Cross-sectional backtests

[javiz]

Thank you for this amazing library @polakowo ! I developed a simple event-driven cross-sectional backtest that i'd like to port to vectorbtpro, but am struggling to find any docs on the topic (or related). I'm not sure the approach is vectorizable - I may need to tweak the approach for it to work with vectorbtpro.

My current pipeline uses historical SP500 constituents as the stock universe (in order to avoid survivorship bias), and on each bar will run the strategy against all symbols and generate bracket orders. It can generate any number of orders, as long as it doesn't go over the portfolio budget. If a symbol is no longer part of the SP500, any open positions are closed.

In porting this I began by creating a large Dataframe with all the historical data, and a mask for whether the symbol is part of SP500 or not. I'm able to run backtests on individual symbols, but am unsure how to run a cross-sectional backtest with a shared budget. I probably just need to get more used to the library, but please let me know if I missed any relevant docs, or whether I should tweak the approach for it to be compatible with vectorbtpro. Thank you!