Backtesting results

The DRL algorithm systematically invests the initial_amount until all the amount is exhausted. However, during validation/trading the entire amount may not be invested (due to hmax) and high initial_amount (1 million). However for the baseline we are using point to point close price returns. This means for the baseline we are considering the entire initial amount whereas for the back test the entire amount may never be invested. This means that the backtest will usually underperform the baseline?

Should the baseline also use a similar investing method...In practice no one would invest 1 million on day 1 even if its an index....alternatively for the baseline we should allocate the initial amount to the index and only allow sell actions?