D&D 5E Here's The Most Common D&D Party Composition

D&D Beyond's latest data-output looks at the composition of the typical adventuring party. The 'traditional' party always used to be Fighter, Cleric, Rogue, Wizard; let's see how that stacks up these days!

These screenshots were compiled by SageAdvice.eu. DDB's developer said "I’m going to be honest: this was really hard to look at from a data perspective right, so what I mean by that is it’s hard to figure out exactly how to chop this data up for it to be the most meaningful that we can make it all right. These are all campaigns where party members and characters within that campaign are taking hit point adjustments, so that’s one of the best senses that we have that something is actually being played”.

​

 

[EzekielRaiden]

When you take just the ones at top, it's not surprising that they are close.

But let's look more at the numbers. For example, with 12 classes there are 1728 different three person parties (assuming straight classed only, allowing duplicates). If they were all equally likely, we'd be seeing 0.05787% each. Instead the most likely are about nine times as likely, which also mathematically shows that there a enough that are below that for the average to work out.

If the lowest are the same amount below the average as the highest are above, then there's about an 1 to 80 ratio in likelyhood between lowest and highest.

Even if no one ever played duplicates, that's still 1320 different combos - they highs listed are still much more common than the average.

If you actually wanted to analyze this, you'd have to do a goodness-of-fit test, though I'm not entirely sure it would apply. There are reasons, entirely apart from player preferences, why some compositions won't appear even without repetitions--meaning that the naive assumption of an expected uniform distribution may be simply wrong, even in principle. E.g. a party that has no spellcasters at all is severely restricted in almost all areas of play, while Sorcerer/Warlock/Wizard/Druid has a ton of conceptual overlap that mere social convention will usually push against. Further, it's entirely reasonable that for probabilities this low, even a very large data set might not have any at all; if the uniform average probability is 1/1320, then even a data set with three thousand points has (1319/1320)^3000 = about 10% chance of any particular equally-likely party to simply never show up at all in the set. Random noise in the sample can shift things dramatically, and with such tiny groupings compared to the overall size, a shift of merely four or five parties--meaning just four or five people out of thousands made a single different chocie--could radically affect these results. Drawing overly-broad conclusions from this data is very, very easy.

 

[TarionzCousin]

yes, it is a giant PITA to run, why do you ask?

Is it a giant pain in the--oh. Never mind.

 

[R_J_K75]

I’m curious about parties with duplicate classes too: two rogue, two wizards, that kind of thing.

I Started my current campaign June, 1 2018. I pre-generated the characters for 2 reasons. The first being that I was sure if he game would last as 3 out of 5 players were new to D&D, so I didnt want to spend a session creating PCs. Secondly I wanted simpler characters. I gave them some simple options to choose from. We ended up with 2 fighters, 2 rogues and 1 wizard. 1 of the rogues got eaten by a bullette and made a ranger. Thats how the party is now.

 

[EzekielRaiden]

Although I admit it depends, in part, on assumptions that might not hold true*, it's looking like the party-of-four stats are based off a sample size of approximately 2500 parties, meaning there's an expected value around 2 for the "uniform distribution of non-repeat compositions" assumption. If we take into account that most four-person groups won't have both a sorcerer and a wizard and similar such things, we can reduce even further the number of expected "popular" parties, potentially down even as far as only 81 or so "popular" options and the rest literally unseen because the sample size isn't big enough.

*Specifically, I'm assuming that we can guess that the .65% to .53% gap is a gap of 3 parties, and the .53% to .49% is a gap of 1 party. This is a pretty reasonable assumption, as it would be quite strange for there to be large and exclusively even(/multiple of 3 etc.) gaps in these values--we expect relatively small gaps. Also, I assumed that these values given were exact decimal percentages, which is very unlikely, they were almost certainly rounded. But by this reasoning from the difference-of-3 between the first two clusters and a difference-of-1 between the second and third cluster, I got the same value both ways, approximately 2500. I think we can say with confidence that that's about the size of the data set for four-person parties, and it's very unlikely that any of the other data sets are particularly larger than that.

 

[Parmandur]

[/QUOTE]

Although I admit it depends, in part, on assumptions that might not hold true*, it's looking like the party-of-four stats are based off a sample size of approximately 2500 parties, meaning there's an expected value around 2 for the "uniform distribution of non-repeat compositions" assumption. If we take into account that most four-person groups won't have both a sorcerer and a wizard and similar such things, we can reduce even further the number of expected "popular" parties, potentially down even as far as only 81 or so "popular" options and the rest literally unseen because the sample size isn't big enough.

*Specifically, I'm assuming that we can guess that the .65% to .53% gap is a gap of 3 parties, and the .53% to .49% is a gap of 1 party. This is a pretty reasonable assumption, as it would be quite strange for there to be large and exclusively even(/multiple of 3 etc.) gaps in these values--we expect relatively small gaps. Also, I assumed that these values given were exact decimal percentages, which is very unlikely, they were almost certainly rounded. But by this reasoning from the difference-of-3 between the first two clusters and a difference-of-1 between the second and third cluster, I got the same value both ways, approximately 2500. I think we can say with confidence that that's about the size of the data set for four-person parties, and it's very unlikely that any of the other data sets are particularly larger than that.

I think you are probably off by a couple of decimal points, given the installed user base involved.

 

[Parmandur]

If you actually wanted to analyze this, you'd have to do a goodness-of-fit test, though I'm not entirely sure it would apply. There are reasons, entirely apart from player preferences, why some compositions won't appear even without repetitions--meaning that the naive assumption of an expected uniform distribution may be simply wrong, even in principle. E.g. a party that has no spellcasters at all is severely restricted in almost all areas of play, while Sorcerer/Warlock/Wizard/Druid has a ton of conceptual overlap that mere social convention will usually push against. Further, it's entirely reasonable that for probabilities this low, even a very large data set might not have any at all; if the uniform average probability is 1/1320, then even a data set with three thousand points has (1319/1320)^3000 = about 10% chance of any particular equally-likely party to simply never show up at all in the set. Random noise in the sample can shift things dramatically, and with such tiny groupings compared to the overall size, a shift of merely four or five parties--meaning just four or five people out of thousands made a single different chocie--could radically affect these results. Drawing overly-broad conclusions from this data is very, very easy.

You are.togjy that drawing conclusions is dicey, as Adam Bradford was at pains to say in the video: but the sample size is significantly northwards of 3000, based on how many people are using the platform (well into the six figures)

 

[FrogReaver]

When you take just the ones at top, it's not surprising that they are close.

But let's look more at the numbers. For example, with 12 classes there are 1728 different three person parties (assuming straight classed only, allowing duplicates). If they were all equally likely, we'd be seeing 0.05787% each. Instead the most likely are about nine times as likely, which also mathematically shows that there a enough that are below that for the average to work out.

If the lowest are the same amount below the average as the highest are above, then there's about an 1 to 80 ratio in likelyhood between lowest and highest.

Even if no one ever played duplicates, that's still 1320 different combos - they highs listed are still much more common than the average.

Classes can duplicate in a party

 

[Blue]

Classes can duplicate in a party

Yes ... most of my post was assuming they could. I even called it out: "with 12 classes there are 1728 different three person parties (assuming straight classed only, allowing duplicates)". That's what I use for everything except the last line.

I have a last sentence that specifically calls out that even without duplicates here are the numbers and the conclusion of the math still holds. Just to show it's true no matter what.

 

[pogre]

I'm curious who gets to play with 8 people in the party.

I'm curious about who wants to play with 8 people in the party.

When I started my current campaign I was concerned about making sure I had at least 4 players at every weekly session. I invited some of my old players - one who was two hours drive away and another who was two-and-a half hours away. I invited a guy who is an LPN and is occasionally on call when we play. Which, is a long way of saying I knew some of the players were not going to make it to every session. I have a ten player pool that I was fairly sure I was going to get five showing up every week.

Over the last year I have averaged eight players per session.

Yeah, that's probably one or two too many, but so far it has been great. It helps greatly that they are a cooperative group of people mostly concerned with everyone having fun.

Our most frequent group composition is: Cleric, Paladin, Monk, Sorcerer, Ranger, Wizard, Druid, and Fighter.

 

[Connorsrpg]

Yeah, data with such small percentages; does it really tell us anything?
For what it's worth (and someone mention playing with large groups. My 5E Elsir Vale campaign right now has 9 players.
Fighter x2
Warlock
Ranger
Druid
Paladin
Cleric
Rogue
Wizard