Methodology

The WristQuant Relative Value (RV) model scores each reference on a 0–100 scale by combining three quantitative signals: a z-score measuring where the current secondary market premium sits relative to its own history, a peer rank measuring how the reference compares to its peer group, and a 3-month price momentum measure. The composite score determines the demand signal (Strong Demand, Neutral, or Weak Demand) assigned to each reference.

All price data is sourced from WatchCharts (watchcharts.com), which provides daily secondary market price history for individual references. Retail prices are sourced from brand websites and authorised dealer listings. Equity data is sourced from Yahoo Finance.

The Issue 01 screen covers 30 references selected from the WatchCharts active secondary market universe using the following criteria:

Liquidity filter: References were required to have sufficient transaction volume on WatchCharts to support reliable price history — in practice, this means at least 12 months of consistent secondary market pricing data as of April 2026. Thinly traded references with sparse or irregular price observations were excluded to avoid z-score distortion.

Brand and segment representation: The screen was designed to span the major segments of the secondary market — sport and tool watches (Rolex, Patek Philippe, Audemars Piguet), dress watches, independent watchmakers, and mid-market references. This ensures the RV signals reflect a range of demand dynamics rather than being dominated by the Rolex-heavy composition of the broader secondary market.

Analytical interest: A small number of references were included specifically because they were relevant to Issue 01's analytical themes: the GMT-Master II Pepsi (discontinuation thesis), the Alain Silberstein Krono Bauhaus and Louis Erard × Silberstein (independent watchmaker opportunity), and several LVMH-brand references (to test the macro correlation argument directly). These deliberate inclusions are noted in the issue where relevant.

The universe will be reviewed and updated each issue. References may be added or removed based on liquidity changes, analytical relevance, or coverage of new market segments. The composition of the screen is disclosed in each issue.

The z-score is computed on the secondary market premium percentage — the percentage by which the current secondary market price exceeds or falls below the manufacturer's suggested retail price (MSRP). For discontinued references without a current MSRP, the z-score is computed on the absolute price level instead.

The mean and standard deviation are computed over the full available price history for each reference (ranging from approximately 2 to 4 years of daily data as of April 2026). A positive z-score indicates the current premium is elevated relative to history; a negative z-score indicates it is compressed.

Note: The z-score is not a statistical significance test. It is a normalisation tool. Given the autocorrelated nature of watch price data, the effective sample size is substantially smaller than the number of daily observations, and the z-score should be interpreted as a relative positioning measure rather than a probabilistic statement.

Each reference is assigned to one of four peer groups based on its primary market positioning: dress watch, sports watch, complication, or tool watch. Within each peer group, references are ranked by their current secondary-to-retail premium percentage (lowest premium = rank 1). The rank is then converted to a quartile (Q1–Q4), where Q1 represents the cheapest references in the peer group relative to retail.

The 3-month momentum signal measures the percentage change in secondary market price over the trailing 90 calendar days. Positive momentum indicates recent price appreciation; negative momentum indicates recent price decline.

Each of the three components is normalised to a 0–100 scale within the full 30-reference universe, then combined using the following weights:

The composite RV Score ranges from 0 to 100. Higher scores indicate references that are cheap relative to their own history and peer group, with positive recent momentum — the combination most likely to indicate mean-reversion opportunity. The backtest (see Archive) shows that the top-quartile RV Score references outperformed the bottom quartile by 3.6 percentage points at 6 months.

The weighting rationale is described in detail in Issue 01. The optimised weights (30/20/50) produced a superior backtest result, but the baseline weights (40/40/20) are used for the live screen to avoid overfitting to a single backtest period.

The demand signal is assigned based on the composite RV Score and a secondary qualitative check:

A sub-signal label is also assigned: mean-reversion (z-score driven), peer discount (peer rank driven), disc. premium (discontinued reference trading above historical average), or disc. discount (discontinued reference trading below historical average).

Condition assumption: All secondary market prices reflect unworn or lightly used examples with box and papers. Prices for worn examples or incomplete sets will differ materially.

Liquidity: The model does not adjust for liquidity differences between references. A Rolex Submariner and a Cartier Panthère trade in very different liquidity environments; the z-score and peer rank treat them symmetrically.

Backtest limitations: The backtest covers April 2022 – April 2026, a period that includes the post-COVID correction and recovery. Results may not generalise to other market regimes. The optimised weights are derived from the same period used to evaluate them, introducing look-ahead bias.

Correlation measurement: The −0.18 rolling correlation between the WatchCharts Index and the Watch Equity Basket is economically and practically indistinguishable from zero, but this does not constitute a formal statistical test. Given the autocorrelated nature of watch price data, the effective sample size is substantially smaller than the number of monthly observations, and the correlation should be interpreted with appropriate caution.