I am using a relative strength model based on Levy, and I have a question regarding the portfolio composition.
How is it possible that my current portfolio consists of different stocks depending on the time period I use for backtesting? In one chart, I am looking at a backtest starting from April 28, 2025, which marks the point when my crisis filter switched back to long.
Should I now trade exactly this portfolio composition and effectively set the system start to the most recent long signal of the crisis filter (which I define based on the S&P 500)?
Or, if I were to start the system today, I would end up with a completely different set of around ten stocks in the portfolio.
You are asking two different questions: how does the backtest behave and how should you start your live trading?
The nature of a backtest is that it will not take any trades before the start date of the backtest. Therefore, on Day 1, all capital and all "slots" are available and they will get used as entry signals occur. If you run two backtests, one starting last week and one starting a month ago, then your initial set of trades will be likely be different, as the backtest with the earlier start date probably has some open trades when the later backtest starts.
How to start live trading is a personal choice. You could start your signal-generating backtest at the same time that you start live trading, effectively ignoring anything that might have happened in the past. Or you can start your backtest earlier, and on the day that you start live trading you can either fully buy into each open position, scale into each position over X days, or ignore currently open positions and only enter when the backtest shows new entry signals. All of these are valid techniques and over the long run should not make much difference on your total P/L.
