Character archetypes
Only two character types in romance are actually a couple
We counted the character types named across 65,510 romance novels and asked which ones share a book more often than chance allows. Two pairs came back with a very loud answer, and the rest of the genre shrugged.
SmittenSpines · 65,510 books · published 4 August 2026
Romance readers talk about character types the way other people talk about star signs. He's a grump, she's sunshine. He's the bad boy, she's the good girl. The shorthand is so established that it gets printed on the cover, and readers buy on it.
Shorthand can be a habit as easily as a fact, though. If you counted character types across tens of thousands of books, would those famous pairs come back as pairs, two halves of one thing? Or just as two popular ingredients that happen to be popular at the same time?
We counted, and the answer came back tidier than we expected. The genre has exactly two archetype pairs that behave like a bonded unit, and everything else in romance is a long way behind them.
- 40×
- grumpy + sunshine, versus chance
- 21×
- bad boy + good girl, versus chance
- 3.0×
- the next strongest pair in the whole genre
- 65,510
- books searched for character-type language
1,016 books, about 26 expected
479 books, about 23 expected
brooding + sunshine
What “more often than chance” means here
Two types can share a lot of books just because both are common. Mentors and villains turn up together in 6,465 books, which sounds like a bond until you notice that each appears in more than a quarter of the corpus on its own. Put two common things in a big pile and they'll meet constantly without meaning anything by it.
So the measure throughout this piece is a ratio: how often two types really do share a book, divided by how often they would if each were sprinkled through the shelf independently. A ratio of 1× is exactly what coincidence produces, and 2× is a genuine leaning. Once you get past 10× the two things have stopped being separate choices and started being one.
Association strength
Two pairs sit in a different universe from the rest of the genre
How much more often each pair of character types shares a book than chance would produce. The dashed line is chance.
Show the numbers as a table
| Pair | Books with both | Expected by chance | Ratio |
|---|---|---|---|
| grumpy + sunshine | 1,016 | 26 | 39.61× |
| bad boy + good girl | 479 | 23 | 20.76× |
| brooding + sunshine | 468 | 157 | 2.98× |
| bad boy + rake / playboy | 216 | 73 | 2.95× |
| brooding + grumpy | 490 | 179 | 2.74× |
| chosen one + reluctant hero | 404 | 148 | 2.72× |
| anti-hero + femme fatale | 475 | 193 | 2.47× |
| alpha + strong-willed heroine | 695 | 303 | 2.29× |
| chosen one + mentor | 1,007 | 489 | 2.06× |
| chosen one + protector | 371 | 181 | 2.05× |
| anti-hero + damsel in distress | 721 | 353 | 2.04× |
| alpha + damsel in distress | 1,178 | 580 | 2.03× |
| bad boy + brooding | 508 | 254 | 2× |
| anti-hero + brooding | 1,265 | 648 | 1.95× |
| brooding + chosen one | 434 | 233 | 1.86× |
| damsel in distress + strong-willed heroine | 310 | 171 | 1.81× |
| anti-hero + chosen one | 235 | 138 | 1.71× |
| alpha + protector | 1,401 | 826 | 1.7× |
| mentor + reluctant hero | 2,419 | 1464 | 1.65× |
| anti-hero + villain | 2,246 | 1366 | 1.64× |
Grumpy and sunshine appear together in 1,016 books. If the two traits were picked independently you'd expect around 26. Bad boy and good girl: 479 books against roughly 23 expected. Neither of those looks like a tendency. Both look like a single authorial decision getting counted twice.
Then comes the cliff. The third-strongest pair in the entire genre is brooding + sunshine at 3.0×, which is a mild preference and nothing more. Alpha and damsel in distress, a pairing the genre's reputation would have you believe is iron-clad, comes in at 2.0×. Real, and completely ordinary.
The genre's actual default is neither of them
Ask instead which single character types appear most often and the famous pairs are nowhere near the top. Grumpy shows up in 2.1% of books, sunshine in 1.9%. What fills the top of the list is damage, plus two roles the genre shares with every other kind of story ever told.
Frequency
Villain, mentor, wounded lead: the three types romance names most
Share of books whose descriptive text names each character type. The wounded lead is picked out in red.
Show the numbers as a table
| Character type | Books | Share |
|---|---|---|
| villain | 17,873 | 27.3% |
| mentor | 17,789 | 27.2% |
| wounded / tortured | 17,051 | 26.0% |
| brooding | 8,475 | 12.9% |
| alpha | 8,214 | 12.5% |
| outsider | 7,853 | 12.0% |
| protector | 6,590 | 10.1% |
| reluctant hero | 5,393 | 8.2% |
| anti-hero | 5,006 | 7.6% |
| damsel in distress | 4,624 | 7.1% |
| caregiver | 4,471 | 6.8% |
| femme fatale | 2,521 | 3.8% |
| rake / playboy | 2,440 | 3.7% |
| strong-willed heroine | 2,420 | 3.7% |
| bad boy | 1,966 | 3.0% |
| chosen one | 1,802 | 2.8% |
| grumpy | 1,384 | 2.1% |
| sunshine | 1,214 | 1.9% |
| cinderella | 1,105 | 1.7% |
| good girl | 769 | 1.2% |
| ice queen | 667 | 1.0% |
| warrior woman | 638 | 1.0% |
| lone wolf | 308 | 0.5% |
| wallflower | 206 | 0.3% |
| girl next door | 75 | 0.1% |
Two of those three are furniture that almost any story carries: somebody to oppose the leads, somebody to advise them. The third one belongs to romance. A wounded or tortured lead appears in 17,051 books, which is 26% of the shelf, and the single most common label anywhere in the corpus is wounded healer, used 7,970 times.
Set the villain and the mentor aside and what's left near the top runs together into one list: brooding, protector, anti-hero, damsel in distress. Ways to be hurt, and ways to be rescued from it.
That puts the grumpy/sunshine finding in a different light. Grumpy is not the genre's favourite hero. The wounded one is. Grumpy is just the only type that arrives with its own mandatory partner.
Who ends up with whom, specifically
Everything above counts types that share a book, which isn't quite the same question as who ends up with whom. A mentor and a villain share plenty of books without sharing a bed.
For a narrower answer we used only the books where the text explicitly attaches a type to the male lead and a different one to the female lead. That leaves a much smaller sample, 4,714 books out of 65,510, because most descriptions never mark the roles that plainly. Treat it as a sketch rather than a census.
Lead pairings
What the leads are, when the text says which is which
Ribbon width is the number of books pairing a given male-lead type with a given female-lead type. Hover a ribbon to isolate it.
Show the numbers as a table
| Male lead | Female lead | Books |
|---|---|---|
| wounded | strong-willed | 124 |
| alpha | wounded | 77 |
| alpha | strong-willed | 71 |
| alpha | damsel in distress | 48 |
| wounded | resilient | 42 |
| knight in shining armor | wounded | 42 |
| alpha | innocent | 41 |
| alpha | reluctant | 39 |
| wounded | wounded | 32 |
| grumpy | sunshine | 32 |
| tortured | wounded | 29 |
| wounded warrior | wounded healer | 28 |
| wounded | free-spirited | 27 |
| protector | wounded healer | 26 |
| bad boy | good girl | 24 |
| wounded | innocent | 23 |
| wounded | determined | 22 |
| redeemed | wounded | 21 |
| protective alpha | wounded | 20 |
| protector | survivor | 20 |
| knight in shining armor | wounded healer | 20 |
| dark | innocent | 19 |
| protective | wounded | 18 |
| charming rogue | wounded | 18 |
The wounded lead wins here too, on both sides at once. The most common explicit pairing in the corpus is a wounded hero with a strong-willed heroine, and the second is an alpha with a wounded heroine. Sit with the ribbons for a minute and the working formula shows up: one of them is damaged, the other is unusually sturdy, and the story is the negotiation between those two facts.
Grumpy/sunshine and bad boy/good girl show up here as well in smaller numbers, which is the useful bit. They're real pairings rather than an artefact of how we counted.
Why any of this matters if you just want a book
It tells you which bits of shorthand actually predict a reading experience. When someone recommends a grumpy/sunshine book, that phrase is carrying real information: the two halves arrive together and the tension between them is the plot. When someone tells you a book has an alpha hero, you have learned much less than you think, because the corpus says alphas turn up next to almost anybody.
It also explains a very specific disappointment. Search a catalogue for “grumpy” and you'll get thousands of books, most of which aren't the thing you meant, because grumpy on its own is only a personality note. The thing you meant has two sides to it, and it has to be searched for that way. Which is the part we're building.
Checking our work
Every figure here comes from one corpus with known limits. It is a snapshot, it leans heavily towards books published between 2020 and 2024, and it works from what gets said about books rather than from their full text. The methodology page sets out what that does and doesn't let us claim, including the measurement we ran to find out how often our own counting is wrong.
The waiting list
Get one email when the search box opens
One email, then nothing. It isn't a newsletter and there's no drip sequence waiting behind it with “5 tips” in the subject line. If that ever changes, you get to say no first.
This form isn't connected to anything yet, and pressing it stores nothing.