ConvexPi

Playground

Real strategy replications, executed in your browser. Each example recomputesthe strategy from its building blocks — reconstructing the Fama-French factors from the underlying size/value/momentum portfolios, or forming a long-short book by ranking industries each month — rather than reading a finished factor off the shelf. It checks the reconstruction against the published series, then splits at the paper's publication year (the McLean & Pontiff test) to ask: did the edge survive out of sample?

Reconstruct the value factor from the 6 size×B/M building-block portfolios

Open in Colab (full data)
Python · runs in your browser
Loading editor…
Everything runs client-side via Pyodide on real Ken-French data — nothing is sent to a server. Edit the code and re-run; prefer a full environment? Copy it into a Colab notebook.

Everything runs client-side via Pyodide on real Ken-French data — edit the code and re-run. For a full end-to-end replication on freshly downloaded data (including true single-name cross-sectional strategies), use the Open in Colab link on each example.

From the replication leaderboard

See all →
Idiosyncratic volatility puzzledormant
Asset-class momentumalive
MAX effect (lottery stocks)dormant
Earnings yield (P/E effect)dormant
Investment (CMA)dormant
Long-term reversaldormant

Verified, out-of-sample-scored replications of canonical strategies — runnable in Colab, open to your contributions.