Standard Data Deep Dive Deep Dive

In the last episode of the podcast Cracker and I reviewed some data from the previous few weeks of Standard in an attempt to understand which deck or decks would be the best choice for grinding the ladder on Arena.

Before we get to the data and ultimately our recommendation, I would like to take you through the data available, how I arrived at the dataset and the criteria I used to arrive at my conclusion.

Starting with the data source, I payed the few dollars to upgrade untapped.gg client to premium, unlocking the tool to give me the details of more than just aggregate win percentage. The tool gives detailed match up statistics, version information and total number of matches played.

With this rich data source now unlocked, I set about discerning a useful and realistic dataset that would form an accurate representation of the meta.

Quick tangent on this, not just because its me, but I feel it important to qualify this is it is a critical component to my analysis. For more than ten years, I have worked as a Business Analyst and Project Manager in the healthcare industry, where my role is to use data to identify areas of patient care, hospital management and revenue that could be improved. Or provide analysis and advice on a decision being made by heads of unit and executive management.

I then form projects and implement these initiatives.

What I have done here is not dissimilar in my approach, although the subject matter is vastly different.

The Dataset

Given there are millions of games played on Arena each month, I needed to focus in on the most relevant data, as well as the data with enough sample size to represent a true reflection of deck performance and match up history. For example, a deck with 100% win rate may have only played 4 games in Bronze, so would vastly skew the numbers.

Here is a summary of the criteria for making it to my dataset

• Standard Best of One queue. I would have preferred Bo3 but the numbers truly demonstrate that Bo1 is hugely popular and I wanted to provide data that was as relevant to as many people as possible.
• Matches played from Gold to Mythic rank. I figured the more casual players in Bronze and Silver are there to play their own decks. This is not a criticism of them, but it is not as competitive as the more solved mid and high tier ladder meta.
• I took the decks with the highest win percentage
• Minimum of 200 matches played against at least half of the other top decks. This gave a rounded reflection of favoured and unfavoured matchups.
• This gave me a sample size of 489,000 matches to work with
• Not reflected in this dataset is play/draw or mulligans

The data was surprising in a number of ways, with a couple of standout decks with a win percentage approaching 60%. The next 10 decks were separated by under 3% aggregate win percentage.

In fact, the differential between best and worst performing decks in this dataset is only 7.5% aggregate. I believe this is a good indicator for a healthy and balanced meta.

For me though, I was not satisfied with only using a single data point for my analysis. If we based our deck decision on aggregate win/loss only, we are taking a significant risk that we are burning wildcards.

The decks

So, after arriving at my dataset, I extracted based on this criteria and was left with the following 12 decks (in descending order of win %):

Here is how they match up against one another. Note that not all decks met the 200 match criteria, so I have left that blank, rather than show data that may not be reflective of the matchup.

Here is my findings. As I mentioned above, selecting a deck based solely on win percentage is dangerous. Look at Gruul Werewolves for example, while it maintains a decent 52% win rate, it only has a single positive match up against the other top decks.

So, now we have our criteria, our match win % and our match up against the other top decks. That should be enough to base a decision on right? Not for me! I wanted to understand a little more. Specifically how represented each deck is and how much frustration at the matchup lottery would play in my ability to pilot a deck enough to climb the Arena ladder.

To do this, I looked at total matches played by each deck, the positive and negative match data as well as the differential between the best and worst matchups

So, it is clear that the Naya Humans deck is a cut above the rest. This isnt surprising with a win rate so much higher than the field. But, I wanted to give a recommendation based on the whole field. Obviously I am not going steer you away from the Naya deck, it is very good. But I would be remis if I didnt provide a viable alternative recommendation.

So looking at the data we have, it is clear that the GW Enchantment deck vastly more popular than any other deck in the field, with Mono White a decent few percentage points back in second.

With the popularity of the Selesnya deck, I want something that will do well against that as it will be my opponent almost one in every four games. The clear standout here is the Azorious Control deck, flexing its massive 62% win rate against the GW deck.

Looking a little deeper, you find that is has an almost break even win rate against the best deck (Naya) and a positive matchup against the next most popular, and best performing deck in Mono White. The same is true for its Orzhov Control matchup.

This positions UW Control with a positive match up against more than 53% of the field without making a massive sacrifice against the best deck (48% win rate) as Naya is less than 10% of the meta.

Looking at its poor match ups, the Izzet and Mono Red decks, while downright awful for the control deck, make up such a low percentage of the meta at 5.6%, I feel you can essentially ignore or write off those matches and happily farm the rest of the field.

So there it is. My recommendation is to play the Azorious Control deck in Bo1 ladder. Happy grinding!