Q4 Filings Reveal 'Smart Money' Rotation
An analysis of Q4 2025 13F filings reveals a significant capital rotation by institutional investors. So-called "smart money" appears to be aggressively shifting sector and asset class exposures in preparation for new market regimes in 2026, providing a rich alternative dataset for quant strategies.
A notable pivot in Q4 2025 came from Duquesne Family Office, where Stanley Druckenmiller slashed his tech exposure to 9.43%, a multi-year low, down from 13.8% the prior quarter. His firm completely exited positions in Meta, SanDisk, MongoDB, and Arm, while initiating significant new stakes in the financial sector via the XLF ETF, the S&P 500 Equal Weight ETF (RSP), and Brazilian equities (EWZ). David Tepper's Appaloosa Management also made significant international shifts, reducing exposure to China to 19.53% from a peak of over 30% earlier in the year. The firm's new focus turned to South Korea, with the iShares MSCI South Korea ETF (EWY) becoming a top buy, alongside a major increase in its holding of memory-chip maker Micron Technology. This rotation wasn't isolated. A broader market trend saw capital move from high-multiple software and AI stocks into "Old Economy" sectors like basic materials, energy, and industrials in early 2026. This "physical pivot" reflects a new focus on the tangible infrastructure—power, materials, and machinery—required to support the AI buildout, a shift away from the software models themselves. The move into cyclicals was evident in fund flows, as sector equity funds saw record inflows of $28 billion in January 2026, with natural resources funds alone attracting a record $7.5 billion. Meanwhile, U.S. equity funds experienced outflows of $34 billion in the same month. This coincided with a challenging start to 2026 for many quant funds, which saw losses as crowded long positions in tech unwound. For quant developers, this environment highlights the increasing value of unstructured data analysis. In 2026, using AI and LLMs to parse SEC filings for sentiment shifts and strategic changes is no longer a niche strategy but table stakes. The surge in AI-related keywords in corporate disclosures itself has become a key dataset, with mentions of "AI agent" jumping 6,550% year-over-year in 2025 filings. The complexity of this rotation underscores the limitations of relying solely on delayed 13F data for direct replication. These filings offer a strategic "road map" of themes influential managers are positioning for—like the AI energy bottleneck or regulatory shifts—rather than a simple buy list. The real edge comes from building systems that can identify these large-scale economic pivots in near real-time.
