This isn’t just a “how should I play this hero?” guide, it is both that and a “how does this hero work?” guide. The goal here is to actually understand the hero. If you have a solid enough understanding of the hero, then it’s not unlikely that you would come to the same conclusions regarding skill build and item choices on your own.

Skills and skill build


Shield is your main survivability tool and the ability that defines the hero. It’s not a mechanically simple ability, so I’ll go into detail.

How it actually works

Assuming you have enough mana to cover the incoming damage, this is how the reduction works:

  1. You take X damage, before any reductions or amplifications.
  2. 0.6X is sent to spell shield.
  3. For every point of damage sent to spell shield, you lose 1/{1.6, 1.9, 2.2, 2.5} points of mana depending on your level of spell shield.
  4. In other words, for X incoming damage, you lose {0.375X, 0.316X, 0.273X, 0.240X} points of mana.
  5. The remaining 0.4X is taken normally. It will be reduced or amplified by your armor, magic resistance, or any other reduction, as well as amplified by mask of madness etc.

From these facts, we can infer that:

  1. Mana shield can potentially absorb the same proportion of damage at any level, but does so at the cost of far more mana at lower levels. Thus, your overall effective health points (EHP) is significantly lower with lower levels of spell shield, not to mention tanking damage is very likely to leave you without mana to cast abilities. Therefore, it’s a spell that you want to have maxed quickly, and you want to use it slightly more sparingly before it’s maxed.
  2. Given that you have enough mana, mana shield is a 60% reduction, which equates to a 2.5 EHP multiplier against all types of damage.
  3. Conventional reductions do not reduce the amount of mana consumed to absorb a certain amount of incoming damage, since they happen entirely after mana shield. The only reductions that synergize with mana shield are those that entirely avoid the damage, such as evasion or magic immunity.
  4. Amplifications work the same way: they won’t increase the amount of mana required to absorb a certain amount of damage.

Actually calculating the EHP from mana shield gets slightly complicated and involves a lot of factors such as what spells you’d like to reserve mana for. Your ult costs 200 mana at all levels, which can be quite significant early on when your entire mana pool might only be 500. You also have to account for mana costs of any items you want to use.

Calculate effective health points with mana shield

Either you will (1) run out of HP before running out of mana, or you will (2) run out of mana first, and then HP.

Let E be your EHP against whatever damage type without mana shield, let M be your mana pool after subtracting mana costs of spells you’d like to cast, and S be your EHP with mana shield.

In case (1), the calculation is simple. Since you had sufficient mana for all incoming damage, mana shield acted as a simple 2.5 EHP multiplier. Thus, in this case, we have: S1 = 2.5E.

Now consider case (2), where you run out of mana and then run out of HP. In this case, all mana was used, so mana shield contributed 2.5M EHP, because you absorb 2.5 points of prereduction damage per point of mana used. Your normal EHP contributed exactly its face value. Thus, S2 = E + 2.5M.

If M = 3/5E, then S1 = S2. Thus, if you have mana equal to exactly 3/5 your EHP without mana shield, you would run out of HP and mana simultaneously, which means this is the break-even point between being limited by mana and limited by EHP. If we start at this value, then if we were to increase our EHP, we would run out of mana first, so S would be S2. Here, S2 < S1. If we were to increase our mana, we would run out of HP first, meaning S would be S1. Here, S1 > S2.

Thus, your effective health points are:

S = min(S1, S2) = min(2.5E, E + 2.5M).

Which items give the best EHP?

The result of this is simple: In the case where you run out of HP before mana, buying additional mana does nothing, and buying EHP gives you 2.5 times the face value of the EHP (likewise for EHP reductions). In cases where you run out of mana before HP, mana gives you EHP at a rate of 2.5 EHP per mana, and normal EHP only gives you face value. But how do you know which situation you’re going to run in to?

Here are some EHP amounts with some various item combinations. The columns of the table, from left to right, are your physical or magical EHP, your max mana, your physical and magical EHP if you were limited by EHP versus limited by mana, and then your final EHP (the lower of the two components to the left). Everything after that is just value analysis.


Here are some of the main things to take away from this table:

  1. 90% of the time, you’re capped on mana rather than HP. This is a very important observation, as it means buying mana will help your EHP, and buying items that increase EHP without increasing mana are often potentially inefficient.
  2. Likewise, since you’re usually capped by mana rather than HP, suffering from a damage amplification has a less significant effect than on a typical hero, since it will only reduce the normal EHP component of your total EHP, not the shield component. You have to be subject to a pretty massive damage amplification (like Slardar’s Amplify Damage) to run out of HP before mana in typical situations.
  3. Items with significant mana costs can really put a dent in the EHP that they would otherwise be giving you.

When to use the shield

The problem with having mana shield on constantly is that often times, you have an easy way of regenerating HP — whether tangoes, salves, or lifesteal — but may not be able to match that regen on your mana. While some builds are more prone to this issue than others, all builds will have to worry about this at least in early laning, when this is especially important due to the lower efficiency of mana shield before it’s maxed.

A typical example is that you start with wraith band and tangoes or any other combination that gives you some okay HP regen, but nothing to really keep your mana up. You’ll want to use mana shield more sparingly. Later, if, for example, you have phase boots, ring of aquila and lifesteal, you’ll have good HP regen from the lifesteal, but the minor mana regen from Aquila won’t be able to keep up with it while tanking neutrals.

One of the main determining factors is whether the damage you’re taking is in small, predictable amounts, hereafter referred to as controlled damage, or in amounts that are either predictably large or unpredictable, called uncontrolled. Examples of controlled damage would be:

  • Casual autoattack harassment from an enemy in lane.
  • Attacks from neutral or lane creeps.
  • Casual nuke harassment.
  • Damage over time and non-stunlock spell combos, since you’ll generally be able to turn mana shield on in the middle of them if they turn out to be too much damage.

Now, examples of situations where you would want to have mana shield on due to potential uncontrolled damage:

  • someone throwing a stun at you, with the possibility of follow-up;
  • a full combo from heavy nukers;
  • a dangerous enemy hero with some form of invisibility. Riki, while uncommon, is the perfect example of this;
  • a hero who can initiate from afar, such as a Storm Spirit.

So, while farming neutrals, you have to look at the enemy lineup and determine whether or not you might suddenly find yourself being initiated on. But, remember that there is absolutely no harm in having mana shield on while you are not taking any damage. Thus, you may act like in the following:

  1. You walk to a neutral camp and attack them. At this point, you should have mana shield on.
  2. As the creeps attack you, you turn it off so you don’t lose mana.
  3. The creeps have been out of the camp too long, so now they start walking back. At this point, you can turn mana shield back on since you aren’t taking any damage.
  4. Creeps attack you again, so you turn shield off.
  5. You finish killing them, so you turn it on and walk to the next camp or lane. Shield toggling is how you sqeeze out a little bit more farming safety, like how Anti-Mage will treadswap frequently while farming the jungle.

Mystic Snake

Often regarded to be mediocre, this skill is actually quite good for a number of things. The only two heroes that the skill maybe should be skipped in lane against are Nyx Assassin (although you can always just hit him with an auto right before the snake would hit him), and Puck. Even then, it still does much more than just a lane tool. Let’s first look at the numbers, and then how to use it.


Mystic Snake costs {140, 150, 160, 170} mana, hits {3, 4, 5, 6} targets, and deals damage plus mana steal to each target. The base damage is {80, 120, 160, 200}, and each bounce increases the damage by 35%, non-compounding. Thus, if d is the base damage at our current level of mana snake, then the nth target (including the first one) will take d * (1 + 0.35 * (n-1)) or d(0.75 + 0.35n). The mana steal is {11%, 14%, 17%, 20%} of each target’s total mana pool.

So the damage done to each target for every level of mana snake is:

  1. 80, 108, 136
  2. 120, 162, 204, 246
  3. 160, 216, 272, 328, 384
  4. 200, 270, 340, 410, 480, 550

As you can see, it starts getting pretty powerful. A max level snake that hits all four units in a creep wave and then an enemy hero will deal 480 damage to them. If you’re against multiple enemy heroes in lane, it becomes even better.

It’s great for farming neutrals too. The total amount of damage that it is possible to do with each level of snake (i.e. the sum of the damage to each target) is {324, 732, 1360, 2250}.

But the mana steal is the nice part, because it allows you to keep casting it. One ranged creep has 500 mana, hence returns {55, 70, 85, 100} mana. In lane you’re likely to hit the ranged creep and one enemy hero. Maybe two if they’re positioned poorly. But what we care about is the sustain. Level one heroes have about 300 mana, therefore hitting the ranged creep and the hero with a level 2 snake gives you back about 70 + 42 = 112 mana, with a net mana cost of 38.

So the lessons to be gleaned from this table are:

  1. Level 1 snake sucks.
  2. Level 2 snake is okay if you can hit both the ranged creep and one hero.
  3. Level 3 snake good, for it can hit all 4 lane creeps and an enemy hero.
  4. Level 4 snake is amazing, in the lane, in the jungle and in battles where you hit many heroes, thus regenerating a lot of mana, especially in the late game.

How to use snake

In lane, snake usage is pretty self-explanatory. Make sure it hits the enemy ranged creep and the enemy hero. The first is usually not hard, although there are rare cases where creeps will be far enough apart to not bounce, usually as a result of the enemy manipulating aggro. The second requires some timing. It’s tempting to use snake to get lasthits and harass at the same time like you would do with a mid hero, but smart opponents will often see that one coming. Thus, you should only try to lasthit with snake if absolutely necessary. Rather, you should aim to use it when the enemy is going to move in for a lasthit. A ranged hero who is really on point can often move away fast enough, but a melee hero usually has to get too close to dodge snake if they want lasthits.

If you’re worried they might dodge, you can send it at the hero first to give them less time to react, but it will obviously do way less damage, so try to avoid that.

One thing that can be tempting to do is to send a snake out such that it will hit an enemy right as they TP to a tower, like you’d do with Mirana’s sacred arrows, Lina’s light strike array, or Pudge’s hook. This is actually ineffective, since they’ll still have the fountain regen buff to recover whatever HP and mana they have lost to the snake.

Level 3 snake is the first level that can hit five enemies (four creeps and a hero), so level 5 is usually when you want to use it far more often. You can still harass with level 2 snake, you just have to have 1 of the enemy creeps dead first (and you have to watch your mana since it doesn’t sustain as much). The same can be used to more effectively harass two enemy heroes with a level 3 snake.

Laning against multiple enemies can range from “field day” to “hell”. If their heroes are equipped to kill you (somewhat unlikely due to your tankiness), it’s pretty bad for you, and you have to try to just get levels. If that’s not the case, your lane should be pretty easy. Obviously you should be aiming to hit multiple enemies with a snake. Often times, uncoordinated pub dual lanes will have one hero intelligently sit back to not get hit by snake, but the other one will move back as the snake comes out in an attempt to avoid it, get hit anyway, and let it bounce to the other guy.

Jungling with snake is a different story. You know exactly how much mana you’ll get back. You generally want to fire the snake at the lowest-HP creep since then you’ll do more damage to the other creeps with the snake. Camps with 3+ manaful creeps are great, as they allow you to come out ahead. You can even pop out of a lane and fire a snake as a way of refilling mana.

The holy grail of camps for Medusa is actually mud golems. The trick here is to turn split shot on, attack both of them until they are low enough to die to snake, then snake them. The snake will kill the first, kill the second, then bounce to the minigolems. Each golem has 400 mana, which gives you 6 * 80 = 480 mana, that is 310 mana after considering the mana cost.

But we’re not done yet. With item dropping (or treadswapping if you opted for power treads), we can squeeze yet more efficiency out of the jungle. Send a snake out, drop mana items, let the snake come back. If you drop a huge item like a skadi, you can potentially get more 600 mana from one cast of snake on a mud golem camp. This can be rather essential if you don’t have a big mana regen item.

Split Shot

By far the simplest of the non-ultimate abilities. Not much to say here. It’s a plain -20% to right click damage while active. It reduces base and bonus, but not attack modifiers. All attack modifiers only apply to the primary target. Does not heal with octarine like cleave does. Total right-click damage output not including modifiers is effectively multiplied by {1.6, 2.4, 3.2, 4.0} assuming you hit the max number of relevant targets.

So, how do you use it? Pretty simple. One relevant target in range? Turn it off. More than one relevant target in range? Turn it on. There’s one target that you absolutely need that +25% DPS on? Maybe turn it off. I say relevant target because not every target is worth sacrificing DPS over. If there’s a squishy target that you can kill and an extremely tanky target that you can’t kill, it’s not worth losing DPS to do some chip damage to the tanky target. People with ghost scepters on obviously shouldn’t be counted as relevant targets. You may also want to turn it off to avoid hitting blademail (if you have huge amounts of DPS, otherwise ignore it) or spiked carapace.

There’s really not much else to say here. Regarding leveling, keep in mind that the hard camps only have 2 or 3 creeps (so only level 1 or 2 is needed for those), and ancients only have 3. Smaller camps may contain more, but those are farmed quickly with snake anyway. Mud golems, the apex camp, are only 2 units, so 1 level of split is fine there. Further levels obviously help with farming, especially if your team is stacking for you, but not nearly as much as the first 2.

Stone Gaze

This ability, simply put, is not very good. Easily the hero’s worst ability. It costs 200 mana, that is 500 EHP, is hard to actually hit enemies with — it’s really on them if they got hit by it and weren’t already disabled — and makes enemies immune to magic damage.

It’s also the hardest to use effectively. The three main uses of it are:

  1. Follow-up on another disable, since that’s pretty much the only way to guarantee you hit it.
  2. Keeping other enemies away from you while you and your team focus someone down.
  3. A shit-hit-the-fan button to let your team disengage from a fight. Not necessarily a losing fight, but maybe your team used a lot of cooldowns on one person and now cannot fight the rest of the enemy team. You really don’t want to use it to try to initiate.

It amplifies physical damage on stoned targets by {30%, 40%, 50%}, which is really not very good scaling.

Skill build: Putting it all together

In a typical game, you should never ever have a level of stats before your other stuff is maxed. In general, your skill build should either be W > E > Q with the first point of ult at 9 or 10, or E > Q with no W.

There skill build for the majority of games is this:

  1. Shield
  2. Snake
  3. Snake
  4. Shield
  5. Snake
  6. Shield
  7. Snake
  8. Shield
  9. Split/Ult
  10. Ult/Split
  11. Split
  12. Split
  13. Split
  14. Ult
  15. Stats
  16. Ult 17-25. Stats

The basic justification is:

  1. Splitshot does nothing in lane, and before you have a decent amount of DPS makes your neutral farming speed with right clicks go from slow to slightly less slow. You also won’t be able to sustain your HP in the jungle before you have lifesteal anyway, unless you kite the creeps around which will further reduce the farm speed.
  2. Snake is good in lane and also the best neutral farming method you’ll have until you actually get some reasonable DPS like a mask of madness.
  3. Shield needs to be maxed pretty early for mana efficiency reasons.
  4. The ultimate isn’t an amazing ability to begin with and costs a large amount of EHP. That’s if you even get it off to begin with, which isn’t guaranteed if the enemy is able to get the jump on you and deal enough damage such that you don’t have enough mana to cast it. On top of that, the scaling is fairly bad, so it’s not worth getting more than 1 point past level 1.

Puck is pretty much the one lane opponent where you do need a significantly different skill build. Usually, you would want 4 points of shield and 3 points of stats at 7, then start leveling split shot and ult. Puck has high base damage so you’ll need the extra damage in order to be able to actually lasthit (although you need less points of stats if you got babysat well enough) and obviously Snake is worthless in lane against a Puck. There’s no threat of dying to the Puck, but you need to make sure you’re securing your own lasthits and denying as much as possible.

Remember, the ult is a rather unimpressive ability with a huge mana cost compared to your early game mana pool. It also has rather poor scaling, so it is not worth upgrading over split shot until split shot is maxed. It can very occasionally be optimal to level it at 6, but these are few and far between.

If you have complete freefarm or early stacks, it can be worth stopping at 2-3 points of shield and getting a point or two of splitshot. Then, either continue leveling splitshot, or go back and max shield. This isn’t a build that you have to do often, since even with splitshot, your farming speed against stacks early on will come from snake anyway.


Items are the big part of playing this hero, and the part that so many people get wrong. Terrible item builds are most of the rest why people think this hero is only good lategame. Most people agree that the hero scales extremely well with items, possibly better than any other hero. Why, logically speaking, would a hero who gets more mileage out of items be weak midgame? Shouldn’t the first big item be just as impactul as the fifth?

The truth is that people often spend their first few big items’ worth of gold extremely inefficiently, then actually start carrying when they get the next few big items because those are actually effective. But we’ll get to that in time.

Starting items

Standard agility carry stuff works here, but you don’t need quite as much regen as other heroes early on. You’ll be buying a little more regen fairly early on in lane. Due to mana shield, one stack of tangoes can last you quite long, and you’ll probably have a support with some spare regen on top of that. Wraith band plus one tango works fine, but you can do wraith band components or circlet and ring of protection plus two tangoes or tango and salve.

Lane progression

This is where the build becomes very situational. These are the items you can buy and when to buy them:

  • Brown boots: You need these at some point, but don’t prioritize over regen.
  • Upgraded boots: Provides a huge boost to hit-trading and lasthitting (especially phase), but again don’t prioritize over regen.
  • Aquila: Provides mana regen thus allowing you to use more mana shield. This extends any HP regen you do have. If you always have mana, each tango goes from 115 HP restored to 287 effective. However, this requires use of the courier, meaning often times you’ll have to spend the money elsewhere first.
  • Stick: Good if enemies are using spells. Especially low level spells, since due to mana shield it can possibly leave you with more than if the enemy had casted nothing. Whether or not it’s worth upgrading depends on the game. I’d say if you want to change up the build and get more damage before a big survivability item, then a wand would be the way to go.
  • Soul ring: Best overall regen item. Not always necessary in lane, but if you do need both mana and HP regen, it’s the best option by far, acting as a mini-perseverance even if you never touch the active. It requires use of the courier but is overall very worth it. It’s also nice for when you’re completely out of mana, as it allows you to get your mana up quickly while lifestealing HP back.

Note: Treads or Phase?

I usually get phase for the laning power and mobility in fights. Treads is perfectly viable too since it allows for some stat switching while using snake.

Next items

Your next items should almost be:

  1. Mask of madness: Single biggest farm accelerator you can buy, plus keeps your HP up while jungling or laning. Due to mana shield mechanics, the main drawback of taking more damage is partially neutralized. A normal hero loses 23% of their EHP when they turn on Mask of madness. Medusa in most situations only loses 10-15% depending on your exact amounts of EHP and mana. Thus you get cost-effective DPS on a DPS-starved hero while being able to ignore most of the drawback. Refer back to the table of items in the mana shield section for some examples.
  2. Large survivability item: This should usually either be Skadi or Bloodstone. Skadi is overall better if you aren’t really sure what you’re doing, since it provides the most EHP of any item (refer back to the table again), a little bit of DPS, and a slow to help chase. Bloodstone will be a little cheaper since you’ll likely already have the soulring, plus gives you plenty of sustain. It’s the best item for games where you’re ahead, games where you need to siege, or just games with a decent amount of activity in general since Medusa is really good at not dying in teamfights. It’s really Medusa’s equivalent of Heart, providing a large amount of bulk and insane amounts of regen. The in-combat regen shouldn’t be underestimated either, as it can easily end up giving you hundreds of EHP even in shorter fights.

Mask of madness, Yasha or both?

Mask of madness and Yasha have similar average DPS, but Mask of madness delivers all of its bonus in a window which has better synergy with your ult, and killing people in general. You don’t ever want to be trying to kill one person for over 12 seconds to begin with. Plus Yasha doesn’t give HP sustain. However, with bloodstone, that’s irrelevant. So with Bloodstone it’s kind of a tossup, but overall I prefer Mask of madness. You can also get both, but you’ll run into slot issues pretty quickly, not to mention it delays survivability items by a pretty long time.

There are a few situational items you can buy at this point instead of Skadi/Bloodstone.

  1. BKB: Sometimes it’s absolutely necessary, such as against Anti-Mage, Phantom Lancer, Invoker, Nyx, or when it just makes their entire team do nothing (see match 1704045855). Not usually worth buying against nukes alone, since you can just tank through them.
  2. Butterfly: Very risky to buy it this early since it’s likely to get countered, but if you really need to cheese it against a right click heavy team it can be viable. It would be similar in concept to a glasscannon Phantom Assassin build with BKB, where you want to steamroll enough such that when enemies get MKBs you’re far enough ahead for it to not be a big deal.
  3. Linkens: Mainly Anti-AM since he’s not going to get an item to pop it (other than Abyssal, but blocking Abyssal is worth it). There are various niche situations where it ends up being really good, such as having on Omniknight on your team since the linkens would block purges.
  4. Hex: Haven’t actually tried this one as a first big item, but could be great for 5 manning.

More big items

After this, anything that gives mana, intelligence or DPS is good. This includes:

  1. MKB: Usually my first big DPS item. The chances that nobody gets evasion on the other team decrease quickly as the game progresses, plus it’s usually not hard to find some channeled or long-cast-time spell to interrupt. With the popularity of solar crest, MKB is usually a must. Plus, the raw damage synergizes well with increased attack speed from Mask of madness.
  2. Crit: Big single-target damage. Most useful when you’re absolutely sure nobody is getting evasion, and that there is a single target you need to focus.
  3. Butterfly: Lots of DPS, and one of the few EHP increases that works well with mana shield, since evaded attacks do not consume mana.
  4. BKB: Sometimes it’s necessary, especially against heavy manaburn or large amounts of chain stun or disarm.
  5. Skadi: Largest bulk item you can buy. Plus a BKB-piercing slow and 25 damage with bonus attack speed. Buildup is okay for this hero.
  6. Mjollnir: You don’t buy this for the DPS, you buy it for the shield. It punishes enemies for attacking you. If nobody’s focusing you to begin with, it’s not the item to buy.
  7. Moonshard: It’s DPS. Nothing to really say. You’ll always want to eat one when you’re otherwise full.

Now, for some much more situational pickups:

  1. Divine Rapier: Very risky. Usually the best strategy is to get one survivability item like Bloodstone or Skadi, then get the Divine Rapier. Steamroll a fight or two with the Divine Rapier, and use the money to buy another survivability item. The problem with rapier is that it forces you to 5 man. You lose the option to split up and farm, since you run the risk of getting caught out.
  2. Assault Cuirass: Very rarely necessary, but against heavy minus armor you might get your EHP so low that you end up capped on EHP rather than mana, so you lose quite a bit of EHP.
  3. Refresher: Not something I’d buy very often. Not very useful stats with the new buildup. Primarily for refreshing a BKB and ult. Note the high active mana cost.
  4. Diffusal: Sometimes necessary against ghost scepters, but also good for purging allies.
  5. Hex: Especially good if you didn’t get Bloodstone since you’ll lack mana regen. The disable is always useful, and it gives tons of mana.

Things to not buy

Items you should absolutely not buy except maybe in extreme edge cases that aren’t worth discussing:

  1. Manta: There’s an entire section on this. See below.
  2. Satanic: You’ll just end up with no mana left, not to mention you have to have enough DPS to make it worthwhile to begin with.
  3. Heart: Same as above.
  4. Blademail: The damage returned will reflect the mana shield reduction, so you won’t return that much.

Is Manta really that bad?

Yes. This is not an exaggeration. Let’s analyze what the item gives:

  1. Movement speed and DPS: okay, but we get this from the raw Yasha or Mask of madness.
  2. Int: The 130 mana is nullified by the fact that Manta costs 125 mana to cast, every time you want to cast it. If you don’t have a way of sustaining the mana, it’s extremely taxing over time.
  3. Illusions’ DPS: Basically nonexistant. We’re talking about a hero with no major source of illusion DPS (no manabreak, diffusal, metamorph, etc), with below-average agility gain for an agility hero. Then there’s the fact that ranged Manta illusions deal an entire 28% outgoing damage. Not to mention, the illusions don’t get mana shield and take 400% incoming damage, so they die to a breeze.
  4. Disjoint and dispel: If the enemy has very little disable that you actually care about (disarm, stun, hex), you can simply absorb it, wait it out, then continue about your business. If they have loads of disables, you need BKB anyway. Most soft disables you can simply ignore. Silences do nothing, attack speed slows can be counteracted by getting more attack speed. Medusa isn’t a hero like Luna where one disable is likely to allow the enemy to pile on you and kill you.
  5. Splitfarming: By the time the illusions with their 20 second duration actually get to some creeps, they have no DPS with which to actually farm.
  6. Scouting: The only thing I can’t argue against. Worth 5000 gold? No. Not to mention if you use it to scout, then you don’t have it available for survivability purposes.
  7. Confusion: Works until the illusions lose half their HP to a random AoE nuke, or because people see the stone gaze, Mask of madness, mana shield visual on the main hero.
  8. Sieging: You’re already extremely tanky so you can just walk up the T3 ramp half the time. If not, you’re Medusa: you’re not in a hurry to end the game.

Untested items that might be good

Haven’t tried these. They sound good on paper but their actual effectiveness remains to be seen.

  1. Medallion and solar crest: Evasion works wonders, plus it gives mana regen. In theory it would be good survivability, and helps with killing important targets. The only downside is that it does nothing against non-physical damage.
  2. Guardian greaves: Gives you a heal and mana restoration, although if I was going to get mana boots early game I’d rather just turn it into a Bloodstone and let someone else get the mek. The problem with GG is that the 1600 gold recipe leaves you at an awkward spot for a few minutes, where the mek ends up hurting your own EHP due to its high mana cost. It also obviously takes up your boot slot, and treads/phase give very cost effective DPS.


This patch sucks

This patch — this section was written during 6.84 but 6.85 and 6.86 did not undo the gold changes — is not a farming patch. You simply don’t get good amounts of gold from farming, unless you’re an extreme farmer like Anti-Mage or core Naga Siren. You have to fight to get appreciable amounts of money. Fortunately, if there’s a hero that’s good at sticking around in fights and not dying, it’s this one. Read up on the section about your ult to learn how to use it properly in fights. You really want to just be 5 manning, especially once you have your first big item but usually before that as well, since that’s the easiest way to actually get that big item in the first place.

When to pick the hero

After you know for certain that:

  1. The enemy mid is not Invoker unless you feel comfortable rushing BKB.
  2. The enemy safelane is not a Phantom Lancer or Anti-Mage with half a brain.
  3. The enemy team does not have a highly competent Slark.

The first point should be fairly self-explanatory. Invoker is one of the few heroes that really necessitates a BKB. Number two is a pretty hard requirement too. Against AM, you want some combination of Linkens, BKB, and Butterfly. Against PL, BKB and Butterfly are good, as well as Mjollnir. Point three is trickier. I don’t think Slark counters Medusa nearly as hard as most people believe. First of all, if Slark goes for Medusa in a fight, he’s probably going to lose his team the fight. Medusa can continue attacking people while Slark hits her. Yes, he’ll be stealing lots of stats, but Slark gets the same amount of stats from hitting Medusa 10 times as he does from hitting supports 10 times. The difference is that hitting supports 10 times results in 1-2 dead supports, whereas hitting a Dusa 10 times results in the supports and the Dusa still being alive, thus helping Dusa fight the rest of the enemy team. There’s also the issue about using Dark Pact to remove her ult, which is a possibility but mainly happens in theory only. Most Slark players simply do not have the discipline nor planning to save a Dark Pact for Stone Gaze in the middle of a teamfight.

Now, you also need to think about your allies. As you’ll see below, you want your supports to be doing fairly specific things. You don’t really want a “win lane lose game” support like Lich, because your issue is generally midgame rather than laning. This means you typically want supports that have gank potential so they can gank mid. You want your mid hero to carry you for a bit. Crystal Maiden is perfect for this due to the mana aura. You’ll usually be fine even if you have a jungler since you can hold your own, and if your jungler isn’t countered too hard you’ll end up with another farmed core to cover your early to midgame.

Overall, I would say the best allies are Crystal Maiden and Techies, for the mana aura and obvious game drag-out capabilities respectively. You want your mid hero to be something that stomps the midgame like Invoker, Queen of Pain or Zeus.

Overall play style

Generally, you want supports to zone the enemy offlane for the first few minutes of the game to give you enough of a headstart to decisively win the lane 1v1. After that, they should be aggressively making space unless your lane is free kills. Whether you’re against one hero or multiple, your supports absolutely need to not leech, as snake levels are how you win the lane. When against multiple heroes, sit back and farm, use mana shield and snake intelligently. You’ll probably want a stick too. The goal is to get levels to win the lane, then use your lane win to get farm and build an even bigger advantage while your supports help the other lanes, especially mid. A well-pampered mid is usually the key to successful Dusa games, since it allows you to fight with your mid hero for easy gold (and Bloodstone charges) despite not having that much DPS yourself.

The goal early game is to get as much for yourself without being a burden, because it’s much better to have a big mid hero to cover Medusa’s weak period than to try to bypass it (which simply doesn’t happen often). Until you get one big DPS item, you’ll likely be able to do work on supports who are disabled by your team, but you’re still mainly just an unkillable blob at this point, which is your goal: you want to stick around in fights for as long as possible to soak up gold, and maybe ult.

You have to remember that unless you’re the sole carry of your team up against multiple enemy carries, or you’re against a carry that counters you like AM and want to end before they get big, you are not in a hurry to end. You still want to group up and take fights as five because it’s simply a more efficient way of getting gold compared farming, but you don’t need to take anything risky, and you don’t need to try to go uphill until you’re ready.

Dealing with counters

No, there isn’t some magic answer to AM/PL. The anti-AM items are BKB, Linkens (one of these two should likely be your first big survivability item), Butterfly, and as much DPS as you can pack in. You need to be able to kill AM and his illusions as quickly as possible. You also need to not be a walking mana void bomb, so going for a big mana pool build like Skadi or Bloodstone doesn’t work all that well.

The main thing you have going for you in this matchup is that you can outfight Anti-Mage in those crucial 15-25 minute fights before he gets Manta. AM will win the farm war if you both try to rice. Thus, you need to press your advantage as much as possible. AM is one of the heroes that you may need to get a Divine Rapier to outcarry, and if he’s the only real rightclicker on their team then it’s not the end of the world if you lose it.

AM’s problem is that he runs out of slots because he’s hugely dependent on Battlefury and Manta. He’s always going to have Boots, Battlefury, and Manta, leaving only three slots for Butterfly, BKB, MKB, Abyssal, and Heart. If he doesn’t go Heart, his physical EHP will be terrible, especially with no Butterfly or after you pick up a BKB. If he doesn’t go MKB, he’ll have trouble hitting you. If he doesn’t go Abyssal, he loses a lot of kill potential on you. If he doesn’t have BKB, he’s vulnerable to disables. So AM has to pick and choose his capabilities, whereas if the Dusa goes something like Boots-Linkens-BKB-Butterfly-MKB-Divine Rapier (not necessarily in that order), then you’re not really missing anything important.

Phantom Lancer is a lot different. He doesn’t crumble as much under early pressure since his early fighting is a lot better than AM’s. He also can keep creating more illusions. Your best bet here is probably Boots-Bloodstone/Skadi-BKB-Mjollnir-Butterfly-MKB. When you pick up the BKB depends a lot on whether your team has enough control to get PL off of you immediately if he goes on you. If they don’t, you need BKB to prevent your EHP from evaporating. If they do, then you can be much more offensive.

Invoker isn’t that much of a counter since you can just BKB, and even if he didn’t have EMP, he has so many disables that he might be worth buying BKB against anyway.

Nyx is much of the same. You’ll likely need a BKB against him anyway so that you don’t stun yourself when your splitshot/mjollnir/snake randomly hits him in a fight, plus he’s got tons of burst and manaburn on top of that. It’s doable without BKB (linkens makes it very difficult for him to solo gank you which can be a problem if he’s massively ahead), but BKB + 5 manning is usually the answer. In terms of laning, you can still use snake, you just need to throw the snake, then throw an autoattack right before the snake hits so that the autoattack procs spiked carapace instead. Nyx loses the mana from snake, the manacost of carapace, and only reflects a weak autoattack. Obviously him keeping you at 0 mana in lane can be a problem, so you’ll likely want wand and soul ring as emergency mana sources. Nyx can’t spam indefinitely unless he bottlecrows.


Leave a Reply