Normative reference

Caratlytics Scoring Specification

Caratlytics Standard CTL-1 Version 1.0 Effective date: 2026-05-09

This document is the normative definition of the Caratlytics Score, methodology version 1.0. It is the canonical reference for every weight, lookup table, gate and edge rule in the scoring engine. The consumer summaries on Carat Hunter are descriptions of this document; where they differ, this document governs.

§1Scope and definitions

The Caratlytics Score is a composite measure on a 0 to 100 scale assigned to a diamond listing: a specific stone offered by a specific retailer at a specific price. Stone level pillars (Quality, Certification) attach to the diamond; market level pillars (Value, Market) attach to the listing.

All prices are expressed in USD using a pinned set of conversion rates that serve as the platform's internal unit of account. Pinned rates keep the time series longitudinal: a score or index movement reflects a retailer repricing, never a currency fluctuation.

§2The composite

The overall score is the weighted arithmetic mean of the four pillar scores. Weights are fixed per methodology version. If a pillar score is unavailable, its weight is redistributed proportionally across the available pillars.

Table 2.1: pillar weights
Pillar	Weight
Quality	35%
Value	30%
Certification	20%
Market	15%

Overall = Quality x 0.35 + Value x 0.30 + Certification x 0.20 + Market x 0.15

§3Quality pillar

The quality score is a weighted sum of up to nine components, each scored 0 to 100 from the tables below, divided by the maximum total weight of 100. Components missing from the stone's record contribute zero to the numerator: the denominator never shrinks. A quality score is only issued when present component weights total at least 60.

When cut is ungraded (common for fancy shapes), its 25 point weight partially redistributes: proportions rise from 12 to 27 points. If proportions are also unavailable, polish and symmetry each rise from 8 to 20 points.

Fancy color diamonds reweight: color carries 30 points (scored on the fancy intensity scale below) and clarity falls to 8, reflecting that color dominates value for fancy stones.

Table 3.0: quality component weights
Component	Weight
Cut	25
Color	20 (30)
Clarity	18 (8)
Proportions	12 (27)
Polish	8 (20)
Symmetry	8 (20)
Fluorescence	5
Culet	2
Girdle	2

Table 3.1: cut, polish and symmetry grades (shared scale)
Grade	Score
Ideal (ID)	100
Excellent (EX)	95
Very Good (VG)	75
Good (GD)	50
Fair (FR)	25
Poor (PR)	10

Table 3.2: color grades, D to Z scale
D	E	F	G	H	I	J	K	L	M	N	O	P	Q	R	S	T	U	V	W	X	Y	Z
100	96	92	86	80	72	64	55	45	35	25	20	16	13	10	8	7	6	5	4	4	3	3

Table 3.3: fancy color intensity scale
Grade	Score
FANCY VIVID	95
FANCY INTENSE	90
FANCY DEEP	85
FANCY DARK	80
FANCY	75
FANCY LIGHT	65
LIGHT	55
VERY LIGHT	45
FAINT	35

Table 3.4: clarity grades
FL	IF	VVS1	VVS2	VS1	VS2	SI1	SI2	SI3	I1	I2	I3
100	97	92	87	80	73	60	50	40	25	15	5

Table 3.5: fluorescence
Grade	Score
None (NON)	100
Faint (FNT)	95
Medium (MED)	70
Strong (STR)	40
Very Strong (VST)	20

Table 3.6: culet
Grade	Score
None (NON)	100
Very Small (VSM)	95
Small (SML)	80
Medium (MED)	60
Large (LRG)	30
Very Large (VLG)	10

Table 3.7: girdle (single values; ranges average their endpoints, then a spread penalty of 0.15 points per point of spread applies, capped at 15)
Grade	Score
Extremely Thin (ETN)	20
Very Thin (VTN)	40
Thin (THN)	70
Slightly Thin (STN)	85
Medium (MED)	100
Slightly Thick (STK)	90
Thick (THK)	70
Very Thick (VTK)	40
Extremely Thick (ETK)	20

Within the window a proportion scores 100. Outside it, the score falls by 4 points per percentage point of deviation from the nearer edge, floored at 10. Table and depth are scored separately and averaged.

Table 3.8: ideal proportion windows by shape (table % and depth %)
Shape	Table min	Table max	Depth min	Depth max
Round	54	58	59	62.5
Princess	64	75	64	75
Cushion	58	68	58	68
Oval	53	63	58	62
Emerald	58	69	61	67
Pear	53	63	56	62
Marquise	53	63	56	62
Radiant	61	69	62	67
Asscher	58	65	61	67
Heart	53	63	56	62
SQ_RADIANT	61	69	62	67
Trillion	50	60	38	48
Baguette	55	75	60	75
Other	54	70	56	70

§4Value pillar

The value score positions a listing's price per carat against the statistical baseline for its category: the combination of shape, carat band, color band and clarity band. Baselines store the category's quartiles and are recomputed daily.

Table 4.1: value components
Component	Weight	Label
price percentile	50	Price percentile (50 points): below the first quartile scores 75 to 100, scaling with distance toward Q1 minus 1.5 IQR; between Q1 and the median scores 50 to 75; between the median and Q3 scores 25 to 50; above Q3 decays toward a floor of 5 at Q3 plus 1.5 IQR.
price vs median	25	Price versus median (25 points): a ratio at or below 0.8 of the category median scores 100, scaling linearly down to 0 as the ratio reaches 1.5.
carat threshold	15	Carat threshold (15 points): prices step up at the industry's magic weights. A stone just under a threshold (for example 0.97 carat against the 1.00 threshold) earns a bonus of up to 12 points above the neutral 50; a stone just over earns a small penalty of up to 5 points below it.
cross retailer position	10	Cross-retailer position (10 points): when the same certificate is listed by two or more retailers, the cheapest listing scores 100 and the most expensive scores 20, scaled linearly. Single listed stones score a neutral 50.

Table 4.2: carat thresholds (threshold, maximum bonus points, bonus zone below, penalty zone above)
Threshold (ct)	Max bonus	Zone below	Zone above
0.50	5	0.03	0.02
0.70	4	0.03	0.02
0.75	5	0.03	0.02
1.00	12	0.04	0.03
1.50	8	0.04	0.03
2.00	10	0.04	0.03
2.50	6	0.03	0.02
3.00	8	0.04	0.03
4.00	6	0.04	0.03
5.00	6	0.04	0.03

Listings from vendor feeds whose prices fall below a plausibility floor of 300 USD per carat are treated as pricing artifacts. Their value score is neutralised at 50 so an implausible price can never rank as a bargain.

§5Certification pillar

The certification score measures trust in the stone's documentation: which laboratory graded it, how complete the published data is, and whether a certificate number is present and on record.

Table 5.2: laboratory reputation scores
Laboratory	Score
GIA	100
AGS	98
GCAL	90
HRD	85
IGI	80
CGL	75
EGL	50
GSI	50

A laboratory not in the table scores 60. No laboratory at all scores 30.

Data completeness (40 points) is the weighted share of 18 fields present on the record, using the weights in Table 5.3.

Table 5.3: data completeness field weights
Field	Weight
carat	10
color	10
clarity	10
cut	8
polish	6
symmetry	6
fluorescence	5
table_pct	5
depth_pct	5
length_mm	3
width_mm	3
depth_mm	3
girdle	3
culet	2
ratio	2
lab	8
certificate_number	6
origin	5

Certificate availability (20 points): certificate number with a certificate document on record scores 100; number plus named laboratory scores 80; number alone scores 60; laboratory alone scores 40; neither scores 10.

Integrity gate: a record claiming both a certificate number and a laboratory but missing color or clarity is capped at a certification score of 30. The claim and the evidence must agree.

§6Market pillar

The market score reads the listing's market context from cross-retailer matching and the listing's own price history.

Table 6.1: market components
Component	Weight	Label
cross retailer verification	35	Cross-retailer verification (35 points): 1 retailer scores 30; 2 scores 65; 3 scores 80; 4 scores 90; 5 scores 95; 6 or more scores 100. Independent listings of the same certificate corroborate that the stone is real and on the market.
price trend	30	Price trend (30 points): the 30 day price movement of the listing. A drop of 10 percent or more scores 95; drops of 5 to 10 percent score 85; 2 to 5 percent score 75; 0 to 2 percent score 65; flat to plus 2 percent scores 50; rises score progressively lower, to 10 above plus 10 percent.
price stability	20	Price stability (20 points): the coefficient of variation of the listing's price history. At or below 0.01 scores 100; 0.03 scores 85; 0.05 scores 70; 0.10 scores 50; 0.20 scores 30; above that, 15.
savings potential	15	Savings potential (15 points): a listing at the cross-retailer minimum (within 1 percent) scores 100; within 5 percent of it scores 80; within 10 percent scores 60; within 20 percent scores 40; beyond that, 20. Stones listed at a single retailer score a neutral 50.

§7Grade presentation

Raw scores are presented to consumers as six letter grades. The engine also annotates raw scores with an eight step descriptive label. Both scales are fixed and published.

Table 7.1: consumer grade bands
Grade	Range	Label
A+	90 to 100	Exceptional
A	80 to 89	Excellent
B+	70 to 79	Very Good
B	60 to 69	Good
C	50 to 59	Average
D	0 to 49	Below Average

Table 7.2: descriptive labels on raw scores
From score	Label
90	Exceptional
80	Excellent
70	Very Good
60	Good
50	Average
40	Below Average
30	Fair
0	Poor

§8Limitations

The score reflects the data observed on the Carat Hunter platform, not the universe of all diamonds. A category with thin coverage has a thinner baseline.
Prices are snapshots at collection time, not live quotes. The platform recollects continuously, but a retailer can reprice between observations.
The score measures the stone and its market position. It does not measure retailer service quality, shipping, returns or setting craftsmanship.
Lab-grown price spreads of 80 to 150 percent across retailers are normal market behaviour, not data error, and the methodology treats them as genuine price dispersion.

§9Conformance

An implementation conforms to this specification if, given identical inputs, it reproduces the engine's scores exactly. The reference implementation is the production scoring engine; the interactive widget on this site is a verified port whose parity is asserted against engine generated fixtures.

See worked examples on real GIA stones