D&D 5E 538's Race/Class data, in relative terms

Unfortunately, the original thread for this topic got hopelessly derailed, so here's a fork. Rather than look at absolute numbers, I thought it might be more useful to compensate for relative popularity of races & classes, showing a better metric of which combinations are considered good or bad.

Mathematically, I took the raw numbers, then divided the values for each race by the square root of that race's mean and divided the values for each class by the square root of that class's mean. Between the two adjustments, you end up with numbers averaging about 1. Roughly, 0.5 = rarely chosen, 1.5 = not surprising, 2.5 = popular classic.


Note: DNDBeyond's free character builder doesn't include Drow Elf or Forest Gnome, so they are under-represented.

FWIW, I am very disappointed by the strong consensus against Half-Elf Barbarian, and intend to play one at the next opportunity. Lightfoot Warlock and a few other stat-appropriate combos also deserve much more love.

Thoughts?

This is pretty cool, thanks for normalizing these numbers.

I haven't looked at the raw data to know if it is possible to tell the following, but I've got a number of permutations that are of interest to me.

If you look at the basic-ony characters (basic race & class, no feats, multiclassing, or non-basic spells), how do the numbers look? If you exclude them how do they look?

What do these numbers look like if you only include the highest level class for multiclass characters?

What multiclass combos are common, which are rare?

If a class is often found in a multiclass, does it usually end up being either primary or a dip (or no trend).

If you restrict it to level 2-4,6-19 characters, that may cut out many "experimental builds" at 1st, 5th and 20th that don't actually see play. How does the race/class combos look then?

I personally prefer dwarf barbarians to fighters. The Builder is a lie!

I played a halfling barbarian (dual wielding short swords) that was hella fun

Blue said:

If you look at the basic-only characters (basic race & class, no feats, multiclassing, or non-basic spells), how do the numbers look? If you exclude them how do they look? . . .

Click to expand...

Good questions. I don't have the raw data; all I'm going on is the table posted at FiveThirtyEight, which listed the number of class-race pairs per 100000 characters. The totals came out to a little over 109000, so about 9% was multiclassing, but I don't know individual character details.

Mephista said:

I personally prefer dwarf barbarians to fighters. The Builder is a lie!

Click to expand...

LOL. I don't blame the Builder for what users do with it. On average, most people follow the common wisdom, even if it isn't necessarily wise.

For example, look at the clear preferences for Half-Elf & Halfling Bard / Dragonborn Sorcerer / Tiefling Warlock, when they really could work in any combination.
In fact, Dragonborn's Str boost is completely wasted on a Sorcerer, whereas a Valor Bard or a Blade Warlock could actually make use of it. But more people go with the thematic choice instead. Likewise, Half-Elf has better stats for Paladin than for Rogue or Ranger, but those classes "feel" right.

Sorry [MENTION=61252]Frankie1969[/MENTION], my mistake. 538 often puts their raw data up on github, but I went back and read the article and this was already broken out by Curse (published of D&D Beyond).