D&D 5E 5e Elo Rating (MMR) for PCs and 2000 monsters (preliminary results 1v1)

[Greggy C]

So I ran our (Fan Content) 5e combat simulator for every monster and a bunch of pregenerated characters and assigned them an Elo Rating (similar to MMR) based on the fight results.

Here are the results:

DnDCombat Compendium

www.dndcombat.com

Now caveat. Only 39% if the compendium is implemented, so someone without a symbol in the testing column, like Yeenoghu, is not at full power.
And there are certainly bugs helping many monsters cheat their way up the ladder.

But I though I would secretly share these early results with enworld*

 

[Stalker0]

Ooooh interesting! So lets walk through how this works:

ELO Rating - So my assumption here is that you did many many MANY 1 on 1 combats between various combinations. The winner increases their ELO a bit, the loser loses a bit..... until eventually ratings shake out. Is that a fair approximation?

So one way of looking at the chart above, is that while Tiamat and the Tarrasque are both CR 30, we could say that Tiamat is a "better CR 30" due to a higher ELO. Is that correct?

So I guess the million dollar question to me is, how do you decide how/when to use spells and various abilities/resources?

 

[Greggy C]

Ooooh interesting! So lets walk through how this works:

ELO Rating - So my assumption here is that you did many many MANY 1 on 1 combats between various combinations. The winner increases their ELO a bit, the loser loses a bit..... until eventually ratings shake out. Is that a fair approximation?

If you click on the Elo number itself you can see which fights they did to get to their elo rating, who they defeated, who they lost to.

 

[Greggy C]

So one way of looking at the chart above, is that while Tiamat and the Tarrasque are both CR 30, we could say that Tiamat is a "better CR 30" due to a higher ELO. Is that correct?

Tiamat is higher because Tiamat never lost, whereas Tarrasque lost to Tiamat. Each individual row the Elo history is a best of 8 fights. If it is a draw it is not recorded.

 

[Greggy C]

So I guess the million dollar question to me is, how do you decide how/when to use spells and various abilities/resources?

Thats a big discussion. The short answer is every spell and action a player or monster can do has a base weight (a number that can be overridden by the owner). There are many factors that adjust that value (e.g. does the breath weapon hit many targets, does the spell disable the opponent) based on circumstance of the current positioning in the battle but ultimately the highest weighted action is selected to be executed. That "intelligence" is always a work in progress, there are bugs, there are only 31% of spells, class features, monster traits etc implemented so all these things affect this preliminary rating. It will be another 6 months before we reach 100% and probably 9 months before it is perfected.

 

[Stalker0]

So I checked out the site. HOLY CRAP!! Amazing, simply, easy to use....the fights run super quick, and you can literally see step by step how each fight went if you really want to. I am incredibly impressed.

 

[Benjamin Olson]

ELO Rating - So my assumption here is that you did many many MANY 1 on 1 combats between various combinations. The winner increases their ELO a bit, the loser loses a bit..... until eventually ratings shake out. Is that a fair approximation?

Note that it's an "Elo Rating" after creator Arpad Elo, not an "ELO Rating" which is only appropriate for comparing things to the Electric Light Orchestra.