Statistical validation and data-quality software for quantitative research.
In development. Pilot program planned for 2027.
01PROBLEM
The best of many backtests often wins by chance.
best of 70new data
in sampleout of sample
Illustration: 70 random strategies with no real edge.
02PRODUCT
Built to take your data and your strategy in, and give a tested result out.
Market datayour vendor
Data checksgaps, stale, splits
Backtestyour strategy
Validationwalk-forward, purge, embargo
Robustnessparameter shifts
Reportintervals, p-values
Rust corePython bindingsCLI
Planned interfaces.
03DATA CHECKS
Bad prices will be flagged before they reach a backtest.
Illustration: simulated daily closes.
04VALIDATION
Test windows will stay unseen, with a purge and an embargo against leakage.
fold 1
fold 2
fold 3
fold 4
fold 5
oldertimenewer
trainpurgetestembargo
Schematic.
05ROBUSTNESS
A real result survives small parameter changes.
Illustration. Axes: two strategy parameters. Colour: test-period Sharpe ratio.
06SCOPE