Your five-star books have a pattern: find your reading taste in your own ratings

If you've rated books on Goodreads for a few years, you're sitting on a small dataset about yourself, and it's more revealing than you'd expect. Your five-star books have a pattern. Most people have never looked for it, because a list of titles hides it well.

This guide shows how to find that pattern with nothing but your ratings and some plain thinking about data. You can do it on paper. It's also exactly what the Match Book app does automatically, so this is an explanation of how it works as well.

Step 1: Decide what counts as a favourite

Ratings are noisy. A four-star book might be one you loved with a small reservation, or one you admired without really enjoying. Five stars is a much clearer signal. A good rule, and the one Match Book uses:

• Five stars count fully. These are your clearest evidence.
• Four starscount half. They're evidence, but weaker.
• Three stars and belowdon't count towards your taste. A three is usually “fine”, which tells you very little.

Weighting like this is a standard way to handle signals of different reliability. You don't throw weaker evidence away, but you don't let it outvote the strong stuff either.

Step 2: Describe each favourite the same way

You can't find a pattern in titles. You need to describe every book along the same dimensions: its mood, its pace, how heavy it is, how it ends, what it's about. Match Book uses fifteen, but even four or five will show you something. Mood, pace, tone and ending are the ones to start with.

This is the step that turns a list of books into data. Once a title has become “eerie, slow burn, balanced, bittersweet”, you can compare it with every other book described the same way.

Step 3: Count, per dimension

For each dimension, add up the weights behind each value. Say you have 20 five-star books and 10 four-star ones. On mood, you might find something like this:

• Dark: 9 five-stars and 2 four-stars, for a weight of 10
• Tense: 4 five-stars and 4 four-stars, for a weight of 6
• Melancholy: 4 five-stars, for a weight of 4
• Everything else: a weight of 5 between them

Out of a total weight of 25, dark accounts for 40%. That's your dominant value on mood, and 40% is its strength: how concentrated your taste is on that dimension.

Step 4: Compare with what's normal

Here's the step most people miss, and it matters more than any of the others. A value being common among your favourites doesn't mean much if it's common among all books. Statisticians call this the base rate.

In Match Book's catalogue of 1,407 books, about 13% read as dark. The most common mood is tense, at about 25%. So if 40% of your favourites are dark, that's roughly 3 times the base rate. That's a real preference, not a coincidence. If 40% of your favourites were tense, it would say much less, because that's close to what you'd get by picking books at random.

The ratio between your share and the base rate is sometimes called lift. High lift means a preference that's genuinely yours. Lift close to one means you just read what everyone reads.

Step 5: Ignore the dimensions you don't care about

You won't have a strong preference on every dimension, and that's information too. If your favourites are spread evenly across first person and third person, narration doesn't matter to you. A recommendation that insisted on matching it would be optimising for noise.

So rank your dimensions by strength, and by lift if you worked it out. The top four or five are your taste. The rest are things you're flexible about, and that flexibility is where surprising books come from.

Three traps to watch for

Small samples

With eight favourites, one book shifts a value by more than ten percentage points. Treat any “pattern” from fewer than about fifteen rated books as a hunch. This is why Match Book lets you start with its sample library. The method is easier to judge on a big dataset than on a thin one.

You only rate what you chose to read

Your ratings aren't a random sample of books. They're a sample of books you already expected to like. If you've never read a slow-burn novel, your data can't tell you whether you'd love one. It can only tell you that you don't pick them. Statisticians call this selection bias. The fix is to deliberately try the other thing now and then. There's a guide on that: explore or exploit.

Taste drifts

What you loved at twenty-two isn't necessarily what you love now. If your library covers many years, compare your last two years with the rest. The difference is often the most interesting thing you'll find.

What you end up with

At the end you have a short description of your taste in dimensions rather than titles. For example: “dark or melancholy, slow to moderate pace, heavy, bittersweet or ambiguous endings, often about grief or identity, genre flexible.” That's far more useful than “I like literary fiction”, because it can find you a thriller or a fantasy novel you'd love.

To use it, go to the collections that match your strongest dimensions, then open any book to see its closest neighbours. Or let the app do the counting: Match Book reads your Goodreads export, draws your taste as a fifteen-spoke web, and finds books that fit it. Getting the export takes a couple of minutes. The steps are in how to export your Goodreads library.