Commodity Factor Replication

Domestic commodity futures | 2010-01-04 to 2022-02-28

Commodity Factor Replication

A public replication framework for testing whether commodity term structure, trend, inventory, position, and volatility signals retain cross-sectional information after a liquid, changing trading universe is imposed.

Aggregate evidence only. No raw data, private database, credentials, or licensed source report is published.

  1. 01 Licensed market inputs
  2. 02 Dynamic tradable universe
  3. 03 Five factor families
  4. 04 Friday quintiles
  5. 05 Next-period returns and IC
13.0%Front/next annualised return
1.84Front/next gross Sharpe
0.071Front/next mean weekly IC
571Weekly IC observations

Research design

Dynamic universe, weekly decision rule

A product enters only after its listing-age and rolling-liquidity rules are met. The daily main and sub-main contracts are approximated by the two highest-volume listed contracts. Friday snapshots form the cross-section.

Signals are sorted into five balanced groups. The portfolio begins earning returns on the next trading date, and the long-short return is half of Group 1 minus Group 5. Weekly IC is Spearman correlation between Friday signals and the next holding-period return.

Read methodology
Stacked chart showing the number of products in the daily tradable universe over time.
Bilingual replication artifact: this author-generated chart retains its original Chinese on-chart labels for fidelity, with English methodology and interpretation provided alongside it.

Curated results

Evidence across factor families

StrategyReturnVolatilityMax drawdownSharpe
Main/sub-main carry, K=909.5%7.0%-11.9%1.35
Front/next carry, K=9013.0%7.1%-9.1%1.84
Spot/main basis, K=508.1%6.4%-11.5%1.27
Warehouse receipts, K=906.2%7.7%-15.4%0.81

Gross historical backtests. Transaction costs, slippage, financing, capacity, price limits, and roll execution are not modelled.

Annual return robustness heatmap for the front-next term-structure factor.
Bilingual replication artifact: Chinese plot labels are preserved; this heatmap reports term-structure robustness across lookback parameters.
Annual return robustness heatmap for the warehouse receipt factor.
Bilingual replication artifact: Chinese plot labels are preserved; this heatmap reports warehouse-receipt robustness across lookback parameters.
Heatmap of correlations between weekly information coefficients for available factors.
Bilingual replication artifact: Chinese plot labels are preserved; this heatmap shows IC correlation among available factors.
Net asset value comparison for available multi-factor combinations under 100-score ranking.
Bilingual replication artifact: Chinese plot labels are preserved; this chart compares available multi-factor NAVs under the 100-score method.
View result notes and provenance

Interpret carefully

Replication evidence, not an investment claim

Vendor coverage, contract mapping, spot coverage, warehouse receipts, and main-contract rules are local approximations. Some multi-factor components are unavailable and are marked as missing instead of inferred.

Read all limitations

Public package

  • Tested research modules and synthetic tests
  • Local data contract and database schema
  • Aggregate tables and author-generated figures
  • Author's Chinese supplementary report