RAPI's 1-carat adds 0.43% in a month, the 0.30-carat 3.3%

A longer series. One month of daily points cannot distinguish a market that has stopped falling from a pause inside a fall, and the band figure would look different on any other window. Rapaport's own monthly series, which runs back years, is the instrument that could settle it, and it is not published free on this page. A restatement of the September points, which this paper has already documented once, would also move every figure here by cents.

A first-to-last change is the weakest thing you can compute from a series, because it throws away every point in between and answers only what the two endpoints happened to be. Two indices that both finish a month up half a percent can have behaved entirely differently: one may have drifted quietly, the other swung two percent in each direction and landed back near where it started. For a trade that has to decide whether to hold inventory or move it, that difference is the whole question, and the change column does not carry it. The high-to-low band does. It states, in one figure, the widest the instrument got over the window, which is the nearest thing a published index gives to a volatility reading. Here the two measures agree, which is itself the finding: the one-carat's change is small and its band is small, so the flatness is real rather than an artefact of where the month happened to begin and end. The 0.30-carat's band, at 4.46%, is wider than its own 3.30% change, which says the small end both moved and wobbled. A reader who takes only the change column would conclude the two sizes behaved similarly in kind and differently in degree. They did not.

The series was read from Rapaport's diamond price index page, fetched direct with a browser user agent, 200, 306,089 bytes to a saved file. The five size blocks are carried in the page's own embedded data as dated point arrays and were parsed programmatically from those raw bytes rather than read off a chart or a summary; '3915.85' returns 2 hits in the saved file. Each block gives 32 dated points from 30 August to 30 September inclusive. The change column here was recomputed independently from each series' own first and last point and compared against the changePct field the page publishes beside it: 0.430 against 0.43, 3.301 against 3.3, 2.590 against 2.59, 0.660 against 0.66 and 1.370 against 1.37, so the page's own arithmetic and this desk's agree on all five. The band column is this desk's, computed as the maximum point over the minimum point less one, with the dates of both extremes recorded. The page's description of the index is quoted at six words and is the page's own claim about itself, not a verified one. The lab-grown comparison comes from CaratRadar, fetched separately this morning, 200, 116,862 bytes, whose own sentence gives a 1ct average of $651 as of 30 September 2026 alongside 1,163,879 active listings; '651' returns 3 hits in those bytes and '650' returns none. The 16.6% division is 651 over 3,915.85 and is this desk's. No figure here is carried forward from a prior filing. The archive was checked by reading every hit rather than counting: searching RAPI and RapNet returns 21 articles and the three nearest were opened in full. The 7 September piece covers 8 August to 6 September and the transition from monthly to daily publication; the 21 September piece covers the five sessions from 16 to 20 September; the 22 September piece covers the restatement of those same five points. This piece covers 30 August to 30 September across all five published sizes. Same instrument, different period and different scope, so the verdict is advanced, not a repeat.

1. Rapaport Trade Price Index (RAPI), daily series by size 30 September 2026
2. Lab Grown Diamond Prices, CaratRadar 30 September 2026