Lava AI Daily-Frequency Heat Panel MCP Data: How to Integrate mentions and weighted_heat into Quantitative Backtesting

Lava's daily-frequency heat panel (MCP tool get_stock_heat_panel, sharing the same source as the HTTP API) aggregates mentions, agencies, weighted_heat, and is_first_mention_day for individual stocks or concepts by natural day, with a maximum of 20 instruments per request and a maximum range of 92 days. Integration into backtesting requires pulling data in batches and segments with boundary deduplication, explicitly aligning natural days with trading days, confirming the data availability time for the current day, and using PIT-based scoring rather than backfilling history with current scores.