Atlas for Quant Finance
Backtests that
show their work
Atlas writes the notebook, pulls the prices, and runs every cell in front of you. Factor tests, ETF backtests and write-ups, with the series id in the code and results you can rerun.
$20/month · Cancel anytime
px = load_prices("2005-01", "2026-01")
rets = px.pct_change()mom = rets.rolling(231).sum().shift(21) port = backtest(rank(mom), rets)
plot_equity(port, benchmark=spx)
Sharpe 0.81 vs 0.58, every cell rerunnable
Example tasks
From idea to backtest to paper
Backtest a 12-1 momentum factor
Twenty years of daily prices, decay charted by decade, with drawdowns, turnover, and Sharpe by volatility regime.
Compare capex across the 10-Ks you upload
Drop in the PDFs and Atlas pulls capex, margins and guidance into one table, with the page cited.
Stress-test the AI capex trade
NVDA, MSFT, AMZN, GOOGL and META against SPY since 2023: drawdowns, rolling volatility and correlation.
Publish the result as a paper
LaTeX with regression tables, figures, and a BibTeX bibliography, compiled to PDF in the same workspace.
Why Atlas
Research you can rerun
Rerunnable by default
Every result is a plain Python file. Change the window or the universe and run it again.
Primary sources, cited
Prices and series pulled at run time, with the series id in the code.
You approve every change
Nothing touches your files without a green light. Review each proposed edit and approve or reject it inline.