MCX cross-contract gold — is the spread an opportunity, or the cost of closing it?
Of 280 flagged days, 279 sat inside the exchange's own printed high/low bands with sufficient volume. Where the structural level itself moved, the flagged days are reported as a level shift, not as an opportunity.
Tested walk-forward on 6 contract pairs: the centring level is fitted on the training window only, then frozen and applied unchanged to data the test never saw. 4 of 6 pairs held a stable structural level.
▶ Don't take our word for it — run it yourself. The live playground executes this exact pipeline in your browser on input you supply.
Minimum gross gap, in basis points, before a cross-contract spread can profit. The wedge is a contractual purity/duty difference and is not tradeable.
| pair | wedge (bps) | structural | bound band |
|---|---|---|---|
| GOLDGUINEA/GOLDM | 240 | 244 | 256-334 |
| GOLDGUINEA/GOLDPETAL | 0 | 28 | 40-118 |
| GOLDGUINEA/GOLDTEN | 0 | 1 | 13-91 |
| GOLDM/GOLDPETAL | 240 | 272 | 284-362 |
| GOLDM/GOLDTEN | 240 | 243 | 255-333 |
| GOLDPETAL/GOLDTEN | 0 | 29 | 41-119 |
The centring level is fitted on the training window only and then frozen. Where the level itself moved, that is reported as a level shift, not as an opportunity.
| pair | verdict | test breaches | cleared | detail |
|---|---|---|---|---|
| GOLDGUINEA/GOLDM | BOUND_EXCEEDED_IN_TEST | 16 | 10 | level stable max residual +691 bps vs bound 256-334 |
| GOLDGUINEA/GOLDPETAL | BOUND_EXCEEDED_IN_TEST | 225 | 213 | level shift +150 bps train level +140 bps vs test median -10 bps |
| GOLDGUINEA/GOLDTEN | NOT_ARBITRAGABLE | 12 | 0 | level shift -14 bps train level +58 bps vs test median +45 bps |
| GOLDM/GOLDPETAL | BOUND_EXCEEDED_IN_TEST | 39 | 21 | level stable max residual +815 bps vs bound 284-362 |
| GOLDM/GOLDTEN | NOT_ARBITRAGABLE | 0 | 0 | level stable max residual -145 bps vs bound 255-333 |
| GOLDPETAL/GOLDTEN | NOT_ARBITRAGABLE | 1 | 0 | level stable max residual +4 bps vs bound 41-119 |
A structural cost cannot jump. These discontinuities are events, not arbitrage. The last column decomposes each break: which leg moved, and whether it moved with the wider gold complex or away from it.
| pair | date | level shift | delta | class | volume | attribution |
|---|---|---|---|---|---|---|
| GOLDGUINEA/GOLDM | 2026-05-06 | +64 → +31 | -33 | structural | x1.0 | GOLDM moved +58 bps away from the complex |
| GOLDGUINEA/GOLDM | 2025-12-17 | +58 → +167 | +109 | step-change | x1.0 | GOLDM moved -61 bps away from the complex |
| GOLDGUINEA/GOLDM | 2025-04-09 | +70 → +32 | -39 | structural | x2.1 | GOLDM moved +52 bps away from the complex |
| GOLDGUINEA/GOLDM | 2024-04-12 | +45 → +79 | +34 | step-change | x1.2 | GOLDM moved -65 bps away from the complex |
| GOLDGUINEA/GOLDPETAL | 2024-11-07 | +246 → -12 | -258 | structural | x0.9 | GOLDPETAL moved +167 bps away from the complex |
| GOLDGUINEA/GOLDPETAL | 2025-02-14 | +103 → -18 | -121 | structural | x1.2 | GOLDPETAL moved +98 bps away from the complex |
| GOLDGUINEA/GOLDTEN | 2025-10-01 | +33 → +64 | +31 | liquidity-event | x2.0 | both legs moved with the wider gold complex |
| GOLDGUINEA/GOLDTEN | 2026-03-25 | +118 → +44 | -74 | structural | x0.9 | GOLDGUINEA moved -61 bps away from the complex |
| GOLDM/GOLDPETAL | 2024-10-28 | +169 → -53 | -223 | structural | x1.1 | GOLDPETAL moved +70 bps away from the complex |
| GOLDM/GOLDPETAL | 2025-02-14 | +37 → -63 | -100 | structural | x1.2 | GOLDPETAL moved +98 bps away from the complex |
| GOLDM/GOLDPETAL | 2026-07-24 | -69 → -24 | +45 | step-change | x1.1 | both legs moved with the wider gold complex |
| GOLDM/GOLDPETAL | 2026-05-08 | -80 → -41 | +39 | step-change | x1.0 | GOLDM moved +51 bps away from the complex |
| GOLDM/GOLDTEN | 2025-12-29 | +18 → -23 | -42 | structural | x1.0 | GOLDM moved -77 bps away from the complex |
| GOLDPETAL/GOLDTEN | 2026-08-04 | +75 → +31 | -44 | structural | x1.5 | both legs moved with the wider gold complex |
| GOLDPETAL/GOLDTEN | 2026-05-11 | +87 → +58 | -29 | structural | x0.8 | both legs moved with the wider gold complex |
| GOLDPETAL/GOLDTEN | 2025-10-29 | +67 → +122 | +55 | step-change | x0.7 | both legs moved with the wider gold complex |
| GOLDPETAL/GOLDTEN | 2025-07-18 | +78 → +47 | -31 | structural | x1.4 | both legs moved with the wider gold complex |
A bhavcopy print is a settlement price, not a quote you can lift. Each flagged day is re-tested against the exchange's own printed high/low bands and liquidity.
| pair | flagged | reachable |
|---|---|---|
| GOLDGUINEA/GOLDM | 16 | 16 |
| GOLDGUINEA/GOLDPETAL | 225 | 224 |
| GOLDM/GOLDPETAL | 39 | 39 |
Two inputs are assumptions, not measurements: the per-bar minting cost (MCX does not publish it) and the import-duty wedge (policy-set, and it has moved). Both are swept across their full range here. The column on the right is how many basis points the bound itself moves — that is the part of our claim resting on an assumption rather than on the contract specification.
| pair | structural (bps) | total bound (bps) | moves by |
|---|---|---|---|
| GOLDGUINEA/GOLDM | 42 – 446 | 93 – 497 | 404 |
| GOLDGUINEA/GOLDPETAL | 14 – 42 | 65 – 93 | 28 |
| GOLDGUINEA/GOLDTEN | 0 – 1 | 51 – 52 | 1 |
| GOLDM/GOLDPETAL | 56 – 488 | 107 – 539 | 432 |
| GOLDM/GOLDTEN | 42 – 445 | 93 – 496 | 403 |
| GOLDPETAL/GOLDTEN | 15 – 44 | 66 – 95 | 29 |