Institutional Signals

Introduction

Institutional investors with $100M+ in assets under management must disclose their holdings quarterly via SEC Form 13F. The system monitors these filings for accumulation signals -- patterns of institutional buying that can indicate informed conviction.

Institutional activity is not inherently predictive. Index funds rebalance mechanically. Many active managers are closet indexers. But a subset of high-conviction managers -- those running concentrated portfolios with demonstrated long-term alpha -- make allocation decisions based on deep fundamental research. When several of these managers independently start building positions in the same stock, the signal is worth measuring.

What Are 13F Filings?

SEC Form 13F is a quarterly report that institutional investment managers with $100M or more in qualifying assets must file with the Securities and Exchange Commission. The filing discloses all US equity positions held at the end of each calendar quarter.

Key characteristics of 13F data:

• Filing deadline: 45 days after quarter end. Q1 holdings (as of March 31) are due by May 15. Q2 by August 14. Q3 by November 14. Q4 by February 14.
• Coverage: US equity holdings including common stock, ETFs, options, warrants, and convertible securities. The filing shows the number of shares held and the market value at quarter end.
• What it does not show: Short positions, fixed income, international holdings, or derivatives (other than listed options). A manager could be net short a stock through unlisted derivatives while the 13F shows a long position.
• Staleness: By the time a filing appears, the data describes positions that are 45 to 135 days old. The manager may have already exited the position. This lag is a fundamental limitation of the data source.

The engine ingests 13F filings automatically after each quarterly deadline, processing them through a pipeline that maps securities to the internal asset universe and computes accumulation signals.

Accumulation Signals

Raw 13F data shows who holds what. The accumulation signal answers a more useful question: is institutional interest increasing or decreasing?

For each stock in the universe, the system computes:

• Curated holders: How many managers from the curated list hold this stock in the current quarter
• New positions: How many curated managers initiated a position this quarter that they did not hold last quarter
• Net shares: The aggregate change in shares held across curated managers (new shares bought minus shares sold)

These raw counts are then percentile-ranked across the full universe to produce the accumulation percentile -- a 0-100 score indicating where this stock falls relative to all other stocks in terms of institutional interest.

Curated Manager List

Not all 13F filers are equally informative. The SEC requires thousands of institutions to file, but the majority are index funds, quantitative shops running thousands of positions, or low-conviction managers whose buying and selling patterns carry no meaningful information about individual stock quality.

The system tracks a curated set of high-conviction managers selected for specific characteristics:

• Concentrated portfolios: Fewer positions means each holding represents a meaningful conviction bet. A manager with 500 positions is spreading bets; a manager with 25 is concentrating on best ideas.
• Long holding periods: Managers who trade frequently are reacting to noise. Managers who hold positions for years are making fundamental assessments.
• Demonstrable alpha generation: The manager's track record shows persistent outperformance, not just one good year.

The curated list is reviewed periodically but remains static between reviews. This prevents the system from retroactively changing the manager universe based on recent performance.

How It Feeds Into Scoring

Institutional accumulation occupies a specific place in the scoring pipeline:

1. 13F data is ingested quarterly and processed into per-stock accumulation signals
2. Accumulation signals are percentile-ranked across the full universe (0-100)
3. Accumulation percentile feeds into compute_catalyst_strength(), where it is averaged with the SUE (Standardized Unexpected Earnings) percentile when both are available
4. Catalyst strength is one of the inputs to Track B (Mispricing), Gate 3 -- the catalyst gate
5. Gate 3 checks whether near-term catalysts exist to close the valuation gap identified by Gates 1 and 2

The relationship is additive, not dominant. A high accumulation percentile improves catalyst strength, which improves the Track B gate score, which contributes to the multiplicative track score. But because Track B uses multiplicative scoring across all four gates, strong institutional accumulation alone cannot carry a stock that fails on valuation, downside protection, or quality.

Smart Money Page

The /smart-money page provides direct access to the institutional data that feeds the accumulation signal, organized into three tabs:

Fund Tracker -- Browse individual manager portfolios from the curated and top-AUM lists. See their top holdings, position sizes, and changes from the previous quarter. This is the raw data behind the accumulation signal, presented per-manager rather than per-stock.

Market Signals -- An aggregate view showing which stocks are most widely held across the curated manager universe, which stocks saw the most new position initiations this quarter, and where crowding risk exists (many managers holding the same names). This tab answers the question: what is the curated manager universe doing collectively?

Clone Lab -- An educational tool that shows what a portfolio would look like if you followed a specific manager's disclosed strategy. Select a manager, choose an allocation approach (equal-weight, conviction-weight, or sector-constrained), and see the resulting portfolio composition. This is a thought experiment, not a recommendation -- past manager performance does not predict future results, and 13F data is already stale by the time it is available.

Verify It Yourself

Known Limitations