Fundamental Desk
Cross-sectional commodity positioning
Where the 9 commodities stand relative to each other on the two fundamentals with the strongest published evidence: futures carry (term structure — backwardation vs contango) and CFTC COT positioning (commercial hedgers, 3-year z-score). Updated daily. Sorted by carry (most backwardated first).
| Commodity | Carry | COT z (3yr) | Signal | Inventory vs 5-yr |
|---|---|---|---|---|
| Platinum | +3.75% | -0.71 | bullish | — |
| WTI Crude Oil | +1.21% | -0.94 | bearish | -15.6% |
| Gold | +0.63% | 0.66 | bearish | — |
| Copper | +0.27% | -0.47 | bearish | — |
| Silver | -0.05% | -0.74 | bullish | — |
| Hard Red Winter Wheat | -1.62% | 0.62 | bearish | — |
| Palladium | -2.77% | -2.27 | extreme short | — |
| Corn | -2.78% | 0.95 | bearish | — |
| Natural Gas | -8.77% | 1.1 | bearish | +6% |
Carry = (front − next)/next futures price; positive = backwardation (tight supply, historically bullish per the Theory of Storage). COT z standardizes commercial-hedger net positioning against its 3-year history.
How these factors backtest
Cost-aware, out-of-sample results on our 40-year engine — the weak ones shown too, not just the winners.
| Strategy | Sharpe | OOS | Ann. | Verdict |
|---|---|---|---|---|
Cross-sectional COT — 4 precious metals COT positioning | 0.27 | 0.40 | +3.6% | Real but modest market-neutral edge; held out-of-sample (OOS Sharpe exceeds in-sample). The only validated fundamental edge so far. |
Cross-sectional COT + trend filter COT positioning | -0.10 | — | -2.6% | Adding a trend filter does NOT help — overlapping signals aren't additive. |
Inventory — nat-gas, long when tight Inventory / Theory-of-Storage | -0.44 | — | -63.5% | Weak — the time-series form of the inventory signal doesn't work, as the literature warns. |
Inventory — crude, long when tight Inventory / Theory-of-Storage | -0.53 | — | -37.4% | Weak — the time-series form of the inventory signal doesn't work, as the literature warns. |
Cross-sectional carry (term structure) Carry / basis | — | — | — | Highest-evidence signal in the literature (~10%/yr). Daily curve capture began 2026-06-28; backtest pending data accrual. |
As of 2026-10-01. Sharpe gross of fees; OOS = out-of-sample (2016+). Cross-sectional COT is dollar-neutral (market-neutral). Past performance does not guarantee future results.