Silver sheds 6.14% over five sessions, gold barely moves
Kitco's board, frozen at its 16 October tick, reads silver at $56.866 against this paper's $60.584 mark of 12 October. Gold gave up $7.80 over the same stretch, and the gold-silver ratio widened from 68.86 to 73.23.
By Carat Capital
What changed
Silver did all the moving. Against this paper's marks of 12 October, silver is down 6.1369% and gold down 0.187%. The ratio between them widened from 68.8614 to 73.2265.
What it means
- If you hedge a silver book: The $3.718 came off across five sessions, not one. The board has not opened since its 16 October tick.
- If you price scrap on a ratio: Gold now buys 73.2265 ounces of silver against 68.8614 five sessions ago, on this paper's division of its own marks.
- What this is not: A close, a fix or a settlement. These are spot bids off a halted board, and two of the paper's reference indices returned no value at all.
Key figures
| Silverdown $3.718, or 6.1369%, on the 12 October mark | $56.866 |
|---|---|
| Golddown $7.80, or 0.187% | $4,164.10 |
| Gold-silver ratiofrom 68.8614, on our division | 73.2265 |
| Platinumdown $67.00, or 3.9575% | $1,626.00 |
Source: Carat Capital's press-time reading of the Kitco board at 22:01 EDT on 17 October 2026, raw bytes saved, 200, 197,364 bytes. All four figures are the bid, the convention this paper has carried since 31 August 2026. The board's own last tick is stamped 2026-10-16T17:00:00Z on every metal. Prior marks are this paper's own printed marks of 12 October, looked up with a script rather than derived.
The board is halted
One timestamp, four metals. Every metal carries Kitco's originalTime of 2026-10-16T17:00:00Z, the venue's own last tick. The page is serving that tick, not a new one.
A re-read confirmed it. A press-time fetch at 22:01 EDT returned the identical four bids and the identical timestamp. The halt is read off the page, not inferred from the weekend.
Five sessions, not one. These changes run from this paper's own marks of 12 October, the last tape filed. No mark was taken between.
| Metal | 12 October mark | Press time | Change | Change % |
|---|---|---|---|---|
| Gold | $4,171.90 | $4,164.10 | −$7.80 | −0.187% |
| Silver | $60.584 | $56.866 | −$3.718 | −6.1369% |
| Platinum | $1,693.00 | $1,626.00 | −$67.00 | −3.9575% |
| Palladium | $1,140.00 | $1,096.00 | −$44.00 | −3.8596% |
| Lab-grown 1ct | $665 | $663 | −$2 | −0.3008% |
| RAPI 1.00ct | 3,903.60 | unreachable | - | - |
| Zimnisky rough | no prior mark | paywalled | - | - |
Notes on this table
Every change is computed against Carat Capital's own printed mark of 12 October, looked up rather than reversed out of a change field, and the division is printed beside the sentence so it can be compared. Gold: 4,164.10 − 4,171.90 = −7.80; −7.80/4,171.90 = −0.187%. Silver: 56.866 − 60.584 = −3.718; −3.718/60.584 = −6.1369%. Platinum: 1,626.00 − 1,693.00 = −67.00; −67.00/1,693.00 = −3.9575%. Palladium: 1,096.00 − 1,140.00 = −44.00; −44.00/1,140.00 = −3.8596%. Ratio: 4,171.90/60.584 = 68.8614 and 4,164.10/56.866 = 73.2265, a widening of 4.3651 points, which is 4.3651/68.8614 = 6.339%. Kitco's own change fields read −5.70, −0.434, +6.00 and −18.00; those are its one-session numbers on its own prior close and are not the changes printed here, because the basis is different.
One leg moved
Silver gave up 6.1369%. The bid fell from $60.584 to $56.866, a loss of $3.718. Nothing in the saved board says why.
Gold stood still. A move of $7.80 on $4,164.10 is 0.187%, which is narrower than the board's own quoted bid-ask spread on the metal.
The ratio widened. Gold divided by silver went from 68.8614 to 73.2265, up 4.3651 points or 6.339% on this paper's own division of its own marks.
Two indices gave nothing
Rapaport refused three times. The one-carat index returned HTTP 403 to a tool request, a browser identity and an identifying header. It is carried as unreachable with no value.
No prior is restated. The 12 October mark of 3,903.60 is not printed as today's level. An unverified price is carried as null in this paper, never forward.
Zimnisky opened and withheld. The rough index page returned 200 at 133,273 bytes and publishes its level only inside a chart image. Status is paywalled, not unreachable.
What to watch
- Monday, 05:00 ETThe first live tick after the halt. Every figure here is a Friday-evening last price carried through a closed weekend.
- MondayWhether Rapaport's index answers. Three identities were refused tonight and the host served this paper cleanly on 12 October.
- OngoingWhether the ratio holds above 73. It has moved 6.339% in five sessions on silver's leg alone.
The story so far
Go deeper
What would change this call
A live session. Everything here is a halted board's last tick, and the first hour of Monday trade can erase a five-session move in one leg. A second page would also harden it: this reading rests on Kitco alone because the halt makes a two-page corroboration test close to meaningless, as a frozen quote will agree with anything near it. The prior marks are this paper's own, which means a fault in the 12 October tape would propagate into every change printed here.
What a five-session gap does to a change column
This paper's change column normally answers one question: what happened since yesterday. Tonight it answers a different one, because no mark was taken between 12 and 17 October, and the column spans five sessions rather than one. That distinction is not cosmetic. A single-session change is a statement about a day's trading and can reasonably be attributed to the day's news. A five-session change is an accumulation, and attributing it to anything in particular requires the intervening closes, which this paper does not hold. So the figures are printed as what they are, differences between two marks five sessions apart, and no cause is offered for any of them. The second thing the gap does is make the venue's own change field useless for comparison. Kitco publishes a change against its own prior close, which is one session, and that is why its silver field reads minus 43.4 cents while this paper's reads minus $3.718. Neither is wrong. They measure different intervals, and a reader comparing them without knowing that would conclude one of the two sources was broken. The habit this house keeps, of looking the prior mark up in its own tape rather than reading a change field off the page, exists precisely so the interval is always the one the paper claims it is.
Method
One page, fetched at press time in this run and saved as raw bytes before any figure was taken: kitco.com/price/precious-metals, 200, 197,364 bytes. Figures are read from the page's own embedded JSON payload, which gives a bid, ask, mid and originalTime per metal: gold bid 4164.1, silver 56.866, platinum 1626, palladium 1096, all stamped 2026-10-16T17:00:00Z. MARKET STATE READ OFF THE BOARD, not inferred from a calendar: the identical originalTime on all four metals, returned twice from two separate fetches ninety minutes apart, is the evidence for the halt, and no exchange holiday is claimed. Under the One-Price Rule the press-time reading was compared against the dated tape file as caratwire filed it at 21:47 EDT and all four bids match to the cent, so no write-back was required and the tape stands as filed; the live tape file carries the same four values. Prior marks were looked up with a script, which names the dated tape file as the file it read, and are not reversed out of any change field. The lab-grown reference is CaratRadar's one-carat average at $663, taken from the 116,865-byte capture in this run's wire folder, whose own stated date is 18 October 2026, one day ahead of this machine's clock and printed as the source gave it. The RAPI and Zimnisky statuses are caratwire's three-knock and re-knock results from the same run.
Sources
- Kitco precious metals board, press-time re-read 22:01 EDT (bid) 17 October 2026
- CaratRadar lab-grown diamond prices, one-carat average 17 October 2026
- Carat Capital, Two holds end at press time 12 October 2026