarya is probably the best example of that. Artanis is kind of like that. I would have said Kharazim too, but he’s been pruned too. There have been some reworks that have been somewhat successful without bashing a theme on us in every tier and not forcing three talents. Lunara for example, but I honestly believe that to make QM a adaptable place, having weird, maybe cooky off builds that can sometimes lend themselves in ranked to be healthy.

18

Mephisto has the best talent tree in the game, he only has one weak talent and it’s placed at 20.
Also Mura’s talent tree is amazing.

19

For me a bad talent tree would be Maiev’s:
Level 4:
More Q damage
More Q damage
More E damage
Level 7:
More Q damage
Reduce armor (for more damage)
E increases your damage
Level 13:
Q grants you armor
W grants you armor
E grants you armor
Level 16:
W deals more damage
Q deals more damage, based on you E
You deal more physical damage

It’s quite absurd - it’s all the same. Damage, damage, damage. Her design is carved in stone and you can’t do anything about it via talents. All her talents do is change the “how”, but not the “what”. In fact, Maiev is perfect example for how maths can determine her whole build, as you just need to pick the most efficient talent… Although it’s a bit hard as most of them have some requirements that can’t be met every single time, and some less effective talents can be a bit more consistent.
She also doesn’t have a single talent that improves anything about her abilities apart from increasing their damage. No range/AoE/radius increase, no talent that allows your E to remain and continue moving after you’ve used blink, giving you 2-nd blink charge for the return trip. No talent to imrpve the duration/cooldown of her trait, or increase her speed by 100-150% while trait is active. She has such a nice potential, but the only talent that can be considered as “improving functionality” would be the physical damage around you when you cast Q. Although by itself it’s also just a damage increase.

For me a good talent tree is one that is flexible and damage talents compete with survivability and utility ones.
For example The Butcher:
Level 1:
Q PvP cdr and mana reduction
Survivability
Q PvE power increase
Level 4:
Q length increase (utility)
E crd (utility)
Q damage increase
Level 7:
W increases your speed (utility)
Dying minions heal you (self-sustain)
E grants you spell armor (survivability)
Level 16:
Q slow lasts longer and doesn’t fade (utility)
falling below 50% HP grants you armor and attack speed (survivability + damage)
basic attacks increase your speed (utility + damage)

While I don’t think that the Butcher is fine as he is, his talent tree is kind a nice and flexible.

1 Like
20

Multiple people have mentioned this and I agree- pick rate alone doesn’t indicate a talent is bad, but it is an indicator in many instances (hence my inclusion of “and/or” in that sentence).

Talent function is also a consideration. Talents that are easier to use or more reliable will see higher pick rates.

Yes. An example is Varian’s Lion Heart. I read a post where someone did that math and showed there’s no way to get more value out of this talent than the other two options. These small math differences can manifest an entire build, as seen in Chen’s Q vs AA-build. The AA-build straight up deals more damage than Q can, and reliably at that. This can be balanced just by doing the math, similar to what is being revealed by Whitemane’s reworked AA-build, but it’s not being done.

In this instance I mean shifting play-style vs the other talents, not the hero itself. When I choose Stormstout Secret Recipe, the way I play Chen is slightly different than Eye of the Tiger. This is rarely an issue, but its worth mentioning.

This is interesting because that looks like a good talent tree to me. And this seems to be the design philosophy of many new Heroes (look at Qhira). People have also been citing Valla as an example of a good talent tree, and she follows this design philosophy. Structuring trees with tier coherency this way makes balancing much easier, but sacrifices creativity. There’s no right/wrong in that regard, just preference. I prefer the tier-coherent structure for most heroes.