Late Game Heroes, Early Game Heroes, and the Horseshoe Heroes.
“If we understand our comp and the enemy team’s comp, we cannot lose.” - Dignitas Bakery, 2017 HGC Western Clash
This will be the first of a series of articles here at Gale Force eSports that discuss different ways to think about drafting in Heroes of the Storm. I know people reading this are going to have questions, because talking about one piece of a larger picture always raises questions, but rest assured whatever they are they will eventually be answered in a future article. Alternatively, you’re welcome to drop by my stream, www.twitch.tv/cavalierguest, and ask me directly when I’m live.
We’ve all heard casters mention “This composition is very late game” or “If they don’t win the early game they won’t be able to come back”. It’s something that gets talked about quite a bit in the competitive scene but which heroes are which? Which heroes are late game? Which are early game? Why? What are the criteria for a late vs. early game hero?
Today we’re going to go over every hero and explain where they fit in this dynamic and why. The first thing we need to remember is that the objective of the game is to kill the enemy team’s core. Whether you do that just by straight pushing and killing buildings, clever rotations for map pressure, objective control, or constantly killing the enemy team so they can’t defend against minion waves, you’re still doing it to kill buildings.
Imagine a scenario where you had a hero with a level 20 talent that read “Your Q now instantly kills buildings, including core.” Would you say that if you get to 20, you basically auto-win the game? Would the enemy team need to make sure they ended the game before 20? This is frequently a dynamic that arises in every game be it Hero League, Unranked Draft, Quick Match, even ARAM, but players don’t realize it. If you are able to recognize when you have an early game composition vs. a late game composition, it will affect everything about how you play that match. When you know you’re a late game comp you just want to stay as even in levels and structures as possible until your team’s power spike. Keep every lane soaked, prevent pushes, and die as little as possible. Even if you lose objectives, so long as you’re continually soaking, you won’t ever be more than two levels behind at most. On the other hand, if you are an early game composition you need to exploit that whether that means securing kills, objectives, camps, buildings, or whatever else.
The first category is early game heroes. An early game hero allows you to kill buildings, kill heroes, or control objectives in the early game. Frequently this means snowballing: you use your strong early game to get a significant level advantage, ideally a full talent tier, so that the relative weakness late game does not matter and then you tempo out the game. This means that every death you allow once you have that advantage is massive, far more so then it would be for a different type of composition, because of the comeback experience mechanic. The absolute worst thing you can do is play to aggressively when you have an advantage as an early game composition. Just slowly strangle out the win with your level advantage.
The second category is late game heroes, which as we’ll see has the most diversity in why they spike so strongly. The criteria for a late game hero is one that wins the game - allows you to kill buildings and the core - in the late game. They can be further defined by the negative quality that a late game hero is difficult or even impossible to make up a deficit with, meaning if you are 3 or more levels behind you are unlikely to ever win. This can be because of talents, scaling, quests, relative power levels, healing throughput, all sorts of things. But the point remains: getting behind with late game heroes is just as much of an auto-loss as getting ahead with them is an auto-win. This should impact your decision making during games.
The third category of heroes is one I think everyone kind of thinks all heroes fall into, no huge spikes or drop offs. Heroes who simply scale normally and have reasonable talent power all the way up. Obviously as the game progresses they get stronger, but they don’t jump up, or fall down, relative to their power at level 1. This means these heroes are always solid, but it also means they will likely lose to late game heroes unless they have one on their team as well, or are paired with early game heroes to snowball.
The fourth category is particularly interesting to talk about. Horseshoe heroes: heroes who have a strong early and late game, but a very poor mid-game. Frequently this is because they have a strong base kit that scales poorly and is fixed by a talent at the 16 or 20 mark that smooths out their power curve. Horseshoe heroes are volatile in HL, UD, and QM. It can seem like you are just steamrolling the opponent, then suddenly you can’t seem to win anything, and then at the end of the game you win one fight and suddenly end the game. Many people attribute this to people losing focus because the game was “so easy.” In fact in almost all cases where this happens, your composition had a horseshoe hero. That knowledge should let you know when to play aggressive and when to play conservatively as the match progresses.
The last category is debateable if it even exists in the average game: the Reverse Horseshoe. That is, a hero with weak early and late game, but terrifyingly strong mid-game. The only real candidate for this category is Varian. However, since Blizzard has promised us more heroes with the multi-class design, I think there may be more in the future, so defining the term now can’t really hurt.
Early Game Heroes:
Auriel: With the caveat that Auriel always needs to be paired with the right Hope generator (battery), manaless healing in the early game puts you in a position to be very aggressive. Paired with something like Sgt. Hammer or Gul’dan, you put on amazing amounts of pressure. There is an argument to call Auriel a horseshoe hero because of Reservoir of Hope, but it isn’t always the right talent choice. Certain pairings, like Cho’Gall and Lunara, really favor it however because of their scaling.
Chen: Drafted correctly, Chen will absolutely dominate a lane, which helps give his entire team a small advantage. He can turn this into an early 10 and then use either of his heroics to get a pick, which his team can then push with because he can drink tower damage, snowballing a lead. If he fails to do this, he drops off very quickly relative to the power of other heroes.
Dehaka: Another lane bully, you’ll notice this is a trend with early game heroes, but this one has a fun advantage of a global and a stun. He can be bullying his lane, global for a pick, walk back to his lane and keep bullying it. When people start grouping in the mid game his power drops considerably. In very long games Dehaka actually does become a horseshoe hero, because of his ability to safely split push with essentially three health bars and two escapes while also having the ability to be at any fight almost instantly.
Ragnaros: Ragnaros essentially has a heroic pre-10 in the form of Molten Core. Having one heroic to zero is much better than having two heroics to one. His wave clear and sustain are strong all game long, but the potential ways to abuse Molten Core in the early game are what really allow Ragnaros to just smash teams. It is worth noting that because of Molten Core Ragnaros is hands down the best hero to carry an otherwise late game composition to their power spikes.
Sonya: Another hero who doesn’t use mana for the early game, Sonya’s ability to sustain off the map and do camps gives her infinite lane pressure. While it is true she benefits a lot from her 16 and particularly her 20 talent to survive in late game teamfights, you know what helps even more? Being multiple levels ahead in scaling, which is something she can do just by pushing and killing buildings with her sustain.
Sylvanas: A composition with Sylvanas fundamentally does not have to play the early game the same way other teams do. You can always push forever, forcing the enemy team to react to you. A pick is a tower. A second pick is the wall. A third is half the fort and the fountain, which makes the fourth pick even more likely. Sylvanas, even on a map without a strong objective, is the single hardest snowballing hero in the game. Protect her, kill people, and abuse it.
Xul: Wave clear enables rotations. Rotations allow for picks. Xul has the best rotational wave clear in the game. Either you are behind in the rotation and you get free tower siege and ammo draining from skeletons, or you are ahead in the rotation and therefore your team always has more time to position for picks, which Xul contributes to with his point and click root. With the correct talents, Xul can sustain in lane literally forever, never having to back. The only reason he drops off is because the lanes get so pushed in that he can’t just push waves all the way to core with his limited mobility. In rare scenarios where both teams are deadlocked, Xul can tilt the balance of a Catapult vs Catapult lane very slowly and sometimes force the enemy team to fight or lose which gives him a minor in horseshoe territory, but we’re talking about the 29-30 minute mark on a small subset of maps. Very unusual.
Zarya: Zarya is straight up one of the strongest heroes in the entire game, though deceptively difficult to play. I categorize her as an early game hero not because she isn’t stupidly strong the entire game, but because she is in fact so much stronger in the early game than almost any other hero that her relative power curve has nowhere to go but down. She is second only to Sylvanas in snowballing a small lead into multiple structures and thus an even bigger lead, but she doesn’t use mana and does a lot more meaningful damage when played well.
Late Game Heroes:
Scaling Advantage Late Game Heroes:
These are heroes whose base stats, be it health, damage, or particular abilities, simply scale better. This means that they are strong any time you are ahead in levels, but also stronger in the late game even when you are even.
Cho’gall: Health scales at 4.5%, Gall’s Q and W both scale at 5%, and Twisting Nether scales at 5%. Cho’gall being so strong when you’re ahead in levels was literally the reason Blizzard decided to implement the scaling changes in 2015, because it was so oppressive in internal testing. From playtesting on PTR I don’t believe this dynamic has really changed, but I will update this once I’ve had a chance to evaluate him more on live.
Lunara: Lunara’s trait damage scales at 5% and as the game goes on she acquires tools to spread her trait to potentially the entire enemy team.
Probius: As much as we all wanted a builder hero, Probius is a combo burst mage. His combo scales at 5%, both Disruption Pulse and Warp Rift. Which means the later the game goes, the more likely it is he can simply one shot someone.
Tracer: Pulse Bomb scales at 5.5%. In the late game, especially with a level lead and Quantum Spike, Tracer can essentially kill any other hero with her full combo. Just due to death timers, these picks are much more impactful the later the game has gone and she has so many tools to secure kills.
Death Timer Heroes:
Murky: Some of his builds are gated by quests, but the huge value of Murky is being as annoying as possible with very little punishment. The longer death timers are, the more worthwhile it becomes to trade a Murky death, or two, or three, for a kill on the opposing team. His split lane pressure is a slow build, as opposed to destroying buildings in the first few minutes, so it takes a while for that to open up lanes as well.
Tyrande: Tyrande is a hero who is extremely good at enabling picks… and awful at doing anything to get value out of them. So you have to wait until the death timers are long enough that you can use the value of those picks. Sentinel also increases in value in the late game as scouting becomes more and more important to winning and denying the enemy team’s play, especially as the lack of structures or different amounts of push in lanes can really restrict your options for seeing where the enemy team is on the map.
Talent Advantage Late Game Heroes:
These are heroes who need a specific talent in order to completely unlock their power. Sometimes it is a heroic, but frequently it is a quest talent that is hard to finish or a 13, 16, or sometimes even a level 20 talent. Unless you are snowballing a lead hard in the early game, don’t expect to end a game until you hit the power spike these heroes need.
Azmodan, Taste for Blood: You really can’t participate in teamfights until ~250-350 stacks. Ideally you don’t want to fight until you’re finished stacking but it rarely works out that way.
Falstad: Having three uncapped quests at level 1 means that Falstad only gets stronger as the game goes on. He does have strong spikes at 7, 16, and 20 as well but it is the slow build nature of Falstad’s level 1 talents that will eventually allow him to force a win. Note that a hero with both wave clear and a global is his or her own ticket for getting to late game. There is a reason Nostromia played Falstad every game possible from Bronze 5 to Masters.
Kharazim, Insight: Until Kharazim finishes Insight he basically isn’t a healer, which is interesting, because Iron Fists Kharazim has the same overall healing throughput but his damage makes fight shorter. Shorter fights need less healing. Ideally your team understands this and allows you to wail away on things without getting in too much trouble.
The Butcher: The Butcher is essentially not a real hero until his trait quest is finished. He has a stun to help out with CC chains and is otherwise comparable to a full minion wave in his damage (minor exaggeration, 118 DPS vs 160 DPS). To compensate, he is close to being 1.5 heroes when his quest is finished.
Abathur: The need for Ultimate Evolution to meaningfully contribute to teamfights cannot be understated. Everything else Abathur does, the hat, the vision utility, the soaking, are tools to keep his team even until 10 and to control the map post-10, but Ultimate Evolution is the key to unlock the puzzle of how to allow your team to be winning so they can exploit that map dominance.
Gazlowe: With Robo-Goblin, Gazlowe has basically the third highest sustained AA damage in the game, with 4 seconds of move speed and armor on demand. At level 20 with Mecha-Lord it is more than a stacked Butcher, though not considering talents Butcher may take. Alternatively, with a good setup Grav-o-Bomb should just win a teamfight. He used to be a horseshoe hero to a degree, but with the PTR changes I’m placing him squarely in the good at 10, gets game winning at 20 category.
Johanna: Johanna’s major weakness as a tank is the lack of a strong engage tool. Both her heroics work as engage tools, so she can finally start tanking instead of clearing waves. From this point onward she just gets progressively stronger.
Tassadar: Archon is essentially required to be relevant in teamfights, outside of specific compositions that are just abusing shields, which is how we most commonly see him in organized play. When your teamfight is tied to a long cooldown you’re not going to be able to take advantage of opportunities as much as other compositions, so you get into a rhythm post-10 of fight at objective, clear waves until next objective. Post-20 with Twilight Archon you can finally just end. All of his level 1 talents being relatively long completion quests doesn’t help either.
Thrall: Thrall isn’t really an assassin until 7, with Follow Through he finally has real kill potential, but he isn’t a playmaker until he picks up either Sundering or Earthquake. Both fix the small deficiencies he has as a bruiser in different ways, depending on if he needs to play more as a flanker or an anti-flanker.
Uther: Uther depends heavily on talents to really enable his team: Divine Shield, Shrink Ray, almost any of his 16 talents, even Gathering Radiance if you abuse the bug. The key to Divine Shield is to always use it aggressively and almost never use it on yourself. Even Divine Storm’s potential to be an area stun is quite a playmaking ability.
Li Li, Bruiser Build: The combination of Gale Force and Mass Vortex is a ton of spread damage but it is Surging Winds that is the payoff. The ability power increase makes every ability, including her heroics, do more damage or healing. The timing can be tricky but this is the level where you should start winning fights that were otherwise close.
Muradin: While he has strong pick potential all game long, Healing Static at 13, along with Avatar at 10, fundamentally changes Muradin’s game plan. E goes from standing for Escape to Engage. The armor plus the healing means Muradin can just start cracking skulls and live. If his team follows up his aggression, he should just win the fight. He continues to have strong spikes at both 16 and 20, but this is where he comes into his own.
The Lost Vikings: There are many misconceptions about TLV as a win condition, like “they are great pushers.” No, they soak. Safely. Even it if means being behind towers while they die to the enemy team. Yes, a strong 4-man can get an early game tempo advantage with TLV by trying to bully the enemy team without missing soak, but wouldn’t a hero with just really good wave clear have most of the same effect as TLV in that case on almost every map? The advantage of TLV is their ability to push in the late game, when being alone becomes infinitely more dangerous because of death timers. You can’t harass TLV at this stage without bringing a group unless you know where the 4-man is. But TLV has Jump!, they can juke, and jive, and waste multiple heroes’ time for a third of theirs. This trend continues all the way to 20 when Fury of the Storm can be picked up and TLV can actually push every lane, forcing constant response from the enemy team lest they lose keeps and core to the Nordic annoyances.
Chromie: Poke and patience are the hallmarks of many late game heroes. Knowing you can secure a kill if you just wait for Cleanse to be used while you poke and be annoying. Chrome is particularly interesting because the most important power spikes in the game (barring Heroics) are 16 and 20 for almost every hero, but she gets them at 14 and 18, which sometimes allows for tempo wins off the back of these unusual spikes. In particular Shifting Sands is a huge shift in her threat level, potentially increasing her poke by 40%.
Kharazim, Iron Fists and Transcendence: Regardless of which one you go, either Cleansing Touch or Way of the Hundred Fists are important capstone talents. Even in the rare situation where Echo of Heaven is the right choice, it has a dramatic impact. Kharazim becomes the best anti-CC support in the game at this tier, or has incredible chase and burst potential, or has some of the best sustained healing.
Li Li, Serpent Build with Serpent Sidekick: Doubling the effectiveness of a basic ability and doubling the advantage of every talent you took for it is very large. Obviously Kung-Fu Hustle is big at 20, but this is the spike you’re looking for.
Medivh: The power of Reabsorption really can’t be overstated. MVP Black played L5 on Towers of Doom in the Super League Season 3 final. They lost every fight before 16 and won every fight after 16. A feat MVP Black repeated against Fnatic at Blizzcon, the only game they won in the 3:1 set. Master’s Touch being finished is also a factor, but that huge power spike at 16 is the real terror of a late game Medivh.
Nova: Despite Nova’s damage, heroes who secure kills but don’t siege or have wave clear just don’t do enough until death timers get longer. In Nova’s case she also doesn’t guarantee kills until 16, with Crippling Shot or Lethal Decoy depending on build, allowing her to start to be a much bigger threat to the backline.
Raynor: Either Bullseye or Executioner are the big spikes here that really round out Raynor’s kit. Giant Killer at 13 and potentially finishing Seasoned Marksmen, or in a good game continuing to stack it past the reward, are also available and help Raynor start to feel like the carry that heroes like Valla are all game long.
Samuro: Absolutely awful in the early game, mostly good for doing camps. At 16 he can finally participate meaningfully in teamfights, with Press the Attack and Harsh winds being particularly nice boosts. On large maps you’ll see top players frequently play him as split pusher who can also do camps, especially with Illusion Master, but lane pressure like that is still a late game win condition.
Stitches: Fishing Hook is the primary culprit here, but Pulverize and even Shish Kabob are also strong in the right circumstances. This is the single biggest tier for Stitches’ playmaking potential and it comes at a time when a single death at the wrong time can often snowball a win. In a more competitive environment with correct follow up Stitches would be closer to an all game hero and is one of the few where coordination can fundamentally change his category.
Valeera: Valeera’s ability to act as a “finisher” and really secure kills in fights spikes quite a lot at 16 with Expose Armor with another boost at 20 in the form of Cold Blood. Until then she can mostly finish off already weakened targets but can’t really just kill a backline squishy hero.
Zeratul: There is a strong argument for making this level 20 for Rewind, but Sentenced to Death is a huge part of the reason Rewind is so incredibly nuts. This is the talent tier when you can finally just delete someone who has taken any kind of poke damage. Void Prison is also very strong of course but its primary function is to enable another hero, this tier is where Zeratul becomes a personal terror for the enemy team on his own.
Leoric: Formerly one of the strongest early game heroes, Leoric shines at 20 with Spectral Leech. There is a strong argument for putting this at 16 as the combination of Crushing Hope at 13 and Unyielding Despair at 16 has very much the same dynamic, intense sustain paired with intense damage, but Spectral Leech is territory where Leoric can legitimately just win a fight and level 20 is late enough that you can then just end the game afterwards. Fun-fact: Old Cloud9 once ran Medic+Leoric as their late game win condition, Stim Drone+Spectral Leech = gg game. The shortened death timer and vision while dead contribute to his late game, as well.
Nazeebo: Vile Infection. Tripling the damage of your trait and allowing it to apply to heroes give you some of the highest sustained damage in the game tied to very low cooldown abilities. A team that gets to 20 with a Nazeebo just basically should never lose. The major exception here is when poke is abnormally valuable, such as on BoE against a team that is committed to Immortal defense, or against certain healers. Those situations are fairly rare however.
All Game Heroes:
Anub’arak: While every talent in his tree is strong, they essentially help him maintain parity. His base kit is simply one of the strongest in the game and everything he picks up merely accentuates it so he doesn’t drop off at all.
Artanis: While huge playmaking potential is opened up at 13 in particular with the right talents, Artanis’ base numbers and synergies in his kit are really all he needs. His talents mostly maintain parity, with very few opening up new options for him.
Diablo: Goes to show how important new talents can be, Diablo used to be the definitive early game tank but post re-work his power curve has been smoothed out considerably and his late game drop off no longer exists.
E.T.C.: Strong base kit,, strong talents that help him maintain parity. Prog Rock is arguably a huge spike when finished but as it isn’t always the right choice and it is the only thing that really qualifies, not quite enough.
Gul’dan: Despite his major builds all having an important capstone at 16, Gul’dan’s constant self-sustain, poke, and waveclear put him in a position to be valuable all game long.
Jaina: Her combo is fairly consistent in its kill potential across all levels, the only real drop off she has is the sheer mobility many heroes pick up and her lack of defensive options make her very slightly worse in the late game because she can just die, but in many ways that is a positioning issue and not a hero issue.
Kael’thas: Similar to Gul’dan, despite the massive power spike he gets at 20, Kael’thas is just strong all game long. He isn’t quite as weak as Jaina despite the same mobility issue because he has a stun, though his focused damage is lower.
Li-Ming: The trend of mages always being solid all around heroes continues. It is worth noting that Li-ming’s Arcane Orb scales at 3%, as does Calamity, her level 7 talent. Full Orb build is her most popular build but has a below-average win rate for her, and it is precisely because it turns an all around hero into a hero who drops off late game but doesn’t get their capstone talent until 16. Both Wave of Force and Disintegrate scale at 5%, which helps her maintain parity along with her strong damage talents.
Lúcio: While Lúcio has strong talents and talent synergies at 16 and 20, they enhance something he is already doing fairly well. His major contribution, sustained healing and movement speed, are the same throughout the entire game.
Malfurion: Fairly generic talents across the board, though the mana sustain at 10 from Twilight Dream is very nice. Overall Malf’s base kit is his primary strength and it never drops off.
Rehgar: Strong base kit, two strong heroics, reasonable talents including a wave clear talent at 1. Rehgar is just solid all the way up.
Tychus: With most of his value coming from his trait. Tychus stays strong consistently throughout the game. He will occasionally have a brief spike right after 10 just due to the siege power of Commandeer Odin that can lead to tempo wins, but for the most part he is a grinder. Slow and steady.
Valla: Sustained damage, wave clear, burst damage, mobility, very fragile. None of that changes at any point in Valla’s career, it is pretty much the same dynamic all the way up.
Almost every horseshoe hero has a strong basic kit that scales poorly relative to the power level of other heroes that is fixed by a late game talent, almost always a 16 or 20 talent.
Scaling Advantage Horseshoe Heroes:
Greymane: A brief spike at 10 increases Greymane’s kill pressure considerably and his sustained damage early game frequently allows for a tempo advantage on structures and lane pressure with camps, but the later the game goes his scaling of 4.5% on his health slowly makes him harder and harder to kill, allowing for better trading especially in poke wars with his Cocktail build. He drops off in the mid game because heroes start to have answers to his mobility but the harder he becomes to kill the less that matters.
Kerrigan: Her early game burst relative to health pools and defensive abilities allows her to just kill people and survive the counter initiation because of her shields. Her shields actually drop off substantially relative to other heroes’ damage, however, and the frequency of fights not around minion waves also hurt both her mobility and sustain. Her 16 tier however basically fixes her shield issue with either Essence for Essence or Aggressive Defense, bringing her back to parity.
Alarak: Alarak’s play making potential goes down pretty dramatically in full teamfights and none of his talents really boost his kit in an impactful way, but his base kit is one of the strongest and best designed in the game. That said, having two heroics that actually synergize amazingly well with each other gives you the ability to instantly win the game by dominating one late game teamfight.
Arthas: His sustain and pick potential in the early game puts Arthas in a position to simply dominate a lane. In full teamfights he is in a rough spot basically until 20, when literally any of his level 20 talents put him in a position to just win teamfights with just a modicum of help from his team.
Azmodan, Not TFB: There are some interesting dynamics with every other Azmo build that is not Taste for Blood. Sieging Wrath does decent damage while allowing for consistent wave clear and lane pressure. Laser does reasonable damage but your wave clear is very lackluster. In either case Demonic Invasion can be picked up to backdoor keeps, in which case your teamfight presence is largely nil but you can still win the game. Generally if you don’t get a tempo advantage with Azmodan OR don’t get a Keep in the first two casts of Demonic Invasion, you’re behind with no tools to catch up, but his 20 talents are actually so strong they can win games. Forced Recruitment has lane pressure, Fifth Circle has a massive slow, Perishing Flame is both a good teamfight heroic and allows for solo back doors of core with exceptional ease.
Brightwing: If Brightwing can’t tempo out a lead with her global, she is highly dependent on her 13, 16, and 20 talents. Post-20 in particular a lot of her talents fix a basic issue with her lack of burst support. The ability to push out a lane and be in the teamfight instantly also makes it very difficult to close out a game against her, giving her team more opportunities to take the win.
Illidan: His dueling potential in the very early game and ability to clear waves, siege, and solo camps puts him in a position to get every possible advantage on a map. There is a cool dynamic here where if he goes The Hunt he can frequently turn his early game strengths into a pure tempo advantage and he actually belongs more in the early game category, but if he goes Metamorphosis then he drops off slightly post-10 until his 16 and 20 power spikes.
Lt. Morales: The sheer healing throughput in the early game is so ridiculous that a team supported by Morales can go absolutely ham and simply out-trade the entire early game. In the mid game she drops off slightly due to mana issues, but they get fixed with Caduceus Reactor 2.0 at 20.
Rexxar: Bullies lane, gets tempo, wins or you suffer a slow and painful decline in overall effectiveness. Unless you get to the very late game and the power of a stun with a five second cooldown and a free ward start to show their real value, especially with Animal Husbandry ticking away, though even his globe quest is a late game spike when finished and all of his level 20 talents are very good.
Sgt. Hammer: The ability to siege a lane and get immediate building damage just means your team needs to protect you. It forces the enemy team to respond, much like a Sylvanas or a Zarya. This allows for an early game advantage. Unfortunately you tend to lose it because of the power of objectives and the difficulty of rotating without a mount, but in the late game with your 13, 16, and particularly 20 talents you become a teamfighting monster that can just win games and is also one of the most effective “We’re behind, just back door core” heroes in the entire game. All this before even considering Orbital BFG cheese to always auto-push a lane.
Tyrael: Tyrael has just enough damage with Purge Evil to be one of the best lane bullies against fragile heroes. His mobility, especially post-7 with Reciprocate, puts him in a position to kill low mobility heroes if he is the bruiser or off-tank in a composition. You can use this tempo to get 10 first and make very aggressive plays with Sanctification to get even further ahead. Sadly the removal of Imposing Will causes him to drop off considerably if you aren’t ahead and Sanctification isn’t up until 16 and 20 are hit, where he gains new tools to make plays with.
Zagara: In the right match ups she will absolutely dominate her lane, easily killing buildings from off-screen with Volatile Acid. Unfortunately Bile Drop and the resulting Roaches scale at 3% and Hunter Killer’s health scales at 3.5%, making it very easy to kill with random area damage in the late game. Nydus is the power spike that fixes her late game, giving your team a strong late game win condition, especially with Endless Creep for insane vision granting.
Zul’jin: The closest to being a late game hero of any of the horseshoe heroes, because of his uncapped quest at 1 and the power of Amani Resilience at 20, but his strong sustain and dueling potential gives him good early game as well. He drops off due to mobility issues in the mid game. Note this means that if you fix his mobility issue, say with a Tyrael or Lúcio, he becomes an all game hero.
Varian, All Builds: With the removal of the stun on Warbringer, his early game is the weakest of almost any hero. He becomes a real hero at 10 and arguably has the largest spike at 10 of any hero. The problem is his health scales at 3%. While this is less of a drop off for Taunt Varian because of armor, both Colossus Smash and Twin Blades Varian in the very late game, around level 24, start to die extremely easily. The effect is mitigated by how good Shield Wall is, especially for full Parry builds, but Second Wind based builds drop off considerably more.