What is Portfolio returns?
How would your portfolio have developed if you had blindly bought/sold based on the signals?
That is, you buy a “buy signal” when it triggers at time t, and then sell at t+1. Conversely, you sell and then buy back for a “sell signal”.
Important: we do not recommend blindly trading on signals in this way.
We simply use this as a way to evaluate in a standardized way how your portfolio would have developed over the time period that we are backtesting on. Portfolio returns are, at a high level, a function of two things:
information + strategy. Since signal triggers represent the information component, we use the most basic strategy imaginable across all signals in order so that we get comparable evaluation metrics for the signals themselves.
Generally you should always use a trading signal as an additional data point in your trading - not the only data point.
We use compounding when calculating portfolio returns (meaning we simulate re-investing the full amount for the next trade), and we use a transaction cost of 0.2% per trade - which should be considered conservative and thus we do not add an additional slippage parameter.