Different energy.
One portfolio.
Six companies. One risk-first strategy. Explore what happens when traditional energy meets the future of power.
Loading research…
Weights estimated on 2019–2022 dataA fair test. A clear comparison.
Growth of 100 · monthly rebalancing · adjusted prices
Loading comparison…
Where the
capital goes.
The minimum-variance model chooses weights from historical covariance, not from a forecast of the highest return.
Zero weights are allowed. This view does not rerun the optimiser.
The energy universe
6 stocks · USD listings| Company | Exposure | Weight |
|---|
Targets are restored monthly in the backtest. Displayed weights are rounded.
Results without the spin.
One period. Three strategies. Same assumptions.
| Strategy | Annual return¹ | Volatility | Sharpe | Max drawdown² |
|---|
¹ Annualised arithmetic monthly mean, not CAGR. ² Measured at month ends, including starting wealth.
The research, opened up.
Six figures generated by your Python analysis.
Built to learn.
Designed to be questioned.
A historical study of an energy-focused portfolio. Every number on this page comes from the Python project.
Waiting for data
01 Separate training from testing
Weights are estimated using monthly returns through December 2022, then held as fixed targets from January 2023 onwards. The optimiser never uses test-period returns. The latest unfinished month is excluded.
02 Minimise variance, not maximise returns
The objective is wᵀΣw. Weights must sum to 100%, with no short selling. The 2,000 sampled allocations illustrate alternatives; they are not an exact efficient frontier.
03 Read the measures correctly
Annual return is mean monthly return × 12. Volatility is monthly standard deviation × √12. Sharpe subtracts the 2% annual risk-free assumption before dividing by volatility. Chart growth compounds actual monthly returns.
04 Know what the model leaves out
The simulation assumes monthly rebalancing without transaction costs, management fees or taxes. Monthly drawdown misses movements within a month. Selecting companies today introduces hindsight, and historical relationships can change.