Why Skill Matters in Hearthstone

Carbonrod and FreeHornPub are back to present another article and our choice of topic for this week was one that players debate and make jokes about constantly; the relevance of skill within Hearthstone. By no means can we assure perfect play throughout due to fatigue however we strived to come as close as possible. The […]

Carbonrod and FreeHornPub are back to present another article and our choice of topic for this week was one that players debate and make jokes about constantly; the relevance of skill within Hearthstone.

By no means can we assure perfect play throughout due to fatigue however we strived to come as close as possible.

The data is available over here.


We are in the camp that believes that skill is measurable and observable within Hearthstone, and that it has a verifiable impact upon win percentage when piloting a given deck.

Sounds obvious right? Well it is not uncommon to hear how “RNG” this game is, and that there is less and less within the control of the players. So, we decided to put it to the test.

The entire article would be written in the point of view of FreeHornPub.


In order to accurately measure the relevance of skill we decided to play a few games. Well, not quite a few, more like 100. These 100 games were all streamed at twitch.tv/freehornpub for anyone who would like to check them out.

These games were specifically, of the Freeze Mage mirror. We chose Freeze Mage as our deck to pilot during these games for a few reasons. Both of us ran a similar standard list, which can be found in this article.

  1. The deck is considered to be among the hardest deck to fully master in the game, moreover the mirror match is heavily knowledge and skill based according to just about anyone you ask, whilst being an uncommon sight on the ladder.
  2. One of us (Freehornpub) is an expert on Freeze Mage, racking in over 1300 games of Freeze, having extensively studied and played the deck, whilst the other (Carbonrod), is a legend viable Freeze Mage player, with over 500 games of Freeze Mage, having piloted it to legend twice in the past but not fully aware of the nuances of the deck.
  3. These decks are just about as far away from RNG focused as you can get, almost none of the cards in the decks have random effects barring Scientist (which is controllable to a degree), and both decks cycle heavily during the match up, drawing on average 20-25 cards of their deck before the game is decided.

Due to these reasons, it made perfect sense to play this deck for our experiment.

Inferences Drawn From Data

As observed from the data, the initial 10 games were slightly beyond what could be called variance with 7 games going to FreeHornPub, however the last 10 games, fell within variance with only 4 going to FreeHornPub

It can be clearly seen that skill does indeed play a major aspect of this game. However, the difference in skill required to ensure a significant gap in the win rate can be overcome in approximately 50 games, should there be an ideal practise partner who demonstrates the correct line of play consistently.

Initial Divergence In Play

During the initial aspects of the testing, there were many notable aspects of play which noticeably were sub optimal. 10 of which will be listed, with the reasoning behind them, and why they were sub optimal.

  1. In a board in which both sides has mad scientist, as the player going first, on turn 3, pinging face was opt for over pinging the opponent’s Mad Scientist, and running scientist to the face.This line is strictly inferior to pinging the opponent’s scientist and going face with your own scientist. This is so as the opponent would be forced to run in their scientist to your own, or have their scientist cleared by the hero power the next turn while always being behind by at least 1 dmg, when considering the best line of play for the opponent was to trade in the 2/1 scientist while developing something else.
  1. Dropping minions over pinging when both sides have secrets developed in the early-mid aspect of the game before the mana has been ramped up for other more optimal plays.This line is strictly inferior as there is no pressure to develop the minions, as they would not be able to attack the opponent’s health total due to the possibility of the secret being an Ice Barrier. As such, damage to the opponent’s life total would be needlessly lost.
  1. Developing loot hoarder or mad scientist against a board which consist of a single loot hoarder while going 2nd over pinging down the enemy’s loot hoarder.This line of play is inferior as the opponent can go face with the Loot Hoarder and develop a secret, rendering you unable to go to his face, netting him a damage lead. However, should the loot hoarder be pinged, both side should be about even in terms of damage more often than not.
  1. Playing Blizzard on turn 7 to discard cards over using Flamestrike.This line of play is a mistake because on turn 8, it is unlikely that the player would be drawing cards due to the fact he just discarded a single card. As such, on turn 8, a single damage from ping would be lost from Flamestrike on turn 8 as compared to casting Blizzard and weaving a ping in.
  1. Opting to start discarding cards to ensure the hand has 9 cards rather than 8 when the turn ends when there is a possibility of choosing to discard useless cards in the mirror.Aiming for 9 cards during the end of the turn means that the next turn, there would be 10 cards in hand. This is bad as this severely hinders any option to draw any cards without milling self. However, with 8 cards, there would be 9 cards in hand the next turn, allowing for Arcane Intellect to be played followed by burning a garbage card or dropping Loot Hoarders without fearing mill.
  1. Dropping Acolyte of Pain and pinging it on an empty board with 7 cards in hand.This is a misplay because of Blizzard + ping or ping + frostbolt or any other such combinations as it would result in milling by 1. However should it not be pinged, as the mage does not have any means of inflicting a single damage from sources other than the hero power, milling of self is hence impossible.
  1. Dropping Emperor Thaurissan without hitting certain combo pieces (Frostbolt+Ice Lance).This is wrong as if you do not get at least 2 Fireball from archmage while drawing natural burns, or 3 Fireballs with lesser natural burns; you will lose to Antique Healbot and a defensive Alexstrasza from simply not having enough damage to lethal the opponent.
  1. Throwing burns at the opponent’s large bodies such as Archmage Antonidas/Emperor Thaurissan after the Fireball cannot be made from one’s own Archmage, without consideration for whether there is sufficient burns for lethal through a Healbot or Alexstrasza. 

    This can be punished should the possibility of a defensive Alexstrasza in conjunction with a Healbot exist, resulting in insufficient burns, shutting out the game.

  1. Using both Doomsayers without it achieving anything but freeing up a slot in the hand.This is wrong as it would give up the possibility of setting up an empty board Emperor/Archmage/Alexstrasza to be dropped on, more often than not, forcing the opponent to throw a component of their burn, usually fireball, to clear the body, resulting in them requiring a larger number of fireballs to be generated from their own Archmage for sufficient burns for lethal.
  1. Intentionally holding Alexstrasza in the hand for extended periods of time, despite opportunities for massive damage from Alexstrasza to the enemy’s health total just for a defensive play, when the opponent knows how many Fireballs he needs to generate from Archmage to counteract it. (which is generally 3-4) 

    This is wrong as Archmage Fireball generation combo basically renders this useless, it is only correct if the enemy does not understand how many fireballs needs to be generated from Archamge to counteract it, which assumes the opponent to be bad (not wise). This also gives up a lot of pressure you can exert to remain ahead in the lethal count, because double iceblock plays a massive part of this matchup.

Change In Line Of Play Over Time

Similar to how Skynet’s ability rapid learning abilities led to the downfall of humanity in the Terminator series, CarbonRod’s rapid ability at grasping the matchup, highly facilitated from imitation of my movement set too brought about an end to the towering win rate, reducing it to what’s effectively 50%. In this section, I will attempt to detail the changes to his line of play, and roughly when it occurs so it would be easily tied into the change in win rate to observe its impact.

The numbering ties directly to the numbering of the above section details of the different lines of play.

  1. Adopted at approximately game 70
  2. Adopted at approximately game 80
  3. Never adopted.
  4. Adopted at approximately game 55
  5. Adopted at approximately game 60
  6. Adopted at approximately game 40
  7. Adopted at approximately game 15
  8. Adopted at approximately game 65
  9. Not consistently employed though shown sporadically from game 60 onwards
  10. Adopted at approximately game 20

Different Lines Of Play One Might Expect

There are many alternative line of play which appeared in regards to this matchup throughout the history of hearthstone. Some advocated by pros, some advocated by no name players on twitch chat. A few of prominence would be attempted to be listed below.

An example of a line of play in which was advocated in the past would be coining Alexstrasza, in which Firebat once commentated was what he believed to give an advantage to the player going second most of the time. This however can be beaten by simply using Healbot in conjunction with Archmage over the next two turns.

Another example of which was once put forth by Forsen was to always use Alexstrasza defensively. While this is indeed correct when playing against another Freeze Mage who does not attempt to fish for any Fireballs from Archmage for the goal of using Alexstrasza defensively is to force the opponent to run out of burns in the deck, hence lack the burns required to enact lethal. However, it can be beaten. To beat is, all that is required is for Archmage to fish for 3 fireballs to overcome the heals from the opponent.


As Carbonrod’s grasp on the finesse and skill required to play the matchup increased, he began to and successfully completed acquiring the vast majority of the knowledge required to play the match up adequately. As this occurred, his confidence and ability to independently take correct lines of play increased – this increased his win percentage thusly.

It’s worth noting that whilst Carbonrod had achieved Legend piloting Freeze Mage, he had never played the mirror more than a handful of times prior to this experiment, and as such was greatly inexperienced initially. This inexperience translated to a poor display of skill and consequently a poor win rate. Though imitation of superior lines of play, he began to grasp the logic behind these lines, and his personal skill rose. This imitation, and the obtaining of the knowledge behind the lines of play, increased both his knowledge base, thus altering his observable skill, but also altered his observable win percentage.

The conclusion that can be drawn from this occurrence is as expected – skill observably impacts win percentages. Whilst we only examined this specific matchup, we have provided an extensive amount of data and reasoning to support this conclusion. As such, we believe this experiment offers a great deal of support to the notion that skill directly correlates to win percentage in Hearthstone, and as such, mastery of a deck – regardless of its strength, will offer great results than playing a deck you do not understand as well.

So what can we take away from this? Quite simply, you will statistically have better odds and a better win percentage piloting a deck you are good at, than with a deck that is good, accounting for meta conditions and assuming an average quality of deck.

If you’ve got a deck you love, and it’s not tier 4, you should research and focus on mastering it; you’ll see rewards and you will climb the ladder by doing so.

Until next time, this is Carbonrod and FreeHornPub signing off.

About the Authors

CarbonRod is a multi-time legend, having achieved top 100 legend multiple times. He is an expert of the rogue class, having played over 3 thousand games of rogue, knowing all of its intricacies, matchups and variants.

FreeHornPub is a player dedicated to Freeze Mage, having played over 1300 games of Freeze Mage, having researched all of its relevant matchups, and math its possibilities, knowing all of its intricacies, matchups and variants. He streams over at twitch.tv/freehornpub

We’re both currently looking for sponsorship or any job in this industry of reasonable terms.

About the author