You want to know if institutions are buying or selling a stock. The 13F database has the answer — but only if you look at the right number.
The right number is share count. Not dollar value.
Here's why, and how to use it.
Why share count > dollar value
A fund's position value changes for three reasons:
- They bought or sold shares (action)
- The stock price moved (market)
- Their total AUM changed from flows (business)
Only #1 is an investment decision. Dollar value mixes all three together. Share count isolates the decision.
| What happened | Shares | Value | Which one tells the truth? |
|---|---|---|---|
| Fund bought 100K shares, stock flat | Up | Up | Both |
| Fund did nothing, stock up 20% | Same | Up 20% | Shares |
| Fund sold 50K shares, stock up 30% | Down | Up ~15% | Shares |
| Fund bought 200K shares, stock down 10% | Up | Roughly flat | Shares |
In scenario 3, dollar value shows the position growing — but the fund is actually selling. In scenario 4, dollar value looks flat — but the fund is aggressively buying the dip.
The accumulation/distribution framework
Accumulation signals (bullish)
- Share count increasing across multiple quarters: Steady buying, not a one-time allocation
- Shares increasing while price declining: Buying the dip = real conviction
- Multiple institutions increasing shares simultaneously: Consensus accumulation
- New institutions initiating positions: Fresh interest from new money
Distribution signals (bearish)
- Share count decreasing across multiple quarters: Steady selling, not a one-time trim
- Shares decreasing while price rising: Selling into strength = reducing conviction
- Multiple institutions decreasing shares simultaneously: Consensus exit
- Institutions completely exiting: Total loss of conviction
Neutral / noise
- Share count flat, value changing: Price-driven, no institutional action
- Small share count changes (<5%): Could be rebalancing, not directional
- Passive manager changes: Index-driven, not conviction-driven
Step-by-step tracking process
Step 1: Get two consecutive quarters of data
For each institution holding the stock, compare Q3 and Q4 share counts.
Step 2: Calculate the change
For each holder:
- Shares increased → buyer
- Shares decreased → seller
- Shares unchanged → holder
- New position → initiator
- Position gone → exiter
Step 3: Aggregate across all holders
Count:
- Total buyers vs. sellers
- Total shares added vs. removed
- Net institutional share change
Step 4: Weight by manager type
Not all buyers are equal:
- Active manager buying = conviction signal
- Passive manager buying = flow-driven (lower signal)
- Market maker changes = hedging (no signal)
Step 5: Look at the trend over 2-4 quarters
One quarter of net buying could be noise. Three consecutive quarters of net buying is a pattern.
The aggregation trap
Be careful with aggregate institutional ownership numbers:
"Institutional ownership increased from 75% to 78%" could mean:
- Many institutions bought small amounts (broad accumulation)
- One massive passive fund had inflows (meaningless)
- The company bought back shares, reducing float and mechanically increasing the institutional percentage
Always look at the individual holder changes, not just the aggregate.
Real-world application
Suppose you're tracking NVDA institutional ownership:
| Quarter | # Buyers | # Sellers | Net shares | Signal |
|---|---|---|---|---|
| Q1 2025 | 300 | 150 | +50M | Accumulation |
| Q2 2025 | 250 | 200 | +10M | Slowing accumulation |
| Q3 2025 | 180 | 280 | -30M | Distribution begins |
| Q4 2025 | 200 | 250 | -15M | Distribution continues |
This pattern shows a transition from accumulation to distribution. The turning point (Q2→Q3) is the key signal — that's when institutional consensus shifted.
Tools for tracking
Manual (free)
- Download 13F filings from EDGAR for your target stock
- Extract share counts per holder per quarter
- Build a spreadsheet tracking changes
- Time-consuming but thorough
Automated (platforms)
- 13F Insight and similar platforms pre-compute these changes
- Search by stock ticker to see all institutional holders and their quarter-over-quarter changes
- Filter by buyer/seller, sort by change magnitude
The bottom line
Institutional buying and selling signals live in share count changes, not dollar values. Track share counts across multiple holders over multiple quarters, weight by manager type, and look for consensus patterns. One fund buying is anecdote. Twenty funds buying over three quarters is data.
Originally published at 13F Insight
Top comments (0)