Commodities provide a source of uncorrelated returns for equity investors. We select the best two ranking commodity ETFs and rebalance the portfolio on a weekly basis.
Recently held (past year): GCC, CPER, KRBN, DJP, UNG
| CAGR | |
| Sharpe Ratio | |
| Winning Trades (%) | |
| Drawdown | |
| Current Portfolio |
87 closed calls: 41 up, 46 down or flat. Losses are included. Result % is the change in system equity over the hold, not a per-share return.
| Symbol | Entry | Exit | Result |
|---|---|---|---|
| DJP | 2026-03-23 | 2026-06-08 | 4.34% |
| UNG | 2026-03-17 | 2026-03-23 | -4.74% |
| KRBN | 2025-05-30 | 2026-03-17 | 16.83% |
| USO | 2025-05-19 | 2025-05-30 | -1.81% |
| UNG | 2025-05-07 | 2025-05-19 | -3.82% |
| CPER | 2025-04-03 | 2025-05-07 | -0.15% |
| SLV | 2025-01-30 | 2025-04-03 | -1.02% |
| UNG | 2024-11-15 | 2025-01-30 | 9.65% |
| GLD | 2023-10-13 | 2024-12-18 | 2.59% |
| USO | 2024-11-04 | 2024-11-15 | -6.59% |
See every closed call for this system · How our machine-learning models time trades
Past performance is not indicative of future results.
Commodities are an important asset class that most investors have inadequate exposure to. ETFs make it easy to gain exposure and our model helps selects the two most probable commodity ETFs. Owning commodities has historically reduced overall portfolio downside and increased Sharpe Ratios. Everyone should have some commodity exposure. Our simple system helps you find the best opportunities in this overlooked space.
Just watch the daily PortfolioAI email for updates to the model and when indicated, take the trade.
PortfolioAI is the result of almost 15 years of research. We started with simple momentum models and technical indicators and have slowly grown our capability to include over 40 different parameters for each security. Early into our research with Machine Learning we developed a simple but effective strategy for labelling our data-set. That insight allowed us to train our MLs to predict whether the next period was likely to be 'Optimal' or 'Non-Optimal' for owning the security. We use an ensemble of ML combinations for each security for additional robustness. Our MLs are retrained frequently and we continue to search for new data sets to add to our work.