[Skill Build Analysis] Ember Spirit

November 27, 2013

With both Ember and Earth Spirit being completely new to many Dota 2 players, I thought they’d be worth looking at how players have been using them.  If you’ve been watching Dotabuff, you saw both heroes starting at extremely low first day win rates (~38% for Ember, 32% for Earth), but both win rates improved significantly over the first week of their release.  Frankly, even though his win rate has been lower, I think Earth Spirit is the much more viable hero of the pair, but Ember’s hero ID comes first so Earth will have to come later in the week.  As usual, here’s a brief rundown of the skills:

EmberSkills

For a more detailed description, here’s the Dota2 page.

Now for the skill builds, I’m going to try something different this week.  First, here are how the primary builds break down.

emberPrimaryUsage

In Normal the playerbase is pretty evenly divided between maxing Sleight of Fist and Flame Guard by 8.  As you move up, Flame Guard clearly becomes the dominant build.  To get a better look into how things are developing, let’s include the Secondaries (basically, the second skill maxed in the build)

emberPrimSecUsage

What we find is that skill priorities are fairly mixed in Normal, but as you move to Very High they make a pretty distinct transition to Flame Guard -> Searing Chains -> Sleight of Fist.  Do win rates support VH players in this decision?

emberWinRates

It appears so.  Discounting the result for Q->Split (a build only featured in 93 games), the Flame Guard -> Searing Chains -> Sleight of Fist hierarchy held up pretty well.

My suspicion is that Sleight of Hand doing so poorly suggests that Ember Spirit works much better as a ganker than a carry.  A complete 6.79 Farm Dependency test will be coming a bit later, but I don’t expect that Ember Spirit’s score will be very high.  At best his carry potential revolves around Sleight, which is comparable to Kunkka’s Tidebringer in that it provides burst potential but no sustained damage.  Shortly after Ember came out, I watched a pro player play him, dominate the game with the highest CS and 6 slot item build including a rapier, a teamwide 2:1 kill advantage, and still lose the game.   What you want from a carry is a hero that can turn that kind of an item advantage into a guaranteed win, and Ember Spirit just isn’t capable at that.

Given that, maybe Flame Guard -> Searing Chains are better because they’re better at being the mid game ganking complement to an actual carry, or maybe they’re just performing better simply because the players going that build aren’t even trying to fulfill the farming carry role.  Honestly it’s probably a bit of both.  Ultimately, I don’t think Ember Spirit’s in a great place right now.  He’s not an amazing solo laner, but he also doesn’t really warrant a protected lane.  Early findings suggest that his overall performance is up in VH, but it’s likely mostly driven by VH players emphasizing ganking over farming.

I looked into his items but didn’t make any graphics because they were really boring.  Basically look at the Dotabuff item page and assume that VH players build the typical VH items (Wand, Drum, BKB, Bottle, etc.) more frequently.

In any case, that does Ember Spirit for now.  Earth Spirit will be coming later in the week.


6.79 Hero Picks and the Rise of Uncertainty

November 27, 2013

Originally published on Team Liquid (Gfx: riptide, Heyoka // Editors: TheEmulator, riptide, Firebolt145)

I’ve received a number of requests to do a hero analysis for the latest Dota patch, and to be honest, I’ve been procrastinating pretty hard. You see, 6.79 is weird. At the pub level everything appears to be behaving as normal, but for the competitive scene 6.79 might be the biggest change for the game since the conversion to Dota 2.

The Pull-Camp Blues

Since at least TI2, Defensive Trilanes (or Defensive Duo with a dedicated Jungler) have been the dominant team composition. Effective use of the pull camp served as a significant source of farm for your supports, and their nearby presence helped to ensure the safety of your primary carry. In more recent patches it became more common to see Aggressive Trilanes designed to disrupt this farming pocket, but an Aggressive Trilane was a risky commitment. If you failed to accomplish much in the laning phase, you could find your trilane supports underleveled and unable to accomplish much in the post-laning phase.

Then 6.79 came out, and with it major changes that throw into question the dominance of the Defensive Trilane. First and foremost among these was the neutral camp swap that made the pull camp an easy camp. This alone greatly reduced the effectiveness of pulling, but on top of that, neutral experience is now also shared among any nearby heroes, which means that an offlaner can safely contest a portion of neutral farm by just standing in the general vicinity of the pull. As a result we’ve seen teams do a number of things with their supports to compensate, such as running 2-1-2, Dual Mid, or building 20 minute Midases.

But these changes don’t just affect supports. For an example let’s take a look at the fortunes of the offlane specialist, Dark Seer.

All competitive stats provided by www.datdota.com and represent the 6.79 patch period as of November 15th

6.79 has seen a massive collapse in Dark Seer’s competitive presence and performance, and before you ask, no it’s not the months of Vacuum cooldown nerfs finally taking their toll. In an environment with entrenched Defensive Trilanes, Dark Seer specialized in being able to find farm from the offlane position in a way that few heroes could match. Even if he didn’t complement your overall lineup and gameplan, he would still be able to use Ion Shell to find the money for a quick Mekansm and an eventual utility pickup like a Scythe of Vyse.

With Defensive Trilanes significantly weakened (and various other changes like the more favorable creep clash point and larger experience area), Dark Seer’s competitive advantage in offlaning is greatly diminished. He’s certainly not a bad hero, provided you’re actually drafting a team that takes advantage of his utility, but he shouldn’t be treated anymore as your one-stop solution to all your offlaning problems. Essentially, you can afford to get a bit more greedy with your offlane selections, so unless you’re specifically building around Vacuum and Surge, Dark Seer isn’t the greatest pickup.

The tendency for risk aversion isn’t exclusive to Dark Seer either; it’s also apparent in the two other offlane survival specialists, Clockwerk and Timbersaw.

Couple caveats. First, while neither win rate is stellar, they’re both doing far better than Dark Seer. Second, neither hero is really that much off from their 6.78 win rate. Clockwerk ended the patch at 46.9% and Timbersaw at 49.0%. Finally, both win rates are still unstable. In the intervening time since I started working on this article, Clockwerk has jumped to 47.2% and Timbersaw has fallen to 45.1%.

Still, both heroes have seen a tremendous amount of picks but haven’t really put up the numbers to justify it, which is a hallmark of an overexposed hero. Neither pickup is bad, provided they fit into your lanes and composition well, but teams are leaning on them too hard (and possibly drafting them too early) as a risk averse pickup. Hero choices in the offlane are more open then ever, so teams shouldn’t feel caged into having to rely exclusively on survival specialists. It’s also worth keeping in mind that the TI3 Champion Alliance ignored Dark Seer almost completely and only used Clockwerk sparingly in situations where Bulldog had already attracted multiple bans.

The Great Compression

The other major development of 6.79 is a whole host of nerfs targeted at many of the dominant heroes of the 6.78 metagame.

Elder Titan and Troll Warlord were not available for the entirety of 6.78
For more information on the 6.79 changes check out http://dev.dota2.com/showthread.php?t=109826

Outside of the inclusion of Treant Protector and the two heroes added to Captain’s Mode late in 6.78 (Elder Titan and Troll Warlord), this list of heroes includes includes 12 out of the 15 heroes with the highest Pick/Ban rate in 6.78. Besides the support trio of Visage, Naga Siren, and Chen, all of the nerfed heroes have also seen substantial reductions in their 6.79 draft relevance. That being said, none of these heroes look completely crippled, with the possible exceptions of Outworld Devourer and Troll Warlord, who are both much less viable in the mid roll than they were in 6.78.

It’s also worth keeping in mind that competitive win rates can be deceptive. We are still early into the 6.79 patch, so the usual warning of “Small Sample Size” applies. But on top of that, a hero’s win rate over such a short time span can be heavily influenced by which teams are deciding to play that hero. For an illustration, let’s take a look at Magnus.

At first glance Magnus looks much improved in 6.79, and maybe the environment does suit him more despite receiving no direct buffs. But when you look at the breakdown of who is playing Magnus, it becomes obvious that a lot of his 6.79 win rate is driven by Alliance being responsible for over 1/3 of his games in the new patch period. Nearly any hero can look good under those circumstances.

While this should give you pause before you accept a win rate at face value, it doesn’t diminish the fact that Magnus is being treated as a priority again, at least in the eyes of Alliance. Too often the community as a whole overreacts to a nerf by completely dismissing a hero, when in actuality that hero is still very much a threat in the right hands. s4 on Magnus is one example of this, but I’d like to draw your attention to the potential latest overreaction.

After a reign of terror in 6.78 that included achieving the 2nd highest ban total of the patch period (behind only Batrider) and a 5-0 sweep in the International 3 Grand Finals, Io received a brutal nerf in 6.79 that led to the hero having the lowest observed public win rate in Dota 2 history. Maybe this was finally the nerf that would put competitive Io down for the count.

Haha, fat chance. While Io’s Pick/Ban rate has collapsed by a historic 70 percentage points, the hero has still managed to put up an impressive 74% win rate, largely on the back of Fnatic’s 7-1 record with the hero in 6.79. For all of the hero’s faults, Relocate is still an oppressive presence in the midgame, and teams not prepared to deal with it can still be caught with their pants down. Fnatic’s play with Io has thus far been dangerous enough that they are the recipient of over 60% of the hero’s total bans in 6.79.

And on a related note, Treant Protector isn’t dead either. He’s fallen pretty far from the must-ban status he enjoyed briefly before the first Living Armor nerfs, but Goblak is back to his old tricks with the hero, including RoX.KIS’s Game 3 Upset of Alliance that eliminated Alliance from the G-League Western Qualifiers.

Pick a Carry. Any Carry.

One consequence of all these nerfs is that 6.79 has an absence of Flavor-of-the-Month picks that you can just mindlessly grab with your first pick like Batrider, Lifestealer, and Io earlier this year. One place where this is particularly clear is in the selection of carry or 1-role heroes.

The most common picks so far have been Lifestealer, Luna, and Weaver, but as we’ve already discussed, Lifestealer and Weaver are a shadow of their former selves. They’re both still viable, but at sub-45% win rates they’re likely being overpicked by teams that haven’t really adapted to the 6.79 changes and are just reflexively grabbing their old staples. Luna has definitely fared the best of the three, but she hasn’t really looked like a must-pick/ban hero. She’s just a solid performer that’s getting more attention in light of her competition being nerfed.

Looking at things from the other end of the spectrum, we see a number of heroes hovering around the 60% win mark, but they’re all relatively unproven with only Gyrocopter and Lone Druid breaking the 60 game mark. In particular we have heroes like Clinkz (who is looking better than ever with his 6.79 buffs), Shadow Fiend, Anti-Mage, and Spectre all doing quite well, but at the same time we know that all of these heroes have huge vulnerabilities during certain phases in the game that haven’t gone away in 6.79.

The basic gist is that, at least for now, 6.79 is a patch with a wide variety of competitive cores to build a team around and no clear frontrunners, and while this could change at any moment, for now there’s a lot of room for a team to build their own set of quirky strategies around the heroes they enjoy. One example of this is Na`Vi’s experimentation with Medusa, a hero that hasn’t seen a ton of play from any other team in this patch period.

Three other things I want to highlight before moving on. First, the fears of an Alchemist nerf in 6.79 appear to have been highly exaggerated. It might be true that the higher levels of his ultimate are now weaker without the bonus HP, but the extra regen on the lower levels appear to be making up for that. His win rate both in pubs and competitive is up, and he should absolutely be taken into consideration when making your carry decisions.

Second, Viper has seen his pick rate rise dramatically in 6.79, but he hasn’t seen much success. I’m not convinced that this trend of Viper mid is working out too well, as he’s just a big floating target for support rotations. If he has the potential to be competitively viable, teams need to rethink how they use him.

Third and finally, Mirana is not on this list for a reason. Moving on…

The Emerging Support Hierarchy

While the 6.79 changes have definitely breathed fresh air into the stagnant pool of offlane picks, the unfortunate side effect is that they appear to have created at least the perception of a strict support hierarchy. Four supports in particular stand out from the rest as being particularly adept at coping with the nerfs to pulling.

*Ranked exclusively among common supports

As patient zero in the XPM arms race, Lich has received a ton of attention in 6.79 drafts since day one and boasts the second highest ban rate in the patch period for all heroes. The 6.79 buff to Sacrifice that allows him to convert friendly creeps into experience has turned him into a hero that no one wants to face. All the attention might be a tad overblown, as Lich still has the same competitive weaknesses that made him a non-factor in previous patches despite his consistent pub success, but for now he remains a top ban priority.

Crystal Maiden has been trending as a strong support since before TI3, but her ability to use Frostbite as a jungling tool has taken extra importance with the need to find replacements for the pull camp. She’s the most picked hero in 6.79, which makes me less critical of her lower win rate among these four supports.

During the first week of the patch, Venomancer’s win rate had dipped below 50%. Since then, however, his performance has skyrocketed, and he now has the second highest win rate among heroes with at least 100 games played. Along the same lines of Crystal Maiden, Venomancer’s Poison Sting buffs have allowed him to function as a semi-jungler.

Finally we have Visage. Of the four he has the least immediate tools for finding farm, but it’s possible that his Familiars make up for this once he hits six. Visage has appeared to be largely unhindered by his 6.79 nerf as long as teams compensate for his early game fragility.

Outside of these four heroes, support options look rather dismal.

 

Abaddon was not available for the entirety of 6.78

With the exception of Earthshaker, representation rates are down across the board. On top of that, win percentages are looking dismal for everyone but Bane and Rubick, who are both holding relatively steady, and Nyx Assassin, who is inexplicably up despite the 6.79 Impale nerf. There are a few lesser used supports that are doing alright (such as Vengeful Spirit at a 50% win rate/10.9% pick rate), but there are no obvious contenders to the top four.

Of course there’s the possibility that an increase in jungling is poaching some of these support slots, but the stats don’t really support this theory.

Collectively, pick rates are down slightly for the big three junglers. Chen and Enchantress are struggling some, lending credence to the theory that the neutral creep XP nerfs in 6.79 have hurt jungling, but Enigma appears to be completely unfazed. His 6.79 performances have been quite good (if a bit sparse), but he’s comparable to Anti-Mage/Spectre/Shadow Fiend in that he has the potential to be shut down hard if teams see the pick coming and react to it.

Push Comes to Shove

Thanks to some indirect buffs, mass Necrobooks have come back in style in 6.79 (a trend with likely more staying power than the mass Midas strategy), but that’s not the only boost push comps saw with the patch. Four of the heroes with the largest win rate boosts in the patch are pushing specialists that were mostly ignored in 6.78, and they’ve all seen increased usage in the latest patch.

Pugna has been the standout performer of the four. He saw very limited use in 6.78 but was the centerpiece in a handful of very convincing victories. With 6.79’s buffs to almost every ability in his arsenal, Pugna is being seen as a real threat, and he now has the 3rd highest bans in the new patch. Pugna’s role of an early pusher is comparable to the success Leshrac saw in the TI2 era. Pugna’s one downside is that he is perceived as needing safe lane farm, which often robs his team of a proper late game contingency plan.

Death Prophet might have taken over from Warlock as the strongest pubstomper that everybody ignores, but she may have also taken over from Warlock as the most successful pub hero that can’t make it work competitively. Her record overall has been pretty dismal, but if you look at the individual games there are a number of games that Death Prophet has lost despite putting up what appears to have been a solid performance. Perhaps she lacks closing power in a competitive environment, but my suspicion is that she’s better than her statline would indicate.

Unlike the rest of this list, Shadow Shaman has been quiet and reliable. He’s comparatively flexible when it comes to his role, and has seen a decent amount of success as a support. It’s unlikely he’ll become a must-pick hero, but he’s a great pick-up if you want to shore up your CC and objective control.

Finally we have Lycan, who like Death Prophet, has seen a good amount of pub success mixed in with some abysmal competitive results. Of the four, he’s the hero I’m most skeptical about, but I wouldn’t write him off completely just yet. It could just be a matter of finding the right sort of line-up or lane setup in order to get him to where he wants to be in the midgame. For now though, I don’t expect too many teams to be eager to try him out in a high stakes scenario.

Conclusions, My Top 5, and Some Other Noteworthy Heroes That Just Didn’t Fit Anywhere Else

So that does it. 6.79 is a crazy patch that throws old laning strategies into question and flattens the hero hierarchy. For now, winning is much less focused on grabbing Flavour-of-the-Month picks, allowing teams the room to experiment and innovate. Now with the nuance taken care of, let’s give you an arbitrary top 5 list based on questionable criteria because this is the internet and we all know what you’re here for.

#5 — Gyrocopter
I really only had a top 4, so this pick might be a bit of a reach. Still, even though Gyrocopter has been relatively ignored this patch, at least compared to the attention he’s used to, there’s no reason to believe he’s fallen off. Maybe his 65.5% win rate in 6.79 is a low sample size fluke, but Gyrocopter has been at a reliable win rate over 1,500 games going all the way back to 6.77. I also have more faith in Gyrocopter as a risk-averse carry pick in high profile games than, say, Dragon Knight who has a similar overall 6.79 win rate. During TI3, Gyrocopter put up a 55.6% win rate compared to 43.9% for Dragon Knight. Ok sure, this was largely on the back of Alliance going 8-0 with the hero. Whatever, moving on.

#4 — Pugna
Creepy skeleton baby is a high commitment pick, but he’s a pushing monster that can put a game away quickly if the other team isn’t prepared for it. He might start to fall off once teams learn how to deal with him, but for now the 6.79 buffs have put Pugna in a very dangerous place.

#3 — Venomancer
During the short stretch of 6.79 in October, Venomancer’s win rate was a not especially noteworthy 49.2%, but in November he’s put up an impressive 62.6%. He’s definitely one of the strongest supports of the patch, and to be honest, I think he’ll end up comparing very favorably to the much more banned Lich.

#2 — Bounty Hunter
Whereas Dark Seer has struggled in 6.79, Bounty Hunter has done the exact opposite. The offlane changes have suited him well as he’s a much less safe laner than the Dark Seer/Clockwerk/Timbersaw trio, and he’s perfect for leeching from the enemies’ jungling. Bounty Hunter currently has the highest win rate of any hero with at least 100 games played, and it’s unlikely that he’ll be falling off any time soon.

#1 — Elder Titan
Despite being nerfed in 6.79, Elder Titan is the most banned hero in the patch. He’s not as strong mid anymore, but he’s still a versatile laner that amplifies his team’s output dramatically. Expect him to continue to eat a lot of bans from teams that have this weird aversion to suddenly blowing up.

Finally, there are a number of heroes on the rise in 6.79 that just didn’t fit into any of the above sections, so here’s a quick list on who to keep an eye out for in the near future.

Click here for a more clear image.


6.79 Hero Analysis [Link]

November 20, 2013

Is now available on Team Liquid.  This one might be tough to archive properly due to some table-breaking tables, so I advise checking it out there for an easier read.


[Skill Build Analysis] Venomancer

November 18, 2013

Venomancer has seen one of the biggest improvements in public win rate in 6.79 and is quickly becoming one of the most dominant supports in competitive Dota with a 58% win rate and the 5th highest pick rate.  Given all the attention he’s receiving, I felt he was an ideal subject for the latest SBA, especially considering how much his skills have changed in the last year.  Between 6.74 and now, the damage on level 4 Poison Sting has increased from 140 to 300, and the scaling on Venomous Gale has changed from 100/170/270/350 to 25/200/375/550.  Those are pretty big damage swings, so how have players adapted to the changes?

First, here’s a quick rundown of the abilities we’ll be examining:

venoSkills

The 6.79 changes:

– Poison Nova no longer ignores invis/fogged units
– Venomous Gale no longer ignores invis units
– Venomous Gale damage over time rebalanced from 15/30/45/60 to 0/30/60/90
– Poison Sting duration increased from 6/8/10/12 to 6/9/12/15
– Poison Sting dps no longer dispels healing or disables dagger
– Plague Wards now have Poison Sting for 50% of the damage at the current skilled level (When both Venomancer and a Plague Ward affect a target, only the highest dps is applied)
– Plague Wards XP bounty increased from 12/12/25/25 to 20/25/30/35

And finally, as a reminder, Skill Build Analysis generalizes hero builds into a primary and secondary skill.  For Venomancer, a Q->W build would be a build that maxes Venomous Gale first (like 4/1/1/1 by 7) and then maxes Poison Sting next (So 4/4/1/2 by 11).

When it comes to builds, Veno players are all over the place:

venoUsage679

There’s no clear consensus at any skill level, though Poisonous Gale -> Plague Wards and Plague Wards -> Poisonous Gale are the least popular builds at all skill levels.  Also interesting is that the overall distribution hasn’t changed that much compared to 6.78 with the largest gain going to Ward builds in VH despite Ward being the only skill that didn’t receive stronger scaling in the patch.

venoUsage678

So after all the changes, what is Veno’s strongest skill?  Well, all indicators point to Poison Sting.

venoWin679

Note all the top performing builds in VH.  Poison Sting (W) is the strongest primary by over 3%, and in the other two trees the Poison Sting is the dominant secondary (Q->W and E->W).  (And for what it’s worth, this appears to have also been the case in 6.78

venoWin678

Based off this, I suspect that the strongest Veno build is 1 point in Gale early for gank attempts, then max Poisonous Sting followed by Plague Ward (such as 1/4/1/1 by 7, then 1/4/4/2).  This build has the highest overall win rate in both High and Very High, and it gives Venomancer a lot of neutral killing potential in an environment where supports are often starved of XP due to the 6.79 nerf to the pull camp.  Venomous Gale builds aren’t bad though and can likely sync well with certain playstyles, but Gale builds should still max Sting relatively early.

I guess that’s pretty brief, but there’s not much more to say.  Hopefully you all find it satisfying enough after the hiatus.  Keep a lookout for something big coming later this week, and maybe something on Earth and Ember Spirit soon, API permitting.