The concept of "250 assets" is a too narrow base for large-scale portfolio optimization, as time is limited for these academic exercises. We manage thousands of instruments covering multiple asset classes and geographies, where liquidity can be a dwindling resource rather than just a parameter. A sample so small masks the true challenges of risk management and diversification in a complex macroeconomic environment. For example, consider our exposure to European government bonds; a simple simulation on 250 assets will never predict the impact of yield variations on the entire portfolio. Attention is a bottleneck, and focusing on such limited test beds ration our intellectual resources poorly.