On the Table, By the Numbers


Guest Article by Carlo Connor-Hill from asoiaf-stats.com

Who and What?

(Editor’s Note: This article was written using pre 1.5 data, so some of the information might be slightly out-of-date at the current time. Additionally, don’t miss the Faction Starter Set Giveaway opportunity at the end!)

I have recently gone live with my community project to provide local, worldwide and faction rankings for ASOIAF at https://asoiaf-stats.com/. The site remains a work in progress, but the more people use it, the more I will work on it! The ranking system has been in development over the past 3 months with a number of trials run across 1.3 and 1.4 in my local area and internationally thanks to the participation of some Youtube personalities you might be familiar with. For those who don’t know me, which is entirely forgivable, I play in Northeast England where the tournament scene is still in its infancy and have only recently begun to interact with others on the main Facebook groups. I am however a professional data scientist, and I am a complete stats nerd, so hopefully you can trust that what I have made not only works but is a very robust system!

Asoiaf-stats.com is also offering a Faction Starter of your choice to one user at the end of February chosen from the ranked submission history. So if you want a chance to win a starter set get submitting ranked data!

In preparation for the release of 1.5, I have been collecting data for the last 4 weeks of 1.4 across 3 separate gaming metas in the UK, USA, and Singapore with close to 100 game submissions. The players range from the CMON invitational top table down to more casual local tournament entrants. The data comes from a mixture of tournament games, with each meta submitting the results of at least one full tournament, as well as practice matches played in preparation for those tournaments. Which brings the question to my mind, does anyone NOT play the game under 2 list tournament format at their casual club night games?

When each game is submitted a modified ELO rating system is used to calculate the probability of each player (and faction) of winning with 2 equally ranked players sharing a 50-50 win probability. This is then compared to the actual result, including the margin of victory, and the rankings of each of the players and factions played is updated. As many players play the same faction it is a relatively easy process to begin to rank the factions irrespective of the ability of the players that are playing them.

So let’s take a look at the data I have collected over the past 4 weeks, hopefully, it can give us some insight into the balance changes made for 1.5 and the kind of insights data collection can give us. Bear in mind that this data is from the small testing pool of thirty players I used to develop the site, the current userbase is more than three times that size and still growing, so the accuracy of my statistics are only going to improve.

What are people actually playing?

People may be surprised to find that NW are not as popular as you might expect considering all the hearsay about them being the best faction. What I find interesting about this chart is that faction popularity strictly follows the number of units available for each faction. I think this is a strong indication of people’s preference to play factions with a large unit and hero pool, or maybe it just shows that NW starter sets are hard to find in the EU and everyone would be playing them if they could get their hands on them!

But how successful are these factions?

Factions are balanced around a 0 rating, + ranking points given by wins and – ranking points in every loss, the exact points won and lost are calculated by the Elo (https://en.wikipedia.org/wiki/Elo_rating_system)  probability of the outcome and the actual VPs scored.

The Good

Across all the data we can see that the NW were the highest rated faction (this will probably surprise few people), but not significantly above Lannister and Free Folk in actual rating (possibly more surprising to many readers). They did however sit a clear margin above the other factions in Win Ratio (an indication that although they win more often their margin of victory in these games is often low). For a competitive game, anything with a win percentage outside of 40-60% is not well balanced and needs to be corrected (see how League of Legends looks to achieve this across 230 separate champion roles https://champion.gg/statistics/#?sortBy=general.winPercent&order=descend).

The Bad

Baratheons and Neutral both have low amounts of play, and are difficult to judge at this stage. Starks, however have collected plenty of data and look to have been significantly underperforming in 1.4! Arguments can be made that they possess some strong units and NCUs, but across a reasonable size data set they have simply not been winning games with any reliability. I have my own personal theories as to why but we will stick to the data itself for today.

The Ugly

So we have looked at the factions and their win ratios, but that is just scratching the surface of both the insights and the levels of balance we can hope for in the game design. Let’s take a look at the commanders, this is where we start to really see some issues. (I have removed all commanders with less than 5 data submissions as a baseline for calculating a reasonable win %) 

Tywin; 100% win ratio! But take this with a huge grain of salt as this entire statistic comes from not only a single tournament in 1 of the 3 metas but also from a single player (who is top-rated in that region). Is he strong? Definitely. Is he too strong? Only more data and more players can say so.

Jon; 73.9% win ratio… Here, in my opinion, is the biggest issue in the entire data set. This win ratio was achieved across 23 games, with many different players using him, and also includes a number of Jon vs Jon matchups, meaning a not insignificant number of his 6 recorded losses come from mirror matches! Even worse than his ludicrous win ratio is that not a single other NW commander sees enough play to make it onto the list. He is simply the obvious NW choice and even when not used as a commander he is almost exclusively used as an attachment instead. Sworn Brothers and Bowen Marsh have both taken a hit in the 1.5 updates, but in my opinion, they have been unreasonably scapegoated with the real balance issue resting with Jon (and Ghost).

Below this, we have a good chunk of commanders residing within the 40-60% range with Tormund reaching the heady heights of 62.5%, a single win above 50%, but he does look to be one of the best FF commanders available during 1.4. Robb scrapes in at 40% win ratio, and yet is the highest rated Stark commander to see significant play…

I think some people may be very surprised to find Tyrion sitting at 20% win ratio. From my personal experience, I would say; that while his control style may be very frustrating to play against it does not necessarily convert into wins. 

Generally, the weaker commanders simply see little play, but oh boy, is Brynden (The Blackfish version) bringing up the rear! I actually don’t know what to say here other than I don’t know what is wrong with him, or why anyone is playing him?

A final word on NCU vs on-field Commander; I believe the choice of an attachment commander vs NCU commander is largely dictated by the game mode being played, with a number of game modes relatively heavily favoring one or the other. I always include both in my 2 list format. But they look generally well balanced between the 2 when you sum across each type, Roose sees some play across all factions, particularly those who either don’t have an NCU commander or don’t favor their own faction option. But the High Sparrow looks to be by far the most popular NCU choice, so let’s see whether the revamped Warrior’s Sons and Champion of the Faith swing that even higher.

1.5 Predictions

There are many great articles and in-depth analyses on the changes made by 1.5, so I won’t go too deep here. When you look at these 1.4 results however you see reasonably clearly why NW took a hit. As I previously mentioned, I believe they have targeted the wrong culprit in Sworn Brothers, but I can see that it is a much easier fix than a change to Jon. 

I think that the Starks needed a bump and only time will tell if they have received it, they had some of the lowest internal balance in terms of units and NCUs, with some units such as Greataxes and Sansa outshining all other competition in their points range but not enough other good options to make fully rounded lists that fare well across the full spectrum of game modes and possible match-ups. 

Lannister improvements see a number of its least used units being brought more in line with the other faction options, an already high-performing faction given more viable choices is always going to be a tough opponent.

Free Folk saw a couple of stealth changes in both the charge and panic mechanics, and they may struggle to find good spacing to achieve the charges a horde list relies upon. But they are the biggest beneficiaries from panic damage being limited to D3+1. I believe them to be, in the hands of an experienced FF commander, the current faction to beat!

Baratheons remain low on unit choices, as will Targaryens upon full retail release, and this will limit how much play they see and how competitive they remain in the short term. Neutrals start to see their faction being rounded out with Heroes 2 and Stormcrows providing significantly more choices. The changes to Varys have hit all factions (barring FF) equally as he saw competitive use across the board. For Flayed Men only time will tell how the change affects the meta. I think the change from 10 to 9 points opens up more list building options, which, as I keep mentioning, is the key to success.

Overall I think the changes are very fair, and are backed up closely by the data. I understand why some NW players are unhappy, not because they are worse (because they have been shown to be the number 1 faction), but because I think the changes made restrict faction choice rather than open it up. But I am not a NW player, so maybe I am wrong?

The Future

With close to 100 games submitted by a player pool of around 30 players in 3 separate metas we can start to put together some top level statistics. To get down to the really interesting stuff however requires a significantly larger data set, an order of magnitude more results to be exact!

Ever wondered which faction is best suited to each of the game modes? 

Want to know what your odds of winning a particular faction match up is? 

Or do you just think the results don’t represent what you are seeing in your local area?

Well, the only thing to do is sign up and start submitting results! Sign up https://asoiaf-stats.com/, get your local group to sign up, find out how you compare to the world at large and maybe even influence the future balance of the game. To help sweeten the deal, http://asoiaf-stats.com is also offering a Faction Starter of your choice to one user at the end of February chosen from the ranked submission history. So if you want a chance to win a starter set get submitting ranked data!

Knowledge is power, and the more submissions we get, the more we can determine who is the true power behind the throne!



Please enter your comment!
Please enter your name here