Run a strategy

Start from an example or paste your own. Declare tunable numbers in PARAMS and return target weights from on_bar(ctx, data). How strategies run

Imports: numpy, pandas and the standard math modules. Full list Ctrl + Enter to run

No choices to make. The score can't depend on a setup that happens to suit the strategy.

2,000 random slices of history, 2 to 8 years each, drawn from the whole US market and 49 industry portfolios going back to 1926 (Ken French data, survivorship-bias-free).
Each slice is scored with the same six sub-scores as a normal report, against buy & hold of that asset over the same days, at 5 bps per side.
The score is the average, capped by how much the strategy beats holding: no edge over buy & hold, at most 50. A rerun usually lands within 2 points.

The result is the Quantgrounds score. Takes about a minute. The detailed tests below the score use the whole US market (1926 to today) as a reference. Choose "Your setup" to pick assets, dates and costs yourself; that gives a diagnostic score for that setup only.

Adjusted daily closes.