Playing With and Against Phantom Assassin

With Phantom Assassin surging in popularity earlier this year, and her being in almost every game during/since the Manifold Paradox event, I thought it would be a good idea to put out some general tips and data.

Dodgem Cars

Since Phantom Assassin has 50% evasion, you’ll notice that when you attack her you end up with more misses than a Saudi king; and they have a lot of wives. I tend to find that most people have enough sense to get an MKB against PA, even on heroes who don’t like getting one. It’s usually one of their first 3 item pickups. But this doesn’t only impact the PA, this impacts her team as well. I see far too many people get Butterfly on someone like a Luna when they’re on team with a PA. I find this to be a poor item choice, since the enemy carries are already going for an MKB. 50% evasion on PA, and then 35% evasion on her teammate tends to be putting too much emphasis on evasion. Even sometimes Halberd can be a bad pickup alongside a PA. I say that it’s a generally a pretty bad idea to get Butterfly when you are playing with PA, no matter how good of an item you think it would be for you. It’s putting too many Phoenix eggs in one basket.

> Safelane Please

Carries tend to have one thing in common that’s not Princess Leia related: they lose more when they have a carry on their team . While Phantom Assassin has seen a massive shift in laning over the past year, many still safe lane. All of the atypical ‘safelane carries’ also utilize the jungle to be successful. This means that PA’s worst teammates tend to be the heroes that want the same as her. Void, Terrorblade, Anti-Mage, Juggernaut, Chaos Knight, Lifestealer and Terrorblade all statistically don’t match up well when on the same team as PA. Despite laning shifts (such as a recent surge in offlane Void, Mid PA, etc), I would call these heroes fairly similar in laning: historically they’ve wanted to safelane, and use the jungle a bit. Even when in a different lane, they have similar ambitions and fill the same place in team (in most pubs).

You actually end up doing yourself a disservice if you pick this sort of hero, PA doesn’t perform too well when she plays alongside them. It’s humorous that people slur low/mid tier Dota for having too many carries, and yet PA still manages a better win rate in these ‘carry centric’ brackets despite the disadvantage. If you want to make the most of a teammate’s PA pick, it’s probably a hero you wouldn’t initially think of: Techies.

Yeah, based on a reliable method (matchups of win rates), Mortred’s best team mate overall (since he was added to Dota 2) has been Techies. Techies happens to go incredibly well with all the heroes that PA does bad with: the ‘safelane carry’. It’s incredibly likely that a offlane duo Techies ends up having such an impact on the enemy carry, that it gives PA the ability to ‘out-carry’. Being able to carry isn’t just about getting farm, it’s about stopping the enemy farm as well. With early kills, making it risky for the carry to push, and constantly restricting enemy movements, Techies ends up putting the PA in pal.

For the Veiled Ones!

Phantom Assassin is likely to only be on your team half the time, which still leaves an ungodly amount of games when you have to play against her. Sand King is one of her top counters in the game. Before IceFrog decided to make PA a nightmare by removing/reducing evasion-disabling from Legion Commander and Doom, I often enjoyed using them. Other times it was a Faceless Void, of course, that was before he removed the evasion-removal Chronosphere had. Yeah…. IF really has been egging on PA in recent history. Regardless, Sand King and his Veil of Discord works well against the Sisters of the Veil.

An essentially instant blink-disable – including mass Magical burst damage – frequently nets the king of the desert a kill. Sand King also has a much easier time getting a blink than heroes of a similar role, which allows him to exploit PA’s early game weaknesses. Omniknight and Razor are also top counters, who both tend to do well against ‘in your face’ heroes like PA or Lifestealer.  Centaur Warrunner can sometimes be a “poor-man’s Sand King”; lots of magical burst while Stampede helps with kiting PA throughout the game.

Easy to Overlook

I don’t like the common trope of “play your own game”. I tend to find that Dota is a team game resulting in one player’s actions having a massive impact on the success and failure of another, and the team as a whole. Terrorblade has 7 base armor, so I believe it’s rather silly for a laning Terrorblade to get a stout shield. Physical damage is not a weakness that needs to be solved in his starting items. Extra stats are far better at boosting his survivability and early game impact. So I tell them. After all, a Stout Shield’s 250 cost is 40% of your starting gold, which is a lot to spend on a hero with 7 armor and less than 500HP (435 base). Sometimes this occurs for Phantom Assassin as well, although not to the same extent. With 4 base armor resulting in about 20% physical reduction, and then 20% evasion on top of that, she looks pretty good against physical damage early on. And yet the vast majority of PAs will get a stout shield (if not a PMS) unquestionably.

Stout shield is generally great on high HP heroes with low/average armor, such as Centaur, or most famously when tanking creeps (such as in jungle). But PA spends a lot of time daggering creeps from afar, and is likely to have her first death predominantly from magic (perhaps from a mid gank such as QoP). It’s 250 gold for a ~53% chance to block 20 damage. I’m not saying you shouldn’t get it, but I think it shouldn’t be anywhere near as set as stone as it is now (and TB should probably get it less too).

When you kill a PA early on in the game, how effective was her 250 gold spent against you? It’s a thought to consider on almost any carry, when you die in the early game (even if you think a support or lack of wards was the cause), how effectively did you utilize your gold to prevent that high-impact early death? And the better part of this exercise, is when you kill an enemy carry in the first 20 minutes of a game, what mistakes did they make that allowed you to kill them? Perhaps their lack of boots due to their Midas rush enabled you to kill them, which you should factor into your decision making process next time you’re the carry. Don’t just learn from your own mistakes, learn from other people’s mistakes as well.

I See You (Later)

PA is an interesting hero. She basically forces you to buy an MKB, which can become problematic if you also need a slot for detection, anti-kite, magic immunity or a dispel. But if we are going to see her as much as we are, I think it’s at least a good idea to attempt to improve your ability to play with and against her. I try to encourage my readers to follow the philosophical viewpoint of Freethought when dealing with Dota 2.  That is to reject dogma and tradition, instead using logic, empiricism, and reason to come to conclusions on heroes, strategy, itemization and all things Dota related. So I’ve laid out some data, some reasoning, and some ideas. Give it some thought and hopefully you weren’t distracted by my terrible Carrie Fisher joke.


See Also: Building Phantom Assassin Effectively – My article from May this year, laying down Key Concepts with the hero while encouraging people to take advantage of her unique timing window.

BTS and DC Have The Best Parts of the $40 Million Dota 2 Ticket Industry Pie

As Dota 2 rose in popularity, more and more people tried to start their own casting team and studio to get a piece of the pie. But in recent times, organizations like JoinDota have become increasingly archaic as Beyond The Summit and DotaCinema claim victory. While Dota 2 expands across the board with incredibly revenues.

If we go back to The International 2, you’ll find that a lot of the casters are no longer leading the charge in the scene. Ayesee is currently casting Dota 2 League Season 5, which has struggled to sell any tickets. While Luminous, Draskyl, and Tobiwan aren’t looking so hot right now either. While Slesh, well, I never really liked his casting anyway. But let me explain why I’m calling the win for BTS and DC.

How Big Is The Industry?

I’ve heard some people complain about there being too many tournaments, but I disagree. I think the problem is that as you get more tournaments, people have a choice, and they end up choosing the product they believe to be superior. AKA, the crappy and/or undesirable tournaments get left in the corner. This isn’t there being too many tournaments, this is simply consumer desires rejecting poor products. We end up with Amazon’s Fire Phones of the Dota world.

If we look at money spent on BTS and DC tournaments in November alone, we’re just shy of a million dollars with revenues of 900k (USD). In October, there was 700k of revenue, and for both of these months I’m only looking at these two tournaments. The Dota 2 ticket industry is on target to hit around 40 million dollars this year alone. That’s 40 million dollars being spent on tickets via the Dota 2 shop in 2014, with about 10 million of it going towards prize pools. While most of this 40 million is Valve’s International, a solid 3-4 million is still going elsewhere. Although primarily to BTS and DC, which is incredibly important. Note that if item sales were included, I can easily see Dota 2 earning over 80 million in revenue for this year, but item data is much harder to come by. But 80 million is a safe estimate.

Ticket competition heats up at The Summit

Beyond The Summit

BTS held the first The Summit in June this year, and it went well. Starting with a base prize pool of 80k, before being raised to 130k by the community via ticket sales. The tournament was enjoyable with a good take on production value. There were ‘In-booth’ cams akin at The International, while also having a friendly talkative atmosphere and couch with the players, and various other cameras and content throughout the tournament. Evil Geniuses took home first place, but BTS were the true winners.

Fast forward to November, and The Summit 2 is in full swing, with the LAN finals starting in a few days time. While the original Summit only raised 50k, which was not even double, The Summit 2 has already increased the base 100k to an incredible 300k. The prize pool has tripled, and is likely to go even higher as there is normally a boost in ticket sales as the LAN starts.

Dota Cinema

Dota Cinema managed to snag Synderyn in time for their first tournament; the Captain’s Draft Invitational around February this year. Synderyn was well received as a player-turned-caster at TI2, but it took a while for him to end up as a ‘full-time’ caster. He was paired with Fail’s of the Week familiar Sunsfan, and they ended up following a incredibly similar path that The Summit took. Starting with a base pool of 20k, it was raised to 38k by the community, just short of doubling. The tournament was unique, humorous, and very enjoyable.

Right now the sequel tournament is being played, also in the Captain’s Draft game-mode. Starting out with 50k, it has already over 250k. That’s a quintupled prize pool, and the same 200k raised by the community that The Summit 2 has achieved thus far. And of course, both these tournaments are still ongoing, and likely to increase in success.

I Choose You, Sunsfanflora

Everybody Else

If we look at the other tournaments, you’ll notice vastly different community response: they aren’t ‘Beyond the Cinema’. Dreamleague Season 2 increased the base 100k, to only 114k. D2L Season 5 has improved 50k to only 58k. Even Starladder S11 has done incredibly worse than S9 and S10, with 80k going to 110k. The Summit 2 earnt more in it’s first 2 days, than Starladder has in its current 16 day run. While DC did it in 12 hours. BTS and DotaCinema hit the ground running as soon as they went on sale, where as other tournaments trod along at a rate slower than Crystal Maiden.

But Why?

To keep things shorter, I will probably go more in depth about why in a future article. But the primary reason is that both BTS and DC are able to create a positive and aesthetic tournament, that are both unique and vastly enjoyable. Synd and Suns have a rapport that is hugely attractive to an English speaking consumer. Far more so than 5 guys talking over eachother and yelling ‘wow’, or consistently getting Rocket Barrage and Battery Assault confused. If Syn and Suns make a mistake, their rapport gets them through it and oddly enough, often adds to the cast. BTS as described a few paragraphs up, are able to create a more serious, but equally unique and enjoyable product. Every studio/company will always have some viewers by their side, loving the way they’ve set up shop, but shifts in the market alongside clever business practices have catapulted BTS and DC into the lead.

Both companies are able to deliver a product that stands out from the crowd. A tournament is no longer simply hosting the event, you have to own the day. Deliver with booth cams, tub interviews, interaction with players, humorous moments, alternative game-modes, in-game cosmetics, compendiums and then some. Tournaments are no longer about just the games, but a series of events and experiences that thrust them in front of the riffraff. Casters are just one piece of the puzzle, and ultimately it doesn’t matter if everyone knows your name, what matters is delivering a product that people are willing to buy.

At the end of the Dota day when Nightstalker comes out to play, I tend to find that the scene is thriving. Ticket Sales make tournaments happen, so using them to judge the success of tournament is far easier than adding viewer numbers from 5 different locations/websites over the course of 50 different games. You may get a few viewers on stream when no one else is playing, but if you fail to convert popularity and desire into a ticket purchase, then you are falling behind the frontrunners.

DotaCinema are likely to hold another two tournaments next year, and The Summit 3 and 4 are planned either side of The International during Feb-April and September-December respectively. With BTS and DC owning the night (like the fourth of July), it’s entirely possible for well known names and organizations to fall by the wayside. Who knows what the future holds, or where the meepos will fall.


Be sure to follow @ZotaTips on Twitter for a regular stream of helpful hints and advice, or to leave comments here if you enjoy the website.

Note that sponsor titles and stylized names have been removed/normalized in this article. All money is in USD.

Data on Server Size and Dota Demographics

Dota 2 is a global game, but I wasn’t sure exactly how big it was in each region. I was determined to know, so I set off collecting data for an entire week in mid August 2014 in order to find out. My method for this article is to use the Waiting Queue to determine the Server Size. I can obtain data based on how many people are currently playing Dota, as well as data on how many people are waiting for a game of Dota per server. I can use this to extrapolate and create models to determine the size of player bases across the globe.

Server Range

Every day each server peaks and troughs at different time. Using this data, I can create a Server Range; the min and max range of players that are searching for a game at any point on each server. I have also taken samples from across the day, to determine the average amount of players searching per server. The average also helps determine the gradient; some servers have a very steep peak, others are much more gentle knolls.

The image is a tad small here. But Russia and EUW are the two biggest servers, they very consistently have a high numbers of players on them. Conversely, in Dubai, Korea, and South Africa there is an incredibly small amount of players. Elsewhere in Asia, both Chinese servers are big, but they tend to fall off really hard. Each day, Chinese Dota outclasses SEA Dota overall, but Chinese servers don’t last very long. In SEA the servers are busy for hours on end, where as in China the servers seem to rise and fall really fast. This causes SEA to have about the average amount of players playing as the two Chinese servers combined, even though SEA peaks lower.  In the Americas, all three servers are similar in size. Brazil has a hard fall off, but still manages to beat USW overall. USE is the biggest server in the region across the board.

My theory is that the bigger your server is, the better the matchmaking will be. Playing on Australia you will likely get rather large difference in people’s unranked MMR, as well as less precise MMR across the board. In EUW however, the pool of players is about eight times bigger, which should cause far better matchmaking. It’s thus better for Dota 2 pubbers to have big scenes in small locations, rather than have small scenes all over the world. Being good on a South African server is meaningless, since you will consistently be put with opponents and allies that are severely under your skill level. This not only makes it harder to get better at the game, but will heavily impact your playstyle; causing you to be much more of a ‘one many army’ dota player.

Player Demographics

Working with my data, I can start to work out the demographics of the Dota population. Some players aren’t very faithful to their servers due to internet pathing, so I’ve split them up into regions. Players are likely to play at least within their region, such as Arabs on either Dubai or an European server, and Peruvians on one of the 3 American servers. While people can play anywhere, the vast majority of players are likely to play at least near them for ping reasons, with a very insignificant amount of say Russians playing in Korea.


Over the course of a 24 hour day in each region, about 46% of the people searching for matches are from Europe, which includes the Middle East and Russia. A solid quarter of players are Chinese, although SEA is not too far behind. In the untitled part is other countries at 2.5%, with 1.9% being Australian.

This data however is susceptible to a couple of outside factors, causing some limitations. In Australia, a very rich county where people game at home, a Dota 2 player can generally play when he wants to. He doesn’t have to travel to a Net Cafe, nor wait for money to afford to use one. In SEA however, Net Cafes are incredibly popular, which also means that the SEA server base is limited by the amount of computers available in Net Cafes, as well as their culture and opening hours. I said earlier that SEA servers were busy for long periods of time, where as China has a much more sudden peak. I actually predict that SEA is probably bigger than what is represented. Overall this pie graph is a tricky one, especially when you consider time zones. Since more time zones will result in lower peaks and higher averages.

Predicting Player Bases and the Miscellaneous Stats

I also know how many people have the Dota 2 client running, which is publicly available information. Generally, there seems to be about 5% of people with the Dota 2 client running searching for a game. It appears that Dota 2 peaks at about 8% of the total monthly player base each day, and that 15-20% of the Dota community play each day.

Obviously the waiting queues get massive surges when a server fail, as people all rush to find a game. The odd thing however is that weekends don’t seem to increase the numbers of players playing. Both in total players online, and in players searching, weekends seem to have no to very little effect on Dota players. Restaurants get peaks on certain days, such as the weekends, whereas Dota doesn’t seem to have a weekly busy period.

This blog has over 200,000 views in the past 12 months, but is 60% USA/UK/AUS/Canada. It has a fair amount of western Europe and SEA;  both Vietnam and Poland are much higher than I would expect, and even have 28 views from Iraq. I imagine the Dota 2 Reddit would have a similar skew, being primarily the ‘five eyes’.


The Server search range is solid data, but other data becomes incredibly difficult to get exactly right. I’ve tried to create logical models and have reliable information, excluding less reliable results from the article.

Another article I am working on is how global events impact the Dota 2 player base. Primarily looking at the Ukraine-Russia crisis and the devastation it has caused to the Russian economy, as well as the Hong Kong democracy protests. My initial data suggests that they both have caused large declines in players. I also should mention that two new South American servers will be online shortly, and will keep my eye on how this impacts the player base. I am however very critical of this decision; the numbers tell me that to keep the player base the most sane, only one server should have been added. A server that caters to Spanish speaking South Americans.

6.82 Win Rate Changes

The new patch has been out for over a week now, which I feel is enough time to do a rundown in both win rate and play rate. I was picky in my data samples because of Techies. So I’ve chosen to compare all 6.82 data including 6.82b, with that of one week in September, a full week with Techies already in the game. Techies addition to Dota 2 should thus have much less of impact on this post. Note that all percents are raw percents, and show actual changes in the percent values.



A lot of increases seem to be due to longer games and the Tier 1s being up for longer. Spectre and Omni received minor changes, but skyrocketed to win rates over 60%. There has only been one other time in the past 2 years where a hero had a win rate over 60% in the last two years; when Centaur Warrunner was added in October 2012. Even Spirit Breaker and Drow Ranger at their worst couldn’t consistently break 60%, and I can’t ever recall a higher win rate. Medusa is another interesting gainer – as she now has a 53.55% win rate – which makes her one of the stronger carries in pub Dota. Lets see if the pros take any notice of her. Some of the hero rises are also a bit odd; Earthshaker and Phoenix are somewhat unexpected.

As for play rate, reworks and remodels take the top 3 spots, but Bloodseeker is entirely ignored. He only went up 0.5%, which is pretty negligible given the time frame. Axe went up, probably as people thought it was the best hero to try out Crimson Guard on. While Spectre was probably as people caught win of the win rate changes, although there’s been a lot of Crimson on her too. People were obsessed with Aghanim’s last major patch, but far less fuss for the new Aghs, with Bane, Phoenix and Puck’s changes mostly ignored, while Timbersaw and his foe Treant managed small gains off the back of their Aghs.



Nerfs to the forever annoying global Tinker and Nature’s Prophet caused some happy losses, while Bloodseeker’s rework will take some getting used to. Phantom Lancer didn’t drop anywhere near enough, off the chart above at only -1.17%. I find Clinkz’ fall rather interesting, I assume people’s experimentation with orbs has resulted in less wins for the hero. This is likely as he’s already improving in win rate as the days tick by. This list isn’t overwhelmed with “early game heroes” either, it’s primarily nerfed heroes. Chen and Bane are highlights, although they both happen to perform incredibly badly against Phantom Lancer, who did get a massive surge in picks post patch. Lone Druid also had a solid loss.

I removed Techies from the play rate, as he’s been decreasing as people move on from the original addition to the game. But apart from that it’s primarily the heroes who were nerfed. Tiny’s decrease is likely not to last; his win rate went up as seen in the first section. Although Tiny was not directly changed in the version, his win rate still shot up a remarkable 2.51%, which is notable.

Building Medusa Effectively

A few months ago I posted an article on Phantom Assassin, and how I thought the majority of pub players were building her ineffectively. I explained a couple of Key Concepts, such as the Evasion Window and put forwards some general ideas on the direction she should be built. I have had over 190,000 views on this site the past year, and this article happens to be my most viewed. So I’m going to do the same with Medusa, because I strongly feel like there are some Key Concepts with the hero that people are failing to grasp.


The Late Gamer?

Medusa has traditionally built as a Hard Carry, and even to this day people love to do so. There are a couple of problems with this however. The first is that Hard Carries are increasingly becoming extinct. Due to the various changes in the game, people are shifting heroes across the board to different roles and building them in ways that make them have a higher impact earlier on. Faceless Void for example – a hero who for years was refereed to as one of the hardest carries in the game – is increasingly being ultra effective when played not as a Hard Carry. To the point where pro players chuck him in the offlane, and have someone else as their 1 position hero. He is still great with lots of position 1 farm and items; people are simply playing differently to make him better earlier as the space for Hard Carries dwindle.

The second problem is that Medusa has received a number of substantial buffs to her early game. Two years ago in 6.74 (TI2) Mana Shield started at an absorption rate of 0.75, and when maxed was increased to 2. Now however, this is an incredible 1.6 to 2.5. Just one level currently gives more than twice as much as an older level 1 of the skill, even giving more than the old level 3. Stone Gaze has also received incredible buffs. Reworked in 6.78, the new skill is now hugely effective in the early game. A previous level 6 Medusa would have a 140 second cooldown on Gaze, now it’s a very useful 90 seconds. Just those two skills alone are now ridiculously more potent earlier on. Her other skills are also equally stronger in the early game, more than they have ever been before.

Why Would I Pick Her?

The issue that arises is that people are building basically the same as they did 2 years ago, when her early game was criminally weak and Hard Carries were much more potent. With dwindling effectiveness of Hard Carries, and a considerably strong Early Game, it doesn’t make sense to me why everyone would still want to tunnel vision the late game. I’ve noticed people like to say things like “if I wanted an X, why wouldn’t I just pick a Y?”, such as “if I wanted a carry to fight early, why wouldn’t I just pick Lifestealer?” People use this X/Y reasoning every year, such as for Support Naga in 2013. In pubs, I tend to find that with no overall consensus on your draft, it often doesn’t matter what you could have picked. You should try your best to make whatever you end up with work well. The point is that Medusa builds are refusing to adapt to change. Everything is changing year after year, but Medusas remain stubborn, and lack any sort of viability innovation. If for whatever reason you want to play Medusa, it make sense to attempt to be the most effective Medusa you can be. And stubbornness is unlikely to let you be this. I mean, a few months ago it was odd to put a “Hard Carry Void” in the offlane, but at TI4 it’s all the rage. IG.YYF did an incredibly potent MoM+BKB rushing Void during this tournament.

I believe this is the most effective way to make Medusa fit into the current meta-game, by understanding that (like Void) she is great when 6 slotted, but very similarly building her in a way where she comes online earlier and is able to be potent with less. She won’t be changed the same as Void – they have different features – but a bit of fresh thinking would do her a universe of good.


The Pro-Scene Pick/Ban Rate of Void and Medusa. Both outcasts until very recently Faceless Void managed to break free with some creativity

Window: The Gaze Predicament

Now that I have explained how Medusa can do much better earlier now than ever before, I want to explain a Key Medusa Concept: The Gaze Predicament. When you use Stone Gaze, you put the enemy in a very precarious predicament. If they look at you, they turn to stone and take added Physical Damage (but less Magical) which also works on Magic Immune units. If they look away, they can’t cast a spell on you nor attack you for the entire 6 second duration of the ultimate. The ultimate also slows both movement and turning, making it harder to do something while you avert your eyes. This is honestly a “Damned if you do, Damned if you don’t” scenario.

This Gaze Predicament creates a very crucial timing window for Medusa; there is a 6 second window where the enemy is guaranteed to be ‘damned’ (The Gaze Window). My concept behind Medusa, is that she needs to make the most of these crucial 6 seconds. I don’t think you should only stack stats so you can just tickle them, you need to be able to substantially wound them during the Gaze Window. Which comes in the form of Damage. I generally find that going something along the lines of Linkens > MKB is the best way to go about things. Linkens gives you a good foundation to exist off, while MKB gives you a massive increase in damage to wound people during the Gaze Window. You don’t want a long drawn out fight where you very slowly widdle away at 5 heroes. You want to do substantial damage in that short Gaze Window, preferable killing someone in it. Void wants to kill someone in a Chronosphere, and I put forth that Medusa also wants to kill someone in a Stone Gaze. This is especially possible with early damage, and against teams lacking in the communication needed to effectively disengage.

How Hero Builds Become The Norm

I don’t think Medusa can follow the same path Void did, but I think she can become more and more like a Weaver. Linkens into Desolator is a very popular Weaver build, and it results in him starting with a good foundation of farm potential and survivability, before receiving a massive burst of damage. Before Desolator became the norm on Weaver, they used to opt for more late game oriented Radiance + tank item like Heart/Vanguard/Hood builds. Most hero trends are because a hero becomes popular in the pro scene, and the pros are building a hero differently than before, so the pubs start doing what the pros do. Pub players continue to play the hero the exact same way, until years later the hero becomes popular in the pro scene again, and pubbers change to this new way. This has happened with heroes such as Racecar Lifestealer and 1 position Gyrocopter among many others. The problem with Medusa is that she hasn’t had her surge in Dota 2 popularity yet, so we haven’t had an innovation shift for her for over 3 years.

But where does this leave us? We know that Medusa is stronger early on than she has ever been before. She no longer has a 140 second cooldown ultimate at level 6, but a much shorter 90 seconds that works a bit differently. While Mana Shield is incredibly stronger early on. I also tend to find that afk farming just doesn’t fit in too well in today’s Dota. Ultimately I hope this article just makes you think about her in a new way, and pushes you towards new way of thinking regarding Medusa. See what you think of a Midusa, or perhaps getting a Drums. She isn’t the same old only-lategame Medusa she was many years ago, so it makes no sense that you should try and build her in the same way. I think getting nice Early-Mid items will do you a world of good, especially since 5-manning when the enemy has to fight into Stone Gaze is incredibly strong.

Hero Win Rate Differences Between Skill Brackets

In this article I have compared the differences in Win Rate difference of a hero in the Normal skill bracket, with that in the Very High skill bracket. Data is from two fortnights from 6.81b in July and August 2014.

Win Rate Range

The first thing I noticed with my data, was that across the board the entire hero pool had a much smaller win rate range in Very High than in Normal. In the Normal bracket, Earth Spirit came last with a terrible win rate of 35%, while Necrophos topped with a 60% win rate. This meant that the win rates for heroes in Normal ranged from 35% to 60%. In Very High however, almost all the heroes that had low win rates in Normal gained a few percentage points, while all the heroes that had high win rates in Normal lost a few percentage points. The Win Rate Range for Very High was 42% to 57%.

Out of the the 30 heroes with the lowest Normal win rates, 28 of them improved upon their win rate in Very High. For the 30 highest win rate heroes, 23 decreased in win rate.

Skill Bracket Winrates

The Top 10 and Bottom 10 Heroes By Normal Win Rate Reveal An Interesting Trend

Biggest Winners

The heroes who gained the most seem to be an interesting collection of heroes, with a strong emphasis on low win rate heroes. Primarily heroes with a high Skill Floor such as Earth/Ember Spirit.

The Top 20 Gainers, With The Top 10 On The Left

The Top 20 Gainers, With The Top 10 On The Left

There were some notable exceptions to the Win Rate Range trend however. Pugna, Enigma, Doom, Skywrath Mage and Faceless Void all improved in win rate, despite not being that low to begin with. Enigma for example went from a solid 51.11% win rate in Normal, to a even better 54.63% win rate in Very High. I doubt much better micro has much to do with Enigma, so perhaps a combination of superior jungle efficiency mixed with effective pushing and the constant pressure of a well placed Black Hole.

Could you argue that a low win rate hero improving is because they have a high skill floor, and if a high win rate hero is improving, they have a low skill floor and a high skill ceiling? Which might explain that while Void is strong in Normal due to the hero being easy to play good enough, as people get even better at placing his Chronosphere he shoots up in win rate. Resulting in Void being an entry level hero that has a lot of room to grow, perhaps akin to Blizzard’s Starcraft design philosophy of “easy to learn, hard to master”.

Biggest Losers

There is a plethora of fascinating data with the heroes who lose Win Rate between brackets.


The Top 20 Losers, With The Top 10 On The Left

There’s a heap of Supports that do much worse in Very High. Over half of the 20 heroes are predominately played as a Support, like Disruptor and Crystal Maiden. I’m honestly not sure what to make of this. A common criticism of low tier Dota is a lack of support heroes, but in Very High where they are more common, some just do worse.

We then get to micro heroes like Beastmaster, Enchantress and Meepo, all of whom lose out. It has been hypothesized in the past that Meepo’s drop is a result of Smurfs skewing the Normal Data. But I disagree with this due to the sheer amount of Normal games, putting forth the idea that it’s probably more likely that Very High players are simply much better at playing against a Meepo. Especially with focusing their attacks so one of the clones dies quickly.

Enchantress has a low 43.69% win rate to begin with, and then drops to 42.09% in Very High which makes her the lowest win rate hero in the bracket. Something is seriously wrong with Enchantresses in Very High for this to occur. Are they trying so hard to smoke gank that they get put behind due to ineffective rotations? Are they being too greedy and their Midas rush is costing their team dearly? Enchantress also happens to have an even lower win rate in ranked VH games, falling another raw percentage point. If you play in Very High, I suggest you leave Bambi on the shelf.

But then what about Beastmaster? Why do Micro heroes like Chen and Visage gain in win rate, while Beastmaster falls off? Do low tier players not think about the hawk vision at all, perhaps causing them to constantly be spotted by it and never kill it? That’s just one possible reason I’ll put forth. It makes sense to me that PA, Sniper and Axe fall off in win rate, but BM is not a hero I would have guessed has a lower VH win rate.

Is It Me or You?

To explain data, there are two main ideas: are noobs really bad with some heroes, or are pros are just really good at dealing with them? There are of course other reasons that can explain it, but these are two reasons people are instantly going to think about. Interpreting Data is a very complex beast, especially when we are using data from tens of millions of games played all over the world. It’s all open to interpretation, and is very susceptible to misinterpretation as a result of a preconceived notions.

On another note, Dotametrics did a similar article two years ago. There are some similarities with the results, like Axe and Sniper being worse in Very High. But there’s also a lot of differences, Beastmaster improved on win rate in his test back then. Whereas now it’s the complete opposite.

Overall, this is definitely a very thought provoking data set.


For non-noob Dota 2 tips and Mechanics, follow @ZotaTips on Twitter.

3 Interesting Pub Trends

I observe a lot of different Pub Trends, but some of them don’t warrant an entire article to themselves. So here is a 5-in-1 interesting stat fueled trends that have been happening over the last few months.

Lightning Slow?

Leshrac’s Lightning Storm has received a bunch of changes over the last few patches, including the addition of a brief slow. However these don’t seem to be paying off as Leshracs who level Lightning are more likely to lose than those who don’t.

The trend is that Leshracs who opt for Attribute Bonus at levels 9, 10, 12, 13 and/or 14 win more than people who get Lightning Storm at that level. Additionally, the Stun + Edict build at the early levels also happens to beat out the Stun + Lightning build. This is very interesting, since a third of Leshracs go Lightning. Cause vs Correlation is a massive argument for Skill Builds (aka, Do people get Lightning more when they are losing?) One factor is that Leshrac has no massive win rate difference between skill brackets, so there’s no false positives from this. Another is that Leshrac builds are determined at Level 2; before most teams have a big advantage.

Statistically one of the better build paths for Leshrac. Note that Level 3/4 are interchangeable.

Statistically a very optimal Leshrac build, with levels 3/4 being mostly interchangeable (light green).

The Lawnmower Leshrac build is incredibly powerful; you press a couple buttons and then run around mowing everyone down with incredible aoe damage. It also allows you to get crucial early game towers. This suggests to me that 6.82 should buff Lightning, since the stats say the skill doesn’t measure up just yet. The recent buffs improved Lightning for the third of people who chose it, but not to the point where they win more than Team Edict. Hmm…. now all we need is a Team Edict t-shirt.

A recent game where I transition into a Semi-Carry after a successful start.

A solid early game thanks to Edict often results in you transitioning into a powerful Semi-Carry

Tusks Are Improving

When Snowball was reworked in 6.81 it effectively raised the Skill Floor of the hero, resulting in a massive drop for his pub win rate to 39%. Luckily this win rate was not forever. Over the weeks the average Tusk player improved as they became more used to the ability. A few minor buffs in 6.81b to Tusk also helped push him along. The trend however is very clear: Tusks are getting ever so slightly better every week.

Tusk had about a 43.6% win rate in June, going to 44.8% in July, and for the week of July 28 – Aug 3, his win rate was 45.1%. Even better, his win rate is higher in the Very High skill bracket. For the week mentioned above he has a 47% win rate in Very High, which is also steadily increasing. His pick rate has remained steady over the last few fortnights, so no shenanigans there.

It snakes up and down, but overall it’s a slow and steady upward inclination. I am very interested to see where the Tusk train will stop, he had about a 47.5% win rate before the rework. Can he overtake it?

Necrophos To Boot

Necrophos has an insanely high pub win rate, but there’s also a wide variety in boot choices for the hero. However what is popular isn’t always the best, so I ran the numbers, ignoring Boots of Speed + Travel as always since the data is too skewed for those options. Boot choice is a great item to see the effectiveness of, since people chose them very early into the game and are less impacted by a hero winning/losing in that game (except Speed/Travels). Radiance has really high win rate because people almost always get Radiance when they are winning (such as on Spectre, Centaur, Bristleback, etc) the 4 main boot options however don’t generally fall to into this trap.


You can see that Power Treads are the predominate boot choice in both Skill Brackets, but decline in Very High as they mostly flow into Phase Boots. I compared the boot’s win rate in that bracket, to the win rate of the hero in that bracket . Interestingly, Necrophos with Arcanes wins more than those with Power Treads. A lot more. Which means I can say that the statistics strongly suggest that Arcane Boots are the best choice for the hero for how he is played in pubs. Which also lead me to saying that whatever role and/or play style Necrophos is playing in games that he gets Arcane Boots in, is the statistically superior way to play Necrophos.

Necrophos uses a lot of mana and being able to safely have enough mana to use your abilities is vital. People underestimate how much mana he uses, especially with a fast Mekanism. Relying on Sadist is often an unreliable option, perhaps even causing you to Scythe someone for Sadist, and not for the skill itself. Which is more important now that Scythe adds +30% respawn timer. I assume that Necrophos with Arcanes are less likely to get levels of Sadist early on, which is important as Skill Builds that skip early levels of Sadist also have a slight win rate edge. Like with Pugna, Arcane Boots also help your entire team with Deathballing, which is when a Necrophos level advantage and constant heals come in handy. These few issues are my two cents on the problem with treads as evident by their winrate.