Leveraging on the Deck(Data) Tracker

Multiplayer Discussion
1

I realize some people may have some deck tracker installed but uses it mainly as a tool just for remembering the cards left in the deck/ cards being played by opponent.

It have another function that collects game data of the past games we played, and I will like to share how i use the data for my games. (Improving my rank)

Contributions

If anyone have any contributions, do feel free to add from your personal knowledge and experience, or via a link.

example:
https://tempostorm.com/articles/four-steps-to-understanding-hearthstone-data-to-improve-your-game

Thanks in Advance.

(Work in progress)
What affects my rank position?

I will briefly explain the 4 aspects, and attach more detailed explanations if I find any. In summary, a combination of the 4 aspects will determine your rank position. (in no order of importance)

Deck Strength

Do not be mistaken deck strength refer to the tierX, netdecks, etc reference.
Deck Strength refers to how strong your deck is with reference to the match-ups you are facing.

Skill

Refers to player’s proficiency with his/her deck, knowledge about the game, and management of other intangible aspects.

Time

Obvious reference to co-relations between numbers of games playable vs time available.

RNG

Since there is heavy RNG elements build throughout the game, individual’s performance will be affected to a certain degree due to RNG. While certain aspect of RNG are unavoidable, it is still important to understand and manage it to the player’s benefit.

After understanding the above, we will realize that there are controllable and uncontrollable aspects where will determine our rank.
Here, we want to work on the more controllable aspect, via data collection and analysis.

Note 1

Based on 1 deck, I would recommend at least 10 games of serious play (no auto-concede, distracted plays, etc) as a bare minimal for any useful data building. (more game will improve the accuracy for further analysis)

Win Rate (WR)
WR is the % representation of the numbers of games that an individual is winning. It is generally normal to see decreasing WR as you progress higher up the ladder, assuming all others aspect being constant.
Below R5: Due to the win-bonus for consecutive wins, it is still possible to progress up the ladder even when at a negative WR, subjected to numbers of games played.

Why WR matters?

Better WR saves time(overall). It is also an indicator of how well an individual is playing the deck.

What does it reveal?

If we breakdown the overall WR into classes, we can have a better picture of how well we are performing vs each classes. From here, we can better understand if we need more work on which match-ups.

WR can also be use as an indicator to determine if certain card is suitable/effective.

How to improve WR?
As stated earlier, we need a minimal number of games to have a relevant data set for analysis. The greater the number of games recorded, the more reliable the analysis can be.
Note: Most data need to be viewed at a larger scope.

Overall winrate

I view WR with a range. (e.g.50-55% or 50-60%)
Depending on which end of the ladder we are referring to, the WR in discussion plays different significance.
I also associate WR with consistency. If we look at the past 3 season’s record. Do we see a consistent WR across all 3 months.
If yes, then it could be we are very proficient with the deck archetype* and it;s execution. Also note, if we are able to maintain a constant WR, and we see progress on rank, then it is still a positive.
If it is inconsistent, then we need to look at the factors affecting it. e.g. Did we switch archetype*

WR vs Mirror matches

I feel this is a very important match-up to work on. Assuming both players have the same deck. The deciding factor will be down to RNG and individual skill. Since, we are not able to have any control on RNG, then the player with the better skill, would win more over a number of games.

WR vs other classes

Rarely will a deck be absolutely favored vs All deck types. Thus, the chosen deck preferably should capitalize on having as much favored against match-ups / lesser not favored against decks.

When we look at the match-up distributions, we can have a better picture to formulate a plan.

WR (Favored against) if we see a slightly positive WR, we might also want to work on practicing vs such match-up to increase our WR. The more we capitalize on the match-up, the more it contributes to improving the WR.

WR (Not favored against) similar in concept as above, but we want to limit our losses as much as possible. However, depending on the decks we play, it may not be beneficial to focus too much energy here for minimal result/return.

Pitfalls
Wrong application of WR data
Data are just numbers by itself. if wrongly used, will lead to undesired results.
This is a big topic and I won’t touch in detail atm.
To avoid, I would recommend discussing your data and analysis method with a few players (preferable higher rank players)

Considerations

Starting With Coin/Without Coin

Depending on how your deck is built, the above scenario may have a bigger impact on your WR. I assume slower decks would benefit more from having the coin, but after viewing through my game data, it proves otherwise.

Since, I may tinker with deck built (or home-brew) I am still exploring if this data can be use to indicate a need for more balancing (more consistent results)

Tech choices/ Card replacement

I also have my own preferences and limitations. e.g. I have never learnt to use Hunter’s Tracking effectively. In my games, I find that by replacing cards that I have more confidence and proficiency with, could help me more.
(this is also a big topic, so will cover briefly atm)

In general, the tech choices should fit the win condition of the deck I am using. I also use the match-ups I am facing as a reference.

e.g. The current mech hunter lists does not contain Silence.
However, I can see Silence being effective with a majority of my matches,
Mech Hunter/pally - (Magnetic) a lesser extend to warrior with magnetize Zillanx,
Rogue (edwin), Mage (doomsayers, freeze)
It act as a partial removal, or enabler for an extended dmg reach over taunts

How to start
There are quite alot being covered above. I would recommend starting simple. Choose 1 area to focus on and once you have achieve a consistent result, then move to the next.
Once you focus on the right thing, the result will automatically follow.
At any time, you feel being stuck. Feel free to ask, as there are many great players here all willing to help.

till then…

Good Luck Have Fun

4 Likes
2

The data is very interesting,i have thought of a lot of additional data that could be provided and used for better analyses. Specially the mulligan data could be improved upon a lot by taking into account possible synergys between cards. To analyze that you would need mulligan data not only for a single card,but also for that card in combination with any other single card and any other single 2 cards. It would be possible to calculate the perfect mulligan based on that data,showing you which cards to toss in any situation. (tossing a card needs to have a +50% value with giving you back a random card,but no one ever made these calculations as far as I know)

You have to think about it very well, how to use the data and what the data actually means and its very interesting and also kinda fun to think about which data could be used for what. In the end though I have come to the conclusion that the sample seize provided by most data trackers is (way) to small to make solid conclusions. Its good for obvious conclusions,where the difference is very big but it is kinda bad at less obvious conclusions where the differences are small and where you would need a lot of data to reliably tell that one option is very slightly better then a different option.

If anyone has specific questions on which data would be good for what analyzis then feel free to ask and I will try give a few suggestions. The use of data in hs is still in its very early development,lots more is possible but the small sample seize will remain a problem.

1 Like
3

I am only working on the free functions available to anyone.
Is mulligan data that you are referring to, via the site reports or paid functions?
If, not where do we asses this mulligan data?

I do know some sites provides card-by-card data on it’s contribution to win-rates via various sort-able options such as mulligan on hand , etc.
I do appreciate if anyone are interested to contribute articles in this area. However, I will also not touch on this, as I prefer to let individual’s analyze thier own data for more personal needs.

1 Like
4

Hi,
I have my 1st draft up.
Appreciate any contributions, feed-backs, views, opinions and/or suggestions.

1 Like
5

I think this mulligan data has never been collected as far as I know. They only look at single cards.
My suggestion (mostly for people who work on ai implementations or theoretical optimal play calculations) would be to also look at combinations of cards (it would be a big sheet, 15 cards (no singleton) each in combination with another 15 cards and then those also with another 15 cards. There will be duplications but it will still be a lot of different combinations that you have to look at ,maybe 3000 combinations in that order of magnitude (against which a sample seize of say 10k games would clearly be way to small). The data should be easy to collect and it would in the end show very small differences (for which a lot of data would be needed to show those small differences reliable). It could however show exactly what cards to throw and when.

It could show interesting results early on though,maybe show cards that bite eachoter or that work very well together. Maybe there is 2 cards that each have 50%+ when looked at them as a stand alone but who will drop slightly below 50% if you keep both in opening hand. Or cards that are less then 50% stand alone but go above 50% when kept together,things like that.

6

I looked at my stats and noticed I suck worse than I thought I did and no idea how or why. I dropped back to 10 and have been conceding to every mage portrait and playing a budget version of dragon mind blast priest.

I honestly think I’m a lost cause.

7

My contents above could be quite heavy for most readers.

I would suggest starting from your overall WR.
Track your overall WR on a weekly basis. Do note that the variables must be the same, e.g. WR for your dragon priest with minimum 10 games of serious play.
Assuming you are playing say 10-20 games, each week, you can see if your WR is increasing or decreasing.
If your WR is increasing, then your rank should also be raising. Note also decreasing WR does not mean a bad sign*, if you are progressing up the ranks. It could mean you have not gain any new knowledge/skill for your deck.

Also, I mention about the 3 aspect that will affect your rank. Thus, another possibility is your deck strength is limiting your progress as well.

If above is too complicated. I suggest you watch other players play your deck, and have discussion on moves that you might play differently from him/her.

8

With a few substitutions, the deck I play is from a streamer that habitually hits 5 to legend with it: however, I had to sub the legendaries.

Lykotic swears by Even Shaman but I had 23-29 in the stats thing: more losses than wins and it was pretty much similar with all my decks. I deleted all my decks and started over with my dragon priest and classic midrange hunter.

9

I could go deeper explaining but it will just get too boring.

The lower your WR, means the more time + games you need to actually progress in rank (below R5) that can be over 100 of games.

Since you decide to focus on 2 decks which is a good start, then you can also apply the same concept on determining which deck suit ladder climbing more.

Note: You may also need to differentiate between Fun deck from Ladder Deck.*

10

My dragon priest IS fun. All Hunter decks are fun but I already have golden portrait and over 600 wins with Hunter.

I was hoping to climb with a different class. And I would actually appreciate advice on how to use the data provided.

I’m not a data analyst by any means (not even as a hobby).

For example, lets say Even Shaman has a 54% win rate but when I use the deck, I have a 45% win rate. Obviously, the problem is me.

But what if I practice with Even Shaman, and I manage a 52% win rate. So where is the other 2%?

Also, if players want to use data, that is fine but for accurate data analysis, you would have to do it like this:

Date:
Time:
Deck:
Server:

I believe that would give us the most accurate data available.

You might be seeing tons of aggro, but I’m seeing tons of control…but we’re on different servers.

11

Dont forget that this is overall % , a lot of people play the deck that has no clue what are they doing and thats why the win rate % drops , some people can have 85% win rate others can have 30%.
you need just practice and you will achieve higher win rate c:
for example a lot of sites giving rogue class 47% win rate and being 7th class … but 2 months in a row im achieving legend with it.^^

12

Sorry I do not get what you meant, (thus I make some assumptions)

I assume the 54% WR is based of some report. If your WR is 45%, the difference is 9%. If based on 10 game, the difference is 1 game! If based off 100 games, the difference is 9 games.
Thus if the sample quantity is too small, the WR comparison makes no sense.
Using a 100 game as a sample, of that 9 games, what are the details that sets you apart from the general player vs yourself?

https://tempostorm.com/articles/four-steps-to-understanding-hearthstone-data-to-improve-your-game
See point 3. Use data to understand your deficiencies in play

I believe this data will be not very useful from a personal point of view, but much better suited for data collecting services/website.

13

Some objective criticism:
Where is the data unanimously confirming, that always going face increases your average winrate?

Without this, you may as well stop considering the importance of data.
One needs to be aware of the basic data discovered, up until now.

14

Sorry I did not get the message, Can you rephrase?

15

If you are not aware of the very basic discoveries, concerning the data that confirms Smorc is best, discussing data to help others in a practical manner is dishonest.
It’s like not understanding that going second increases or decreases the winrate of certain decks.
Only blasphemous and truly indefensible.

16

Sorry, I truely and sincerely do not understand what you meant.

17

I was using it as an example. I also posted a new thread with my questions for more information on how to understand this stuff.

Personally, I find it boring and I think it kills HS for me. I like feeling - I like the feeling of scoring a win. I like the feeling of losing a fair match. Instead of being cranky all the time because some number says I’m too stupid to play a deck well.

It’s a huge letdown to make all this progress only to discover you’ve made none.

18

There is some part of the game that will be boring, especially when competitiveness is involved.

Many people opt for Fun as a priority which is absolutely fine as well.

Driving a car can be a fun experience, But, before one can drive a car, he/she needs to get through the boring part to obtain the license. (theory lessons, tests, etc)
One can also ignore getting the license, and start to drive a car for fun, but there are consequences.

There will be better people out there that have better knowledge and experience. Thus, I fully support you to continue asking for help.

19

I gotta upvote OP for great usage of collapsible sections.