Track a benchmark's return series with eight tradeable assets, holding non-negative weights that sum to one. The assets can only approximate the benchmark, so perfect replication is impossible; the job is to minimise tracking error, the mean squared gap between the portfolio's return and the benchmark's over the history. A regression on the simplex. Score is mean squared tracking error (scaled), to minimise. 8-D.
| Algorithm | Tracking error | Evals |
|---|---|---|
| — no runs yet — | ||
The JS objective is a line-for-line port of
example_applications/index_tracking and agrees to floating-point
tolerance. The dashed line is the benchmark return, the solid line the portfolio's;
the bars are the chosen weights.