# The Watch Ruler — method and sources

**1,605 watch models · 56 brands · pulled 2026-08-16**

## Where the data comes from

Every specification and price in this file is the manufacturer's own published
figure as carried by **Watches of Switzerland (US)**, an authorised multi-brand
retailer, read from its public product index. 4,022
products were pulled; 1,605 remain after dropping any
watch missing one of the four measured characteristics and collapsing colourway
variants of the same model to their cheapest example.

**Nothing here is estimated, modelled or filled in.** A watch missing a metric
was dropped, never guessed.

## The score

Four characteristics, each chosen so that it is a published fact *and* something
you can point at on a photograph of the watch:

- **Water resistance** — the case, how deep it survives — **30%**
- **Case durability** — the metal, how it survives being worn — **25%**
- **Movement** — the caseback, mechanical, or a battery — **25%**
- **Complications** — the dial, what else it tells you — **20%**

Each watch is ranked against all 1,605 others on each
characteristic and converted to a percentile, then the four percentiles are
combined with the weights above. Percentiles use the **midrank** convention
(`pct = (worse + (tied − 1) / 2) / (n − 1)`): water resistance takes only a
handful of distinct values, so without it two watches with identical specs would
score differently depending on their position in the file.

Where two watches score identically, **the cheaper one ranks higher**. Four
coarse inputs cannot separate 1,605 watches into
1,605 distinct scores, so ties are a real property of
this data rather than something to hide.

## What the score deliberately does NOT measure

- **Chiming complications** (minute repeaters, sonneries) — the source has no
  field for them. This understates a handful of very expensive watches, and it
  is the single largest known limitation of this table.
- **Hand-finishing, decoration and provenance.**
- **Resale value and market liquidity.**
- **Accuracy in service** — chronometer certification is not published per
  reference in this source.
- **Price**, which is deliberately not an input, so that comparing the score
  against the price is not circular.

**Case durability grades the metal on surviving being worn, not on what it
costs.** Gold and platinum are soft and score low; ceramic, hardened titanium
and carbon composites score high. That is a judgement, it is the only one in the
table, and the `case_material` column is included so you can re-grade it.

## What it found

- The correlation between price and score across the whole set is **negative**.
- Watches around $9,550 average a higher score than the
  most expensive tenth of the market.
- The highest-scoring watch is the **Panerai Submersible Carbotech** (PAM01616) at
  $19,900.

## Columns

`pct_*` columns are the percentile the watch reached on that characteristic,
0–100. `variants_listed` is how many colourways of the same model the retailer
listed; the row shows the cheapest.