Thanks for the comment, had a similar discussion over email recently.
The portfolio builder is based on weekly data to make it speedy enough for real-time queries. Thus the base for the performance calculation in the monthly table is always the last week of the preceeding week, here March 27. So for April the calculation is currently for the three weeks since 3/27 until 4/21 (week of 4/17).
Doing the math on daily data you probably took 3/31 until 4/24, this explains the gap. Below the underlying data so you can reconcile, see especially the 1.74% in the week of 3/27.
The initial idea for the tool was for building a portfolio with long-term data, not real-time performance tracking – which undoubtly is raising now questions. Let me see how I can match better between the daily strategy performance and the portfolio builder.
Here an Excel file with the data.