Run the 20/100 SMA strategy

lean/demo-generated-data/demo_sma.py is a complete dynamic-universe strategy over native LEAN files.

Strategy

For every member of sweden100:

  1. update a 20-day and 100-day simple moving average from raw daily closes;
  2. enter long when SMA 20 crosses above SMA 100;
  3. exit when SMA 20 crosses below SMA 100;
  4. equal-weight all bullish assets to 95% total target exposure.

The indicators warm naturally from streamed daily bars. No bulk history request is made. With the current fixture, the slow SMA becomes ready on the 100th 2024 session, leaving 151 sessions for signals.

Run

cd /home/hi/qrt
cd lean/demo-generated-data
uv run lean backtest demo_sma.py --no-update

A successful run logs SWEDEN_SMA_CROSS_VERIFIED.

The verified synthetic result produced:

  • 100 ready SMA pairs;
  • 154 bullish entries and 106 bearish exits;
  • 107 batched rebalances;
  • 448 filled orders and zero invalid orders;
  • 47 final long positions;
  • +1.806% net profit and 1.4% maximum drawdown;
  • zero failed data requests.

Three final-session orders remain submitted as market-on-open orders because the dataset contains no following XSTO session.

Batched target semantics

LEAN’s list overload expects PortfolioTarget.quantity to be a portfolio percentage and internally calculates share quantities:

targets = [
    PortfolioTarget(symbol, target_weight, "20/100 bullish cross")
    for symbol in bullish
]
self.set_holdings(targets)

Do not pass PortfolioTarget.percent(...) outputs into that list overload. Those objects already contain calculated share quantities and would be converted a second time, creating oversized orders.

Production changes

The fixture uses ConstantFeeModel(0) and synthetic quote/volume data. Before interpreting performance, implement realistic Swedish fees, spreads, fills, lot/tick rules, liquidity, market-impact, and brokerage constraints.

Back to top