Three related reports — choose one.
A five-model quantitative-finance benchmark — 296 tasks and 1,480 blinded judgments across five models, with downloadable datasets.
What do QCA-generated research repositories actually establish? A source-pinned investigation of two article reproductions — Avellaneda–Stoikov market-making and skfolio portfolio optimization — with results, alternatives, and per-artifact evidence.
QuantBrain's advisory (35B) and code (397B) models measured against their base and competitor comparators across three notebooks — Russian financial-advisory and applied-code task banks, with paired uplift-confirmation analysis.