I've Built a Chess Rating System for Warhammer

In my article about why I built Siege Ladder, I mentioned about being inspired by the likes of Lichess and Chess.com, this is no more apparent than with my decision to build in an Elo Ranking System into the games you play and record here.

For those unfamiliar with Elo, it's a mathematical rating system that provides you points for every game you play, irrespective of the tournament/event you are competing in. Originating in Chess, it can actually be easily applied to any "zero-sum game", i.e. a 1v1 game system where there is a clear win/draw/loss outcome. 

I found myself attending tournaments in the Great British Hobbit League with no idea of whom I about to face. Has this person played 100+ games, or is this their first tournament? I find it very valuable to know what I'm up against, and that was a driving force behind building this into Siege Ladder.

I could have looked up their GBHL ranking in the two minutes spare before a game, but then that isn't the full story, not everybody attends that particular league religiously, I don't myself. I could be playing 3 x games every week in my local club, or just be in town for a weekend and be the top player in the national league in my own country, that all matters too.

In the past I have spent a lot of time playing Chess via Lichess, particular during the Covid years where I used it as a way of interacting with friends that I couldn't see in person. I was also fortunate enough to work with Chess24 (before they were acquired by chess.com) in my time at IBM, so I had a fair understanding of that world, and could easily see how that could translate to Warhammer.

Building Elo into Siege Ladder #

I launched Siege Ladder recently (December 2025), and as things stand it supports the Middle Earth Strategy Battle Game (MESBG), but my ambitious plans stretch beyond that, and the player profile view is designed with that in mind.

On the left-side of the player profile view, I have left space for many game systems. I've started with MESBG, where here, you can see FlaccidAmigo's Elo Rating for MESBG is 1548. Similar to how, in Chess, you have different ratings for Rapid, Blitz, Classical, etc. I want to built out the same here but with ratings for MESBG, 40k, AoS, etc.

The "~" next to the number is used as an indicator to say that he has recorded less than 20 x ranked games on Siege Ladder, so the rating itself is still open to changing quite a bit, I'll cover this in a little more detail shortly.

One thing I wanted to make sure was that not every game recorded onto Siege Ladder should impact this rating, otherwise I could see it sucking the fun out of the games if every game mattered. That's why, when recording games in Siege Ladder, you can choose whether or not it's a "Ranked" game or not, with the option defaulting to "Casual". This goes the same for events on Siege Ladder too.

For example, the What about Elevensies 2025 event, that was part of the GBHL is a "Ranked" event, with every player's Elo being impacted by the results in that tournament. On the other hand, the recent HH Christmas Blitz (pictured above) was a Casual Event, so whilst each player finishing in a podium position was still given an award (shown in Matt Freemantle's Profile View), the games from that event did not affect the player's Elo Rating.

How Elo Works #

I don't want to scare anyone off with complicated maths (though, I'm sure many of you are just as into it as I am!), but I did want to spend a little time explaining how Elo works.

As mentioned previously, the core of Elo is about comparing the Expected Score with the Actual Score. 

Calculating the Expected Score #

For each player, the first step is to work out their Expected Score, i.e. out of 1, how likely are they to win? This just considers the player's rating and their opponents rating:

The use of 400 here is taken from the formula used in the Chess systems. Whilst it can be modified, I wanted to keep it as consistent as possible so that Chess, and this system, use the same scale. Mathematically, it means that a 400-point difference in rating equates to 10:1 odds.

The Expected Score can be interpreted as:

• 0 to 0.5: The player is expected to lose.
• ~0.5: The player is expected to tie.
• 0.5 to 1.0: The player is expected to win.

Reducing Impact over Time #

The final step to the formula is our K-Factor. This is the largest value that a player's rating can change from a given game. It is good practice to have this value reduce over time to ensure your Elo can stabilise, and also give new players a chance to have their Elo (which starts at 1,500 for all new players) quickly reflect their actual ability.

At Siege Ladder, we made the decision to have the following K-Factors:

• K = 32; First 30 matches.
• K = 24; 30-100 matches.
• K = 16; All matches thereafter.

Let's Look at an Example #

In a recent game against Ben Bartlett, I went into the game with an Elo of 1,544. Ben's Elo was 1,457. Using the Elo calculator I've made freely available on Siege Ladder I can see the possible outcomes:

The beautiful element of Elo here is that, even if the game is a draw, our Elo's are impacted, with Ben's increasing and mine decreasing because that would not have been the expected result.

Game System Leaderboards #

Now with this system in place, it's possible to create an event-agnostic leaderboard for each game system.

You can find the leaderboard for MESBG here, and I'll be adding more as I branch Siege Ladder into more gaming systems too.

Ideas for the Future  #

Two key areas I want to improve Siege Ladder in are as follows:

• Surfacing Elo in Events: Currently the events view doesn't expose another player's Elo, you still need to navigate to the player's profile to view it. I want it to be very clear and easy to see the Elo for every player you're about to face, similar to how Longshanks does this for their own rating system when viewing the registrations for a given event.
• Elo for Armies: This would be a bit experimental, but I think it should work. I want to create an Elo for each army in MESBG, so for every ranked game stored in Siege Ladder, that impacts the Elo of both the player and the army separately. In theory, in a perfectly balanced game, there shouldn't be any bias either way, but it would be great to find out. It also adds richness to the data that most people use at the moment which is just the win % of each army. Instead, Elo would consider the quality of the army before each game and the subsequent outcome. Getting more advanced, I may even be able to consider impacting the player's Elo with the army they have chosen to play. Beat the Balrog with Wraiths on Wings would give you a higher impact on your Elo than if you'd beat The Shire with Rohan for example.

I'm really excited to see Siege Ladder grow, and I'm very grateful to all of those that are using it already. We recently had our first live event run it, and you can read more about that in my write-up here.

Please do also reach out via the Discord Server, share your thoughts and feedback, those already on there will vouch, I'm very responsive and keen to improve this in any way I can.