01 / Paper trading · simulated money
Track record
How the two portfolios have done since the first paper-trading day, against the S&P 500 (SPY). Closing values, published after the market close.
Paper trading · indexed to 100
Show data as a table
02 / Portfolios
Portfolio detail
Holdings at the last close, alphabetical. Weights and quantities are not published.
03 / Backtest
Out-of-sample backtest
Before going live, the rules of A and B were tested on past data without using anything the system could not have known at the time: the forecasting model is trained only on the years before each test period. Execution costs included (0.1% per trade).
Read with caution.
- These are simulations: no real money was invested over those periods.
- Hindsight in the stock list. B's universe includes companies picked today, some of them because they have already risen a lot. Running B on the 50 largest S&P 500 stocks only, the 2019–2026 return drops to 17.5% a year, in line with SPY (17.9%), with a smaller maximum drawdown (−24.7% vs −33.7%).
- The forecasting model showed no measurable edge out of sample: the result comes mostly from the rules (momentum, trend filter, turnover control).
- Seven years with only one real bear market (2022): a small sample.
- Past performance, simulated or real, is not indicative of future results.