OpenFacet – From Market Depth to Market Scarcity: Estimating How Hard a Diamond Is to Find

A price benchmark helps a buyer assess a quote. Finding a comparable diamond is a separate problem, and a reasonable price does not guarantee much choice among stones.

OpenFacet now estimates relative market scarcity alongside price. The score measures findability in the covered retail market, not geological rarity.

Our previous market view showed available offers across carat, color, and clarity. It made useful differences visible: common sizes often had substantial listing depth, while larger stones and some grade combinations had much thinner supply.

Those counts also reflected how we collected the data. One source might report its full matching inventory; another might return a limited page of results. Some specifications were checked regularly, others occasionally. The same stone could appear through more than one seller. A blank cell could mean limited supply, a failed request, or a specification we had not searched that day. The numbers were useful, but they were not a complete inventory.

Estimating the gaps

We cannot search every combination continuously. An availability estimate has to work between observations and for specifications we check less often.

The baseline is fitted from historical offer observations—not carried over from the earlier snapshot or its scores. Marketplaces are calibrated to a common depth scale using specifications observed on multiple marketplaces. The fitted baseline is then smoothed across related specifications. A result-page limit can look like a supply ceiling, while a single observation can make one specification appear unusually abundant or scarce.

Recent observations correct that baseline and inform estimates for nearby specifications. Closer matches and better-supported observations carry more influence. Specifications without a fresh search still receive a prediction; missing data is never interpreted as zero supply.

An early version compared color and clarity combinations only within each carat band. That described unusual grades within a size category, but removed much of the difference between a well-supplied one-carat market and a thinner six-carat market. We kept those size-related supply differences in the score. Sources are aligned to a common depth scale before the model compares specifications across carats. A particular color-and-clarity combination at one carat may still be harder to find than a more common combination at six.

Reading the score

For weights between published carat points, such as 1.23ct, we interpolate the scarcity estimate between nearby points. Pricing is calculated separately, including its adjustments near carat thresholds; those price adjustments are not applied to scarcity.

The score runs from 0 to 100, with higher values indicating greater estimated market scarcity. Its definition is:

$$ R(i) = 100 \times \operatorname{PercentileRank}\left(-\hat{z}(i)\right) $$

Here, $\hat{z}(i)$ is the modeled log market depth for specification $i$. PercentileRank is expressed from 0 to 1 and compares the complete modeled grid across carats, colors, and clarities within the same shape/cut family. Each of the 960 reference combinations has equal weight; the ranking is not weighted by market demand or sales volume. Lower estimated depth gives a higher scarcity score. A score of 80 is neither a count of stones nor an 80% probability that a search will fail. A rank of 90 means lower modeled availability than approximately 90% of the 960 reference specifications.

Classification labels: The consumer interface and professional specification explorer share five human-readable labels and rank bands: Widely Available (0–<20), Moderately Available (20–<40), Limited Availability (40–<60), Rare (60–<80), and Very Rare (80–100). The market matrix retains numeric ranks for closer comparisons.

The professional page shows two views. Scarcity across carat sizes plots the median and 10th–90th percentile range across 80 color/clarity combinations. The shaded band represents variation between specifications, not model uncertainty. The scarcest-quarter frontier marks the earliest published carat band from which a specification remains at rank 75 or higher at every subsequent published band through 6 ct.

A buyer can use the score to decide whether to broaden a search. A jeweler can use it when discussing alternatives or the difficulty of finding a replacement stone. It does not promise a delivery time, establish a fair price, or make a harder-to-find stone a better investment.

The availability model supports two shape families—round and cushion—each ranked within its own reference grid. The professional market visualization currently shows round diamonds only; its 960 reference specifications are not ranked against cushion specifications. Other fancy shapes are not covered, and their availability is not borrowed from round or cushion models. Length-to-width ratio is not currently an input to the scarcity score.

The estimates depend on the retail listings we can observe. Vendor coverage, listing quality, repeated inventory, and search limits still affect the evidence. Modeling cannot recover every stone that was never visible to us.

Validation means withholding observed specifications, predicting them from the remaining evidence, and checking the errors. As fresh data accumulates, we also need to compare the scores with subsequent availability. Filling the grid is not proof of predictive accuracy.

Explore the market view, use the diamond price calculator, or read the pricing methodology for context on the underlying benchmarks.