13F filings: what they show, what they hide, and the CUSIP problem
23.4 million institutional positions from 9,985 managers — filed 45 days late, long-only, and by CUSIP rather than ticker. All three facts change how you read them.
Institutional managers with over $100 million in US equities must disclose their holdings quarterly, on Form 13F. It is the closest thing retail investors have to seeing what large money owns.
It is also late, partial, and filed in an identifier most tools cannot resolve.
Three limits you have to hold in mind
It is 45 days stale. The deadline is 45 days after quarter end. A position you read in mid-February describes 31 December. In a fast-moving name the manager may be out entirely.
It is long-only, US-listed equity. Short positions do not appear. Neither do bonds, most derivatives, cash, or non-US listings. A fund that looks aggressively long may be hedged in instruments the form never touches. You are seeing one leg.
It is a snapshot, not a flow. You get the position at quarter end, not the path. A holding that is unchanged quarter on quarter may have been sold and rebought entirely.
The CUSIP problem, which is where most tools quietly break
13F reports by CUSIP, not by ticker.
If your pipeline does not map CUSIPs to instruments, the institutional block for a given company comes back empty. And an empty block does not look like a failure — it looks like a finding. “No institution holds this” is a strong statement, and it is the statement a missing join produces by default.
We hit exactly this. Our AI evidence pack had a fund_holding block that kept
coming back empty, and the reason was that the table keys on CUSIP while
everything else keys on our internal instrument id. Without the hop through the
CUSIP map, the block silently reported nothing where 23.4 million positions
existed.
There are now 23,450,940 positions from 9,985 managers resolved onto the same instrument pages as the price and the filings.
What is genuinely informative
- Change in position size, quarter over quarter, by manager. Direction and magnitude carry more than the level.
- Concentration. A position that is 8% of a manager’s book means something different from the same dollar amount at 0.3%.
- New positions and full exits, which are the cleanest signals the form produces.
- Disagreement. Two large holders moving opposite ways in the same quarter is where the question is.
What it is not
It is not a trading signal on its own, and the 45-day lag is only part of the reason. The bigger problem is the one we found while testing our own event studies: a change in reported holdings can be almost entirely price-driven. If a stock rose 40% and a manager did nothing, the dollar value of their position rose 40%. Reading that as accumulation is reading the price back to yourself.
Our AI assistant flagged this on SPY: institutional holdings had fallen 5.4% to $342.06bn in Q1 2026 while the price was in an uptrend, and it listed the conflict as a factor — while noting explicitly that the move “may reflect price-driven valuation changes rather than active selling.”
That hedge is the correct reading, and it is the sort of thing a summary written for fluency would have flattened into a story about institutions selling.
13F positions, insider transactions and SEC filing facts sit on the same page as the price. Open it without an account.