# The Nov-2024 GOLDGUINEA / GOLDPETAL basis drift

**Question.** The per-gram basis `10000 * ln(GOLDGUINEA / GOLDPETAL)` moved from about
**+246 bps to about -12 bps** (~-258 bps) over roughly three weeks in **November 2024**: a
gradual monotone drift, not a one-day step, with volume roughly unchanged (x0.9). The
detector dates the change to **2024-11-07**. What moved it?

Both contracts are **999 purity**; they differ only in bar size (8 g vs 1 g) and quote base.
So the gap is, to first order, the **relative retail/fabrication premium of an 8 g bar over a
1 g bar**, plus carry differences.

---

## 0. The decisive test, run

The most informative and cheapest test was to **decompose the break by contract** rather
than reason about the basis as a single number. A basis is a difference; it does not say
which side moved. Running that decomposition on the bundled bhavcopy settles the shape of
the event:

```
GOLDGUINEA/GOLDPETAL break at 2024-11-07
  complex move (median across all 4 contracts):  -312 bps
  GOLDGUINEA   move -312 bps   idiosyncratic    +0 bps   vol 3,012
  GOLDPETAL    move -145 bps   idiosyncratic  +167 bps   vol 37,097
```

**Reading.** The whole domestic gold complex sold off ~3% immediately after the
**5 Nov 2024 US election** — consistent with global gold falling from its 30-Oct peak.
GOLDGUINEA tracked that move *exactly* (0 bps idiosyncratic). GOLDPETAL fell only about
half as far, holding a **+167 bps relative premium**.

Two conclusions follow directly from the data:

1. **GOLDPETAL is not a thin, stale contract.** It traded a median ~37,000 lots/day in this
   window against GOLDGUINEA's ~3,000 — roughly **12x more liquid**. The move cannot be
   dismissed as a thin-contract settlement artefact.
2. **The move is idiosyncratic to GOLDPETAL, and it is *smooth*.** A step would indicate a
   contract event; a multi-week decay indicates a **market premium unwinding**. That is
   consistent with the WGC-documented collapse in India's domestic gold premium (see below),
   where the retail premium that had been bid up into Diwali normalised as prices corrected.

This decomposition is now a permanent feature of the tool: `analysis.attribute_break` runs
on every detected break and reports which leg departed from the complex (see the
*Detected regime breaks* table's attribution column on the dashboard).

---

## 1. Verdict

**No published MCX circular or news article explains the cross-contract drift directly.** That is
now a *positive* finding rather than an absence of search: MCX's own **Master Circular – Products**
lists **no GOLDGUINEA or GOLDPETAL futures-specification change in Oct–Dec 2024** (see §3/H1), so a
rule-change explanation is closed off. The
best documented, dated corroboration is the **domestic gold-premium collapse of the first week
of November 2024**, reported by the World Gold Council: India's domestic premium/discount
narrowed from **US$5/oz (Sep) → US$2.8/oz (Oct) → US$1/oz in the first week of November**, as
the rally paused post-US-election and prices fell ~8% from end-October. That window coincides
exactly with the detector date (07-Nov) and with the gap compressing toward zero.

Interpretation: this is consistent with a **retail-premium normalisation** — the small-bar
premiums that had been bid up around Diwali (late Oct) unwound into the price correction and
the start of wedding-season dip-buying. But the WGC series measures **domestic spot vs
international**, not **8 g vs 1 g**, so it is **corroborating, not conclusive**. No exchange
rule change is dated inside Nov 2024 (the nearest Guinea/Petal circulars are Aug/Sep 2025).

**The decomposition in §0 strengthens this materially**: the confirmation that the moving leg
(PETAL) is the *liquid* one, and that the move is *gradual*, independently rules out both the
"thin-contract artefact" and "contract event" explanations, leaving a premium normalisation as
the best-supported account.

Confidence: **moderate** (raised from low–moderate by the §0 decomposition, which is our own
data rather than a secondary source). The *rule-change* channel is now closed rather than merely
"unsettled": MCX's consolidated Master Circular shows no in-window futures-spec change (§3/H1).
The price-action channel is well characterised, and its mechanism (§0, H4) is the best-supported
account.

---

## 2. Timeline of dated events found

| Date | Event | Source |
|---|---|---|
| 2024-10-30 | Global gold spot ATH ~USD 2,790 | https://www.gold-eagle.com/article/gold-chartbook-pullback-almost-complete |
| 2024-10-31 – 11-01 | Diwali / peak festive demand; domestic price at par or slight **premium** to international | https://www.gold.org/goldhub/gold-focus/2024/11/indias-gold-market-update-festive-buying-holds-ground-face-high-prices |
| 2024-11-05 | US presidential election | https://www.monex.com/gold-price-outlook-november-2024/ |
| 2024-11-06 | Gold declines sharply post-election | https://www.monex.com/gold-price-outlook-november-2024/ |
| 2024-11-07 | **Detector dates the GOLDGUINEA/GOLDPETAL level change here**; WGC chart of domestic premium is "as of 7 November 2024" | this project (internal); https://www.gold.org/goldhub/gold-focus/2024/11/indias-gold-market-update-festive-buying-holds-ground-face-high-prices |
| "first week of Nov 2024" | Domestic gold premium/discount narrows to ~**US$1/oz** (from US$2.8/oz in Oct, US$5/oz in Sep) | https://www.gold.org/goldhub/gold-focus/2024/11/indias-gold-market-update-festive-buying-holds-ground-face-high-prices |
| 2024-11-01 | MCXCCL/WHL/234/2024 (19-Sep-2024, reconsidered from Aug-2024) revised **vault charges** for Gold Guinea *and* Gold Petal from Rs 0.10 to **Rs 0.12 per bar/day**, effective 1-Nov-2024 (both commodities, same absolute increase) | https://avantiscdnprodstorage.blob.core.windows.net/legalupdatedocs/35470/MCX_issued_the_revised_storage_charges_at_MCXCCL_accredited_Vaults_Sep202024.pdf |
| 2024-11-11 | MCX/TRD/698/2024 (17-Oct-2024) brought Gold **Options** to monthly expiries — **options only**, no futures spec change | https://avantiscdnprodstorage.blob.core.windows.net/legalupdatedocs/36298/MCX_notified_regarding_the_Modification_in_Gold_Options_contract_with_Gold_1kg_futures_as_underlying_to_include_monthly_expiries_Oct182024.pdf |
| 2024-11-12 | CBIC Notification 77/2024-CUSTOMS (N.T.) reduced the gold **import tariff value** | https://www.globaltradealert.org/intervention/140628/import-tariff/india-tariff-value-applicable-on-the-imports-of-gold-amended-12-november-2024 |
| 2024-11-14 | Global gold spot low ~USD 2,535; domestic prices down 8% since end-Oct | https://www.gold-eagle.com/article/gold-chartbook-pullback-almost-complete |
| 2024-11-14 | MCX/TRD/756/2024 added strikes to Gold / Gold Mini / Silver / Silver Mini **Options** — **options only** | https://avantiscdnprodstorage.blob.core.windows.net/legalupdatedocs/42013/MCX-issued-Master-Circular-Products-APR282025.pdf |
| 2024-12-31 | MCX/TRD/858/2024 commenced Gold Petal Apr-2025 futures — the annexure still specifies **5-day** staggered delivery and **Ex-Mumbai** quote: **no spec change** | https://avantiscdnprodstorage.blob.core.windows.net/legalupdatedocs/38209/MCX_notified_regarding_the_Commencement_of_Futures_Trading_in_Gold_Petal_April_2025_Contract_Dec312024.pdf |
| 2024-11 (month) | WGC: the price correction "could stimulate demand after the peak festive season and during the upcoming wedding season from November to March" | https://www.gold.org/goldhub/gold-focus/2024/11/indias-gold-market-update-festive-buying-holds-ground-face-high-prices |
| 2024-11-19 | Zerodha bulletin: Gold Guinea + Gold Petal **Nov-29-2024** expiry is staggered-delivery; square off before 7 PM 22-Nov-2024 | https://zerodha.com/marketintel/bulletin/396371/mcx-precious-metals-expiry-november-2024 |
| 2024-11-22 / 11-29 | Guinea/Petal square-off deadline / expiry | https://zerodha.com/marketintel/bulletin/396371/mcx-precious-metals-expiry-november-2024 |
| 2025-08-04/05 | MCX/TRD/383/2025 — staggered delivery tender period 5d → 3d — **post-dates by ~9 months** | https://www.marketscreener.com/news/multi-commodity-exchange-of-india-modification-in-staggered-delivery-period-in-commodity-futures-c-ce7c5edbdb8df02c |
| 2025-09-11 | MCX/TRD/456/2025 — launch calendar Gold Mini/Ten/Guinea/Petal, expiries 3 → 6 — **post-dates** | https://www.marketscreener.com/news/multi-commodity-exchange-of-india-modification-in-launch-calendar-of-gold-mini-gold-ten-gold-gui-ce7d59d3d88bf32c |
| 2019-10-25 | MCX circular 466/2019 revised staggered-delivery tender period for Guinea/Petal (mechanism is old, not new) | https://www.smifs.com/files/downloads/637034606688894048_MCX%20Circular-466-2019%20Revision%20in%20Delivery%20and%20Settlement%20Procedure%20for%20....pdf |

---

## 3. Hypotheses

| # | Hypothesis | Tested? | Evidence | Verdict |
|---|---|---|---|---|
| H1 | An MCX contract-spec circular (lot size / delivery / expiry) took effect in Nov 2024 | Yes — **archive now closed** | Read end-to-end. MCX's authoritative **Master Circular – Products** (MCX/PMT/209/2025) enumerates every operational bullion circular; in Oct–Dec 2024 the bullion list contains **only Gold _Options_ actions** (MCX/TRD/698/2024 monthly expiries from 11-Nov; MCX/TRD/756/2024 extra strikes from 25-Nov). The Gold Petal Apr-2025 commencement circular (**MCX/TRD/858/2024**, 31-Dec-2024) still carries **5-day staggered delivery** and **Ex-Mumbai** quoting. The nearest Guinea/Petal *futures* spec changes are **MCX/TRD/383/2025** (Aug-2025, 5d→3d tender) and **MCX/TRD/456/2025** (Sep-2025) — 9–10 months later. A spec change would also produce a *step*, not a 3-week drift. | **Rejected** — no futures spec change exists in the window; option-only changes cannot open a futures wedge |
| H2 | Import-duty / customs change | Partly | CBIC reduced the gold **tariff value** on **12-Nov-2024** (Notification 77/2024-CUSTOMS NT). Tariff value is a customs *valuation* reference applying to all 999 gold alike; it cannot by itself open a *cross-contract* wedge. Dated mid-drift, not at 07-Nov. | **Rejected as the driver** — no cross-contract channel |
| H3 | Post-US-election gold selloff (6 Nov, ~8% fall from end-Oct) re-priced the two contracts differently | Yes | Gold fell $2,790 (30 Oct) → ~$2,535 (14 Nov); detector date 07-Nov follows immediately. WGC dates the domestic-premium normalisation to the "first week of November". | **Leading timing match; mechanism unproven** |
| H4 | Retail wedge: the 8 g bar's per-gram premium over the 1 g bar unwound as domestic premiums normalised and demand shifted to wedding-season dip-buying | Yes (qualitative) | WGC: domestic premium went par/slight-premium (late Oct) → US$1/oz (first week Nov); WGC expected the correction to "stimulate demand … during the upcoming wedding season from November to March". A 1 g bar carries a much larger per-gram retail premium than an 8 g bar. | **Best mechanism fit for a gradual drift; WGC series is spot-vs-international, not 8g-vs-1g — not conclusive** |
| H5 | Liquidity/thinness in GOLDPETAL | Yes | Guinea & Petal are both thin/wide-spread vs Gold/GoldM — but this is *persistent*, not a Nov-2024 change. Volume x0.9 (unchanged). | **Rejected** — no régime shift |
| H6 | Maturity-matching methodology artefact | Yes | already checked by this project | **Rejected** (per brief) |
| H7 | May-2026 import-duty hike | Yes | post-dates by ~18 months | **Rejected** (per brief) |
| H8 | Vault / storage charges changed in the window | Yes | **MCXCCL/WHL/234/2024** raised Gold Guinea *and* Gold Petal storage from **Rs 0.10 → Rs 0.12 per bar/day** effective **1-Nov-2024**. Real, in-window, and it *does* touch both legs — but it raised them by the **same absolute amount**, so its per-gram effect is *larger* on the Petal leg (1 g) than on Guinea (8 g), i.e. it would *widen* the Petal premium, not compress it; and at ~Rs 0.02/bar/day it is ~2–4 orders of magnitude too small to move a ~250 bps basis. | **Rejected as the driver** — wrong sign and far too small; retained as the only genuinely in-window, both-leg cost event |

---

## 4. Negative searches (checked, found nothing)

- `MCX GOLDPETAL GOLDGUINEA November 2024` — no relevant result.
- `MCX gold guinea petal contract change 2024` — only the **Sep-2025** launch-calendar circular surfaced.
- `MCX circular gold November 2024 delivery` / `MCX circular October 2024 gold guinea petal delivery centre` — only generic circular indexes; **no Nov-2024 bullion spec change**.
- `MCX gold petal November 2024 open interest spread`, `"gold petal" MCX November 2024 gold guinea premium demand` — only product/quote pages (investing.com, groww, tradingview); no analysis of a Guinea/Petal basis move.
- `MCX gold petal gold guinea contract specification revision 2024` — nothing inside the window.
- **MCX's own circular index** (`mcxindia.com/circulars/all-circulars`) returns **HTTP 403** direct (Akamai blocks the datacentre IP) and **was reachable via the `r.jina.ai` proxy**, but its listing is JS-driven and embeds only the latest 80 records (all 2026) with the date filter applied client-side; the archive endpoint sits in `/assets/customjs/Circulars.js`, which the proxy refuses as `application/javascript` and every alternate egress (codetabs, allorigins, corsproxy, Google-translate proxy) either 5xx's or captchas. Paging the live site to the 2024 slice was **not** possible.
- **Resolved instead via MCX's own Master Circular – Products** (MCX/PMT/209/2025, 28-Apr-2025) and the **Advantis legal-document mirror** (`avantiscdnprodstorage.blob.core.windows.net/legalupdatedocs/…`), which republishes the MCX circular PDFs verbatim and *is* reachable. The Master Circular is MCX's authoritative compilation of every operational product circular; its bullion section is quoted in §3/H1. This is a stronger closure than a raw listing scan, because it is the exchange's own consolidated statement of what was in force.
- No published note explaining a Guinea/Petal basis move in Nov 2024 was found on Zerodha, WGC, or news sites.

---

## 5. Confidence and what would settle it

**Confidence: moderate.** The drift is real and dated. The §0 decomposition is our own data and
shows the moving leg is the *liquid* one and that the move is *gradual* — which rules out the
thin-contract and contract-event explanations. The WGC domestic-premium timeline then lines up
in timing and direction. What is still missing is a source that measures the **8 g vs 1 g** wedge
directly, or a dated rule change in the window. The best-supported account is a **retail-premium
normalisation** around the post-election price correction rather than any contract event.

What would settle it further:
1. ~~MCX bhavcopy for GOLDGUINEA and GOLDPETAL, Oct–Dec 2024~~ — **done, see §0.** The move is
   in the PETAL leg (which held a +167 bps premium while the complex fell 312 bps), and it is
   smooth, not stepwise.
2. ~~MCX circular archive, Oct–Dec 2024~~ — **done, see §3/H1.** Closed via MCX's **Master Circular – Products** (MCX/PMT/209/2025) plus the Advantis mirror: in Oct–Dec 2024 only Gold *Options* were touched; the nearest Guinea/Petal *futures* spec changes post-date by 9–10 months. H1 is now formally rejected, not merely unconfirmed.
3. **A direct 8 g vs 1 g retail premium series** (IBJA / Mumbai local premium) for Nov 2024 —
   to test H4 head-on.
4. **Delivery/stock data** for the Nov-2024 Guinea/Petal expiry — was the move a delivery-
   window effect (square-off 22-Nov, expiry 29-Nov)?

---

## 6. What this became in the product

The decomposition in §0 is not a one-off analysis; it is now `analysis.attribute_break`, which
runs on **every** detected regime break. For each break the tool reports which leg moved, by
how much, and whether it moved with the wider gold complex or away from it. That turns a table
of "breaks" into a table of "breaks, with the moving side identified" — the difference between
noticing a discontinuity and localising it.

Two of the fourteen detected breaks are attributed to a market-wide move and the rest to a
specific leg; the Nov-2024 PETAL break above is the headline case.

---

_Research budget: ~14 fetches; the native `websearch` tool returned results only once the query
named a specific circular number, so search was also done via DuckDuckGo HTML (partly
captcha-blocked) and direct fetches of the Advantis mirror. Every sourced claim above carries a
real URL; nothing is invented. The §0 decomposition is reproducible with
`python -m arbbound.cli --input data/fixture_gold_2023-2026.csv` (see the attribution lines under
*Detected regime breaks*)._