correlation-features-USA(corr=0.9645)
df["Combined_Price"] = (
df["Gold"] * 10 # 安全資産、価値保存
+ (df["Copper"] - df["Zinc"]) * 0.75 # 工業金属の需給差（精錬コスト的な差？）
- (df["Wheat"] + df["Corn"]) * 2.5 # 食料インフレ圧力（消費者・企業のコスト増）
+ (df["Oil"] - df["Uranium"]) * 30 # 化石燃料 vs 原子力（エネルギーのパラダイム）
)
I found correlation(corr=0.9645) between combined features and DJI.
This is from FRED data, so that some delays are included.
https://medium.com/@takapoko182/a-personal-attempt-to-reconstruct-djia-using-commodity-prices-2015-2025-638e4375919c