To time-align comparisons against the benchmark, it would be convenient if the backtester could preload the batch transform window, so that the algorithm had enough data to trade on the first day of a backtest.
Example: At present, if a batch transform has a 60-day window, the algorithm won't start trading until about three months (60 trading days) after the backtest starts. Meanwhile, the benchmark fluctuations have been recorded throughout those first three months. So at the end of one year, the results compare a year of benchmark performance to only nine months of algorithm performance.
If the backtester preloaded the window, the algorithm and the benchmark would both be measured over the same full year, and the comparison would be more meaningful.
For live trading, preloading would be even more important. It would be painful to have to wait three months to start trading while the batch window slowly filled up day by day.